diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,141986 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 1000,
+  "global_step": 23598,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.028248587570621e-06,
+      "loss": 1.5851,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.056497175141243e-06,
+      "loss": 1.2643,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.084745762711864e-06,
+      "loss": 1.3893,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.112994350282486e-06,
+      "loss": 1.2454,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.141242937853107e-06,
+      "loss": 1.3615,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.169491525423729e-06,
+      "loss": 1.3839,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.19774011299435e-06,
+      "loss": 1.3906,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.225988700564972e-06,
+      "loss": 1.3025,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.254237288135593e-06,
+      "loss": 1.4215,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.282485875706215e-06,
+      "loss": 1.4113,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.310734463276836e-06,
+      "loss": 1.3774,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3389830508474574e-06,
+      "loss": 1.2773,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.367231638418079e-06,
+      "loss": 1.3063,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3954802259887004e-06,
+      "loss": 1.3608,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.423728813559322e-06,
+      "loss": 1.3337,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4519774011299434e-06,
+      "loss": 1.3235,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.480225988700565e-06,
+      "loss": 1.3891,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5084745762711864e-06,
+      "loss": 1.3538,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5367231638418077e-06,
+      "loss": 1.3358,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5649717514124294e-06,
+      "loss": 1.3885,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.593220338983051e-06,
+      "loss": 1.3151,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6214689265536724e-06,
+      "loss": 1.2601,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6497175141242936e-06,
+      "loss": 1.3647,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6779661016949153e-06,
+      "loss": 1.3327,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7062146892655366e-06,
+      "loss": 1.402,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7344632768361583e-06,
+      "loss": 1.3426,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7627118644067796e-06,
+      "loss": 1.2635,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7909604519774013e-06,
+      "loss": 1.3603,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8192090395480226e-06,
+      "loss": 1.2678,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.847457627118644e-06,
+      "loss": 1.2893,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8757062146892656e-06,
+      "loss": 1.3865,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.903954802259887e-06,
+      "loss": 1.1642,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9322033898305086e-06,
+      "loss": 1.4395,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.96045197740113e-06,
+      "loss": 1.2341,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9887005649717516e-06,
+      "loss": 1.3596,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.016949152542373e-06,
+      "loss": 1.3293,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.045197740112994e-06,
+      "loss": 1.2486,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.073446327683616e-06,
+      "loss": 1.1159,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1016949152542375e-06,
+      "loss": 1.3779,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.129943502824859e-06,
+      "loss": 1.2358,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.15819209039548e-06,
+      "loss": 1.2124,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.186440677966102e-06,
+      "loss": 1.4325,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2146892655367235e-06,
+      "loss": 1.2974,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.242937853107345e-06,
+      "loss": 1.3252,
+      "step": 44
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.271186440677966e-06,
+      "loss": 1.2978,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2994350282485878e-06,
+      "loss": 1.2664,
+      "step": 46
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.327683615819209e-06,
+      "loss": 1.1012,
+      "step": 47
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3559322033898303e-06,
+      "loss": 1.246,
+      "step": 48
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.384180790960452e-06,
+      "loss": 1.402,
+      "step": 49
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4124293785310738e-06,
+      "loss": 1.2586,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.440677966101695e-06,
+      "loss": 1.1195,
+      "step": 51
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4689265536723163e-06,
+      "loss": 1.2245,
+      "step": 52
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.497175141242938e-06,
+      "loss": 1.2357,
+      "step": 53
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5254237288135597e-06,
+      "loss": 1.2645,
+      "step": 54
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.553672316384181e-06,
+      "loss": 1.3015,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5819209039548023e-06,
+      "loss": 1.3317,
+      "step": 56
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6101694915254236e-06,
+      "loss": 1.1705,
+      "step": 57
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.638418079096045e-06,
+      "loss": 1.192,
+      "step": 58
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6666666666666666e-06,
+      "loss": 1.2233,
+      "step": 59
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6949152542372883e-06,
+      "loss": 1.3453,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7231638418079096e-06,
+      "loss": 1.2694,
+      "step": 61
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.751412429378531e-06,
+      "loss": 1.2139,
+      "step": 62
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7796610169491525e-06,
+      "loss": 1.2634,
+      "step": 63
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8079096045197743e-06,
+      "loss": 1.3389,
+      "step": 64
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.836158192090395e-06,
+      "loss": 1.2382,
+      "step": 65
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.864406779661017e-06,
+      "loss": 1.433,
+      "step": 66
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8926553672316385e-06,
+      "loss": 1.2189,
+      "step": 67
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.92090395480226e-06,
+      "loss": 1.2666,
+      "step": 68
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.949152542372882e-06,
+      "loss": 1.2102,
+      "step": 69
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.977401129943503e-06,
+      "loss": 1.2967,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.0056497175141245e-06,
+      "loss": 1.2245,
+      "step": 71
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.033898305084745e-06,
+      "loss": 1.2043,
+      "step": 72
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.062146892655367e-06,
+      "loss": 1.2824,
+      "step": 73
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.090395480225989e-06,
+      "loss": 1.322,
+      "step": 74
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.1186440677966105e-06,
+      "loss": 1.2108,
+      "step": 75
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.146892655367232e-06,
+      "loss": 1.296,
+      "step": 76
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.175141242937853e-06,
+      "loss": 1.2428,
+      "step": 77
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.203389830508475e-06,
+      "loss": 1.2615,
+      "step": 78
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.231638418079096e-06,
+      "loss": 1.3277,
+      "step": 79
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.259887005649717e-06,
+      "loss": 1.1316,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.288135593220339e-06,
+      "loss": 1.2148,
+      "step": 81
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.316384180790961e-06,
+      "loss": 1.2345,
+      "step": 82
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.3446327683615824e-06,
+      "loss": 1.1651,
+      "step": 83
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.372881355932204e-06,
+      "loss": 1.2112,
+      "step": 84
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.401129943502825e-06,
+      "loss": 1.2927,
+      "step": 85
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.429378531073447e-06,
+      "loss": 1.0823,
+      "step": 86
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4576271186440676e-06,
+      "loss": 1.1778,
+      "step": 87
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.485875706214689e-06,
+      "loss": 1.2654,
+      "step": 88
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.514124293785311e-06,
+      "loss": 1.2316,
+      "step": 89
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.542372881355933e-06,
+      "loss": 1.2764,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.570621468926554e-06,
+      "loss": 1.3156,
+      "step": 91
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.598870056497175e-06,
+      "loss": 1.2063,
+      "step": 92
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.627118644067797e-06,
+      "loss": 1.3162,
+      "step": 93
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.655367231638418e-06,
+      "loss": 1.2254,
+      "step": 94
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6836158192090395e-06,
+      "loss": 1.277,
+      "step": 95
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.711864406779661e-06,
+      "loss": 1.2467,
+      "step": 96
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.740112994350283e-06,
+      "loss": 1.1348,
+      "step": 97
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.768361581920905e-06,
+      "loss": 1.2484,
+      "step": 98
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7966101694915255e-06,
+      "loss": 1.4226,
+      "step": 99
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.824858757062147e-06,
+      "loss": 1.246,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.853107344632768e-06,
+      "loss": 1.295,
+      "step": 101
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.88135593220339e-06,
+      "loss": 1.2571,
+      "step": 102
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9096045197740115e-06,
+      "loss": 1.329,
+      "step": 103
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.937853107344633e-06,
+      "loss": 1.252,
+      "step": 104
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.966101694915255e-06,
+      "loss": 1.4049,
+      "step": 105
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9943502824858766e-06,
+      "loss": 1.1897,
+      "step": 106
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.0225988700564974e-06,
+      "loss": 1.2696,
+      "step": 107
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.050847457627119e-06,
+      "loss": 1.3015,
+      "step": 108
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.07909604519774e-06,
+      "loss": 1.2297,
+      "step": 109
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.107344632768362e-06,
+      "loss": 1.1863,
+      "step": 110
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.135593220338983e-06,
+      "loss": 1.2502,
+      "step": 111
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.163841807909605e-06,
+      "loss": 1.1854,
+      "step": 112
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.192090395480226e-06,
+      "loss": 1.3345,
+      "step": 113
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.220338983050847e-06,
+      "loss": 1.2265,
+      "step": 114
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.2485875706214685e-06,
+      "loss": 1.2789,
+      "step": 115
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.27683615819209e-06,
+      "loss": 1.164,
+      "step": 116
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.305084745762712e-06,
+      "loss": 1.1893,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.333333333333334e-06,
+      "loss": 1.1759,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.361581920903955e-06,
+      "loss": 1.1158,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.389830508474576e-06,
+      "loss": 1.2981,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.418079096045198e-06,
+      "loss": 1.2653,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.446327683615819e-06,
+      "loss": 1.1933,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.4745762711864405e-06,
+      "loss": 1.2671,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.502824858757062e-06,
+      "loss": 1.2275,
+      "step": 124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.531073446327684e-06,
+      "loss": 1.179,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.559322033898306e-06,
+      "loss": 1.2004,
+      "step": 126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.587570621468927e-06,
+      "loss": 1.2029,
+      "step": 127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.615819209039548e-06,
+      "loss": 1.2095,
+      "step": 128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.644067796610169e-06,
+      "loss": 1.1818,
+      "step": 129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.672316384180791e-06,
+      "loss": 1.3022,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.7005649717514124e-06,
+      "loss": 1.1924,
+      "step": 131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.728813559322034e-06,
+      "loss": 1.2135,
+      "step": 132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.757062146892656e-06,
+      "loss": 1.2238,
+      "step": 133
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.7853107344632776e-06,
+      "loss": 1.2195,
+      "step": 134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.813559322033898e-06,
+      "loss": 1.1831,
+      "step": 135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.841807909604519e-06,
+      "loss": 1.2602,
+      "step": 136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.870056497175141e-06,
+      "loss": 1.2777,
+      "step": 137
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.898305084745763e-06,
+      "loss": 1.2664,
+      "step": 138
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.926553672316384e-06,
+      "loss": 1.2965,
+      "step": 139
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.954802259887006e-06,
+      "loss": 1.1657,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.983050847457628e-06,
+      "loss": 1.1844,
+      "step": 141
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.0112994350282495e-06,
+      "loss": 1.2737,
+      "step": 142
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.0395480225988695e-06,
+      "loss": 1.31,
+      "step": 143
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.067796610169491e-06,
+      "loss": 1.2514,
+      "step": 144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.096045197740113e-06,
+      "loss": 1.2063,
+      "step": 145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.124293785310735e-06,
+      "loss": 1.1355,
+      "step": 146
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.152542372881356e-06,
+      "loss": 1.1766,
+      "step": 147
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.180790960451978e-06,
+      "loss": 1.3486,
+      "step": 148
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.2090395480226e-06,
+      "loss": 1.2281,
+      "step": 149
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.23728813559322e-06,
+      "loss": 1.1778,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.2655367231638415e-06,
+      "loss": 1.256,
+      "step": 151
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.293785310734463e-06,
+      "loss": 1.3494,
+      "step": 152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.322033898305085e-06,
+      "loss": 1.2966,
+      "step": 153
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.350282485875707e-06,
+      "loss": 1.1729,
+      "step": 154
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.378531073446328e-06,
+      "loss": 1.1986,
+      "step": 155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.40677966101695e-06,
+      "loss": 1.2482,
+      "step": 156
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.435028248587572e-06,
+      "loss": 1.2089,
+      "step": 157
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.463276836158192e-06,
+      "loss": 1.2002,
+      "step": 158
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.4915254237288134e-06,
+      "loss": 1.2957,
+      "step": 159
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.519774011299435e-06,
+      "loss": 1.3125,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.548022598870057e-06,
+      "loss": 1.2566,
+      "step": 161
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.5762711864406785e-06,
+      "loss": 1.2481,
+      "step": 162
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.6045197740113e-06,
+      "loss": 1.1581,
+      "step": 163
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.632768361581922e-06,
+      "loss": 1.2789,
+      "step": 164
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.661016949152542e-06,
+      "loss": 1.2336,
+      "step": 165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.689265536723164e-06,
+      "loss": 1.2376,
+      "step": 166
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.717514124293785e-06,
+      "loss": 1.212,
+      "step": 167
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.745762711864407e-06,
+      "loss": 1.2483,
+      "step": 168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.774011299435029e-06,
+      "loss": 1.1397,
+      "step": 169
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.8022598870056505e-06,
+      "loss": 1.2432,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.830508474576272e-06,
+      "loss": 1.3325,
+      "step": 171
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.858757062146894e-06,
+      "loss": 1.2333,
+      "step": 172
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.887005649717514e-06,
+      "loss": 1.3392,
+      "step": 173
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.915254237288136e-06,
+      "loss": 1.1765,
+      "step": 174
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.943502824858757e-06,
+      "loss": 1.2127,
+      "step": 175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.971751412429379e-06,
+      "loss": 1.2286,
+      "step": 176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.000000000000001e-06,
+      "loss": 1.1191,
+      "step": 177
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.0282485875706224e-06,
+      "loss": 1.2334,
+      "step": 178
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.0564971751412425e-06,
+      "loss": 1.1833,
+      "step": 179
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.084745762711864e-06,
+      "loss": 1.2826,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.112994350282486e-06,
+      "loss": 1.2034,
+      "step": 181
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.141242937853108e-06,
+      "loss": 1.1629,
+      "step": 182
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.169491525423729e-06,
+      "loss": 1.1439,
+      "step": 183
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.197740112994351e-06,
+      "loss": 1.3088,
+      "step": 184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.225988700564973e-06,
+      "loss": 1.2005,
+      "step": 185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.254237288135594e-06,
+      "loss": 1.2861,
+      "step": 186
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.282485875706214e-06,
+      "loss": 1.3588,
+      "step": 187
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.310734463276836e-06,
+      "loss": 1.2553,
+      "step": 188
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.338983050847458e-06,
+      "loss": 1.1586,
+      "step": 189
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.3672316384180795e-06,
+      "loss": 1.2286,
+      "step": 190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.395480225988701e-06,
+      "loss": 1.0967,
+      "step": 191
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.423728813559323e-06,
+      "loss": 1.2553,
+      "step": 192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.451977401129945e-06,
+      "loss": 1.1946,
+      "step": 193
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.480225988700565e-06,
+      "loss": 1.3432,
+      "step": 194
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.508474576271186e-06,
+      "loss": 1.215,
+      "step": 195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.536723163841808e-06,
+      "loss": 1.2303,
+      "step": 196
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.56497175141243e-06,
+      "loss": 1.2081,
+      "step": 197
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.5932203389830515e-06,
+      "loss": 1.2244,
+      "step": 198
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.621468926553673e-06,
+      "loss": 1.2901,
+      "step": 199
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.649717514124295e-06,
+      "loss": 1.2581,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.677966101694917e-06,
+      "loss": 1.1412,
+      "step": 201
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.706214689265537e-06,
+      "loss": 1.1681,
+      "step": 202
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.734463276836158e-06,
+      "loss": 1.1455,
+      "step": 203
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.76271186440678e-06,
+      "loss": 1.1957,
+      "step": 204
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.790960451977402e-06,
+      "loss": 1.2205,
+      "step": 205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.819209039548023e-06,
+      "loss": 1.2916,
+      "step": 206
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.847457627118645e-06,
+      "loss": 1.2733,
+      "step": 207
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.875706214689267e-06,
+      "loss": 1.1752,
+      "step": 208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.903954802259889e-06,
+      "loss": 1.189,
+      "step": 209
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.932203389830509e-06,
+      "loss": 1.2037,
+      "step": 210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.96045197740113e-06,
+      "loss": 1.1258,
+      "step": 211
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.988700564971752e-06,
+      "loss": 1.2549,
+      "step": 212
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.016949152542372e-06,
+      "loss": 1.1728,
+      "step": 213
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.045197740112995e-06,
+      "loss": 1.1625,
+      "step": 214
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.073446327683617e-06,
+      "loss": 1.2757,
+      "step": 215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.101694915254239e-06,
+      "loss": 1.2613,
+      "step": 216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.129943502824859e-06,
+      "loss": 1.1722,
+      "step": 217
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.15819209039548e-06,
+      "loss": 1.2255,
+      "step": 218
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.186440677966102e-06,
+      "loss": 1.2496,
+      "step": 219
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.214689265536724e-06,
+      "loss": 1.3336,
+      "step": 220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.242937853107344e-06,
+      "loss": 1.2329,
+      "step": 221
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.271186440677966e-06,
+      "loss": 1.2192,
+      "step": 222
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.299435028248589e-06,
+      "loss": 1.1392,
+      "step": 223
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.32768361581921e-06,
+      "loss": 1.2563,
+      "step": 224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.35593220338983e-06,
+      "loss": 1.2376,
+      "step": 225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.384180790960452e-06,
+      "loss": 1.256,
+      "step": 226
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.412429378531074e-06,
+      "loss": 1.2667,
+      "step": 227
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.440677966101694e-06,
+      "loss": 1.1751,
+      "step": 228
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.468926553672316e-06,
+      "loss": 1.1624,
+      "step": 229
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.497175141242938e-06,
+      "loss": 1.1552,
+      "step": 230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.52542372881356e-06,
+      "loss": 1.2652,
+      "step": 231
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.553672316384181e-06,
+      "loss": 1.2127,
+      "step": 232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.581920903954803e-06,
+      "loss": 1.1883,
+      "step": 233
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.610169491525424e-06,
+      "loss": 1.2733,
+      "step": 234
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.638418079096046e-06,
+      "loss": 1.2906,
+      "step": 235
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.666666666666666e-06,
+      "loss": 1.2139,
+      "step": 236
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.694915254237288e-06,
+      "loss": 1.3526,
+      "step": 237
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.72316384180791e-06,
+      "loss": 1.2074,
+      "step": 238
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.751412429378531e-06,
+      "loss": 1.1678,
+      "step": 239
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.779661016949153e-06,
+      "loss": 1.2493,
+      "step": 240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.807909604519775e-06,
+      "loss": 1.2055,
+      "step": 241
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.836158192090396e-06,
+      "loss": 1.2013,
+      "step": 242
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.864406779661016e-06,
+      "loss": 1.3417,
+      "step": 243
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.892655367231638e-06,
+      "loss": 1.2848,
+      "step": 244
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.92090395480226e-06,
+      "loss": 1.3038,
+      "step": 245
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.949152542372881e-06,
+      "loss": 1.1888,
+      "step": 246
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.977401129943503e-06,
+      "loss": 1.1569,
+      "step": 247
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.005649717514125e-06,
+      "loss": 1.2284,
+      "step": 248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.033898305084747e-06,
+      "loss": 1.1808,
+      "step": 249
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.062146892655368e-06,
+      "loss": 1.2796,
+      "step": 250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.090395480225988e-06,
+      "loss": 1.2249,
+      "step": 251
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.11864406779661e-06,
+      "loss": 1.1821,
+      "step": 252
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.146892655367232e-06,
+      "loss": 1.2285,
+      "step": 253
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.175141242937853e-06,
+      "loss": 1.1967,
+      "step": 254
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.203389830508475e-06,
+      "loss": 1.2545,
+      "step": 255
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.231638418079097e-06,
+      "loss": 1.2288,
+      "step": 256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.259887005649719e-06,
+      "loss": 1.2052,
+      "step": 257
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.288135593220339e-06,
+      "loss": 1.1603,
+      "step": 258
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.31638418079096e-06,
+      "loss": 1.2647,
+      "step": 259
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.344632768361582e-06,
+      "loss": 1.2799,
+      "step": 260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.372881355932204e-06,
+      "loss": 1.3341,
+      "step": 261
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.401129943502825e-06,
+      "loss": 1.1452,
+      "step": 262
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.429378531073447e-06,
+      "loss": 1.2027,
+      "step": 263
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.457627118644069e-06,
+      "loss": 1.1708,
+      "step": 264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.48587570621469e-06,
+      "loss": 1.2239,
+      "step": 265
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.51412429378531e-06,
+      "loss": 1.1802,
+      "step": 266
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.542372881355932e-06,
+      "loss": 1.1572,
+      "step": 267
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.570621468926554e-06,
+      "loss": 1.2183,
+      "step": 268
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.598870056497176e-06,
+      "loss": 1.3114,
+      "step": 269
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.627118644067797e-06,
+      "loss": 1.2111,
+      "step": 270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.655367231638417e-06,
+      "loss": 1.2363,
+      "step": 271
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.683615819209039e-06,
+      "loss": 1.2103,
+      "step": 272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.711864406779662e-06,
+      "loss": 1.2004,
+      "step": 273
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.740112994350282e-06,
+      "loss": 1.2551,
+      "step": 274
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.768361581920904e-06,
+      "loss": 1.2384,
+      "step": 275
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.796610169491526e-06,
+      "loss": 1.2099,
+      "step": 276
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.824858757062146e-06,
+      "loss": 1.1463,
+      "step": 277
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.853107344632768e-06,
+      "loss": 1.2266,
+      "step": 278
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.88135593220339e-06,
+      "loss": 1.127,
+      "step": 279
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.909604519774011e-06,
+      "loss": 1.2857,
+      "step": 280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.937853107344633e-06,
+      "loss": 1.3142,
+      "step": 281
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.966101694915254e-06,
+      "loss": 1.2078,
+      "step": 282
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.994350282485876e-06,
+      "loss": 1.245,
+      "step": 283
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0022598870056498e-05,
+      "loss": 1.092,
+      "step": 284
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0050847457627118e-05,
+      "loss": 1.194,
+      "step": 285
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.007909604519774e-05,
+      "loss": 1.1918,
+      "step": 286
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0107344632768361e-05,
+      "loss": 1.2166,
+      "step": 287
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0135593220338983e-05,
+      "loss": 1.2799,
+      "step": 288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0163841807909605e-05,
+      "loss": 1.3147,
+      "step": 289
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0192090395480226e-05,
+      "loss": 1.2205,
+      "step": 290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0220338983050848e-05,
+      "loss": 1.3037,
+      "step": 291
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0248587570621468e-05,
+      "loss": 1.1909,
+      "step": 292
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.027683615819209e-05,
+      "loss": 1.2266,
+      "step": 293
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0305084745762712e-05,
+      "loss": 1.2753,
+      "step": 294
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0333333333333333e-05,
+      "loss": 1.1384,
+      "step": 295
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0361581920903955e-05,
+      "loss": 1.1889,
+      "step": 296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0389830508474577e-05,
+      "loss": 1.1909,
+      "step": 297
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0418079096045198e-05,
+      "loss": 1.1956,
+      "step": 298
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.044632768361582e-05,
+      "loss": 1.1715,
+      "step": 299
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.047457627118644e-05,
+      "loss": 1.2805,
+      "step": 300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0502824858757062e-05,
+      "loss": 1.1892,
+      "step": 301
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0531073446327683e-05,
+      "loss": 1.1837,
+      "step": 302
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0559322033898305e-05,
+      "loss": 1.2623,
+      "step": 303
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0587570621468927e-05,
+      "loss": 1.1505,
+      "step": 304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0615819209039549e-05,
+      "loss": 1.3072,
+      "step": 305
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.064406779661017e-05,
+      "loss": 1.2206,
+      "step": 306
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.067231638418079e-05,
+      "loss": 1.2726,
+      "step": 307
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0700564971751412e-05,
+      "loss": 1.1035,
+      "step": 308
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0728813559322034e-05,
+      "loss": 1.1339,
+      "step": 309
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0757062146892655e-05,
+      "loss": 1.2238,
+      "step": 310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0785310734463277e-05,
+      "loss": 1.1413,
+      "step": 311
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0813559322033899e-05,
+      "loss": 1.19,
+      "step": 312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.084180790960452e-05,
+      "loss": 1.1521,
+      "step": 313
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0870056497175142e-05,
+      "loss": 1.2121,
+      "step": 314
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0898305084745762e-05,
+      "loss": 1.2314,
+      "step": 315
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0926553672316384e-05,
+      "loss": 1.2593,
+      "step": 316
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0954802259887006e-05,
+      "loss": 1.2328,
+      "step": 317
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0983050847457627e-05,
+      "loss": 1.1297,
+      "step": 318
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1011299435028249e-05,
+      "loss": 1.194,
+      "step": 319
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.103954802259887e-05,
+      "loss": 1.2408,
+      "step": 320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1067796610169492e-05,
+      "loss": 1.3089,
+      "step": 321
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1096045197740112e-05,
+      "loss": 1.2581,
+      "step": 322
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1124293785310734e-05,
+      "loss": 1.1215,
+      "step": 323
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1152542372881356e-05,
+      "loss": 1.2453,
+      "step": 324
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1180790960451978e-05,
+      "loss": 1.2283,
+      "step": 325
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.12090395480226e-05,
+      "loss": 1.3633,
+      "step": 326
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1237288135593221e-05,
+      "loss": 1.2979,
+      "step": 327
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1265536723163843e-05,
+      "loss": 1.1819,
+      "step": 328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1293785310734464e-05,
+      "loss": 1.2214,
+      "step": 329
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1322033898305084e-05,
+      "loss": 1.0919,
+      "step": 330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1350282485875706e-05,
+      "loss": 1.2865,
+      "step": 331
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1378531073446328e-05,
+      "loss": 1.1804,
+      "step": 332
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.140677966101695e-05,
+      "loss": 1.1646,
+      "step": 333
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1435028248587571e-05,
+      "loss": 1.2135,
+      "step": 334
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1463276836158193e-05,
+      "loss": 1.1981,
+      "step": 335
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1491525423728815e-05,
+      "loss": 1.2072,
+      "step": 336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1519774011299436e-05,
+      "loss": 1.246,
+      "step": 337
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1548022598870056e-05,
+      "loss": 1.1662,
+      "step": 338
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1576271186440678e-05,
+      "loss": 1.1513,
+      "step": 339
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.16045197740113e-05,
+      "loss": 1.1162,
+      "step": 340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1632768361581922e-05,
+      "loss": 1.283,
+      "step": 341
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1661016949152543e-05,
+      "loss": 1.2172,
+      "step": 342
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1689265536723165e-05,
+      "loss": 1.2456,
+      "step": 343
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1717514124293787e-05,
+      "loss": 1.2978,
+      "step": 344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1745762711864407e-05,
+      "loss": 1.3128,
+      "step": 345
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1774011299435028e-05,
+      "loss": 1.1954,
+      "step": 346
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.180225988700565e-05,
+      "loss": 1.2243,
+      "step": 347
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1830508474576272e-05,
+      "loss": 1.2147,
+      "step": 348
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1858757062146892e-05,
+      "loss": 1.1303,
+      "step": 349
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1887005649717515e-05,
+      "loss": 1.2725,
+      "step": 350
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1915254237288137e-05,
+      "loss": 1.1153,
+      "step": 351
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1943502824858759e-05,
+      "loss": 1.2672,
+      "step": 352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1971751412429379e-05,
+      "loss": 1.1884,
+      "step": 353
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2e-05,
+      "loss": 1.0894,
+      "step": 354
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.202824858757062e-05,
+      "loss": 1.2366,
+      "step": 355
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2056497175141244e-05,
+      "loss": 1.2408,
+      "step": 356
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2084745762711864e-05,
+      "loss": 1.2729,
+      "step": 357
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2112994350282485e-05,
+      "loss": 1.1234,
+      "step": 358
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2141242937853109e-05,
+      "loss": 1.3314,
+      "step": 359
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2169491525423729e-05,
+      "loss": 1.2377,
+      "step": 360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2197740112994352e-05,
+      "loss": 1.1499,
+      "step": 361
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2225988700564972e-05,
+      "loss": 1.1771,
+      "step": 362
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2254237288135592e-05,
+      "loss": 1.2206,
+      "step": 363
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2282485875706216e-05,
+      "loss": 1.2576,
+      "step": 364
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2310734463276836e-05,
+      "loss": 1.1282,
+      "step": 365
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2338983050847457e-05,
+      "loss": 1.2592,
+      "step": 366
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2367231638418079e-05,
+      "loss": 1.2217,
+      "step": 367
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.23954802259887e-05,
+      "loss": 1.2116,
+      "step": 368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2423728813559324e-05,
+      "loss": 1.2157,
+      "step": 369
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2451977401129944e-05,
+      "loss": 1.1902,
+      "step": 370
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2480225988700564e-05,
+      "loss": 1.2785,
+      "step": 371
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2508474576271188e-05,
+      "loss": 1.2696,
+      "step": 372
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2536723163841808e-05,
+      "loss": 1.229,
+      "step": 373
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.256497175141243e-05,
+      "loss": 1.2189,
+      "step": 374
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2593220338983051e-05,
+      "loss": 1.2417,
+      "step": 375
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2621468926553673e-05,
+      "loss": 1.1883,
+      "step": 376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2649717514124293e-05,
+      "loss": 1.2409,
+      "step": 377
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2677966101694916e-05,
+      "loss": 1.2738,
+      "step": 378
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2706214689265536e-05,
+      "loss": 1.2076,
+      "step": 379
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.273446327683616e-05,
+      "loss": 1.1554,
+      "step": 380
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.276271186440678e-05,
+      "loss": 1.1499,
+      "step": 381
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2790960451977401e-05,
+      "loss": 1.2587,
+      "step": 382
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2819209039548023e-05,
+      "loss": 1.0356,
+      "step": 383
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2847457627118645e-05,
+      "loss": 1.1678,
+      "step": 384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2875706214689265e-05,
+      "loss": 1.1316,
+      "step": 385
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2903954802259888e-05,
+      "loss": 1.2196,
+      "step": 386
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2932203389830508e-05,
+      "loss": 1.2466,
+      "step": 387
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.296045197740113e-05,
+      "loss": 1.2566,
+      "step": 388
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2988700564971752e-05,
+      "loss": 1.1732,
+      "step": 389
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3016949152542373e-05,
+      "loss": 1.2952,
+      "step": 390
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3045197740112995e-05,
+      "loss": 1.2457,
+      "step": 391
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3073446327683617e-05,
+      "loss": 1.1791,
+      "step": 392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3101694915254237e-05,
+      "loss": 1.1576,
+      "step": 393
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.312994350282486e-05,
+      "loss": 1.1417,
+      "step": 394
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.315819209039548e-05,
+      "loss": 1.2494,
+      "step": 395
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3186440677966102e-05,
+      "loss": 1.2586,
+      "step": 396
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3214689265536723e-05,
+      "loss": 1.1838,
+      "step": 397
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3242937853107345e-05,
+      "loss": 1.2237,
+      "step": 398
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3271186440677967e-05,
+      "loss": 1.222,
+      "step": 399
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3299435028248589e-05,
+      "loss": 1.2455,
+      "step": 400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3327683615819209e-05,
+      "loss": 1.2128,
+      "step": 401
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3355932203389832e-05,
+      "loss": 1.2037,
+      "step": 402
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3384180790960452e-05,
+      "loss": 1.1472,
+      "step": 403
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3412429378531074e-05,
+      "loss": 1.1799,
+      "step": 404
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3440677966101695e-05,
+      "loss": 1.2762,
+      "step": 405
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3468926553672317e-05,
+      "loss": 1.1657,
+      "step": 406
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3497175141242937e-05,
+      "loss": 1.2312,
+      "step": 407
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.352542372881356e-05,
+      "loss": 1.1721,
+      "step": 408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.355367231638418e-05,
+      "loss": 1.2752,
+      "step": 409
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3581920903954804e-05,
+      "loss": 1.2586,
+      "step": 410
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3610169491525424e-05,
+      "loss": 1.2071,
+      "step": 411
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3638418079096046e-05,
+      "loss": 1.2295,
+      "step": 412
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3666666666666667e-05,
+      "loss": 1.2373,
+      "step": 413
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3694915254237289e-05,
+      "loss": 1.2281,
+      "step": 414
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3723163841807909e-05,
+      "loss": 1.2505,
+      "step": 415
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3751412429378533e-05,
+      "loss": 1.262,
+      "step": 416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3779661016949153e-05,
+      "loss": 1.2398,
+      "step": 417
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3807909604519776e-05,
+      "loss": 1.2118,
+      "step": 418
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3836158192090396e-05,
+      "loss": 1.2469,
+      "step": 419
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3864406779661018e-05,
+      "loss": 1.0644,
+      "step": 420
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.389265536723164e-05,
+      "loss": 1.2273,
+      "step": 421
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3920903954802261e-05,
+      "loss": 1.2483,
+      "step": 422
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3949152542372881e-05,
+      "loss": 1.1994,
+      "step": 423
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3977401129943504e-05,
+      "loss": 1.1006,
+      "step": 424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4005649717514124e-05,
+      "loss": 1.1728,
+      "step": 425
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4033898305084744e-05,
+      "loss": 1.2448,
+      "step": 426
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4062146892655368e-05,
+      "loss": 1.2404,
+      "step": 427
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.409039548022599e-05,
+      "loss": 1.2351,
+      "step": 428
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4118644067796611e-05,
+      "loss": 1.2767,
+      "step": 429
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4146892655367233e-05,
+      "loss": 1.2716,
+      "step": 430
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4175141242937853e-05,
+      "loss": 1.1175,
+      "step": 431
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4203389830508476e-05,
+      "loss": 1.223,
+      "step": 432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4231638418079096e-05,
+      "loss": 1.2931,
+      "step": 433
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4259887005649716e-05,
+      "loss": 1.2082,
+      "step": 434
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.428813559322034e-05,
+      "loss": 1.2985,
+      "step": 435
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4316384180790962e-05,
+      "loss": 1.2196,
+      "step": 436
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4344632768361582e-05,
+      "loss": 1.1933,
+      "step": 437
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4372881355932205e-05,
+      "loss": 1.2205,
+      "step": 438
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4401129943502825e-05,
+      "loss": 1.261,
+      "step": 439
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4429378531073448e-05,
+      "loss": 1.2783,
+      "step": 440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4457627118644068e-05,
+      "loss": 1.3303,
+      "step": 441
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4485875706214688e-05,
+      "loss": 1.2318,
+      "step": 442
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4514124293785312e-05,
+      "loss": 1.1439,
+      "step": 443
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4542372881355932e-05,
+      "loss": 1.2758,
+      "step": 444
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4570621468926554e-05,
+      "loss": 1.214,
+      "step": 445
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4598870056497177e-05,
+      "loss": 1.1787,
+      "step": 446
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4627118644067797e-05,
+      "loss": 1.1341,
+      "step": 447
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.465536723163842e-05,
+      "loss": 1.2708,
+      "step": 448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.468361581920904e-05,
+      "loss": 1.1415,
+      "step": 449
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.471186440677966e-05,
+      "loss": 1.1891,
+      "step": 450
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4740112994350284e-05,
+      "loss": 1.2975,
+      "step": 451
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4768361581920904e-05,
+      "loss": 1.2233,
+      "step": 452
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4796610169491525e-05,
+      "loss": 1.2743,
+      "step": 453
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4824858757062149e-05,
+      "loss": 1.2496,
+      "step": 454
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4853107344632769e-05,
+      "loss": 1.2677,
+      "step": 455
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4881355932203389e-05,
+      "loss": 1.2703,
+      "step": 456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4909604519774012e-05,
+      "loss": 1.235,
+      "step": 457
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4937853107344632e-05,
+      "loss": 1.3033,
+      "step": 458
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4966101694915256e-05,
+      "loss": 1.2094,
+      "step": 459
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4994350282485876e-05,
+      "loss": 1.3131,
+      "step": 460
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5022598870056497e-05,
+      "loss": 1.1532,
+      "step": 461
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5050847457627119e-05,
+      "loss": 1.3309,
+      "step": 462
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.507909604519774e-05,
+      "loss": 1.2021,
+      "step": 463
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.510734463276836e-05,
+      "loss": 1.2279,
+      "step": 464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5135593220338984e-05,
+      "loss": 1.168,
+      "step": 465
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5163841807909604e-05,
+      "loss": 1.1626,
+      "step": 466
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5192090395480226e-05,
+      "loss": 1.2612,
+      "step": 467
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5220338983050848e-05,
+      "loss": 1.3576,
+      "step": 468
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.524858757062147e-05,
+      "loss": 1.2825,
+      "step": 469
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.527683615819209e-05,
+      "loss": 1.1959,
+      "step": 470
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5305084745762714e-05,
+      "loss": 1.1271,
+      "step": 471
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5333333333333334e-05,
+      "loss": 1.25,
+      "step": 472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5361581920903958e-05,
+      "loss": 1.1935,
+      "step": 473
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5389830508474578e-05,
+      "loss": 1.3119,
+      "step": 474
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5418079096045198e-05,
+      "loss": 1.2977,
+      "step": 475
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.544632768361582e-05,
+      "loss": 1.1371,
+      "step": 476
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.547457627118644e-05,
+      "loss": 1.2944,
+      "step": 477
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5502824858757065e-05,
+      "loss": 1.2322,
+      "step": 478
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5531073446327685e-05,
+      "loss": 1.1134,
+      "step": 479
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5559322033898305e-05,
+      "loss": 1.2193,
+      "step": 480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5587570621468928e-05,
+      "loss": 1.3061,
+      "step": 481
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5615819209039548e-05,
+      "loss": 1.1969,
+      "step": 482
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.564406779661017e-05,
+      "loss": 1.1489,
+      "step": 483
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.567231638418079e-05,
+      "loss": 1.276,
+      "step": 484
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5700564971751415e-05,
+      "loss": 1.2141,
+      "step": 485
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5728813559322035e-05,
+      "loss": 1.1993,
+      "step": 486
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.575706214689266e-05,
+      "loss": 1.2296,
+      "step": 487
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.578531073446328e-05,
+      "loss": 1.174,
+      "step": 488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5813559322033902e-05,
+      "loss": 1.2634,
+      "step": 489
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5841807909604522e-05,
+      "loss": 1.2316,
+      "step": 490
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5870056497175142e-05,
+      "loss": 1.2685,
+      "step": 491
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5898305084745765e-05,
+      "loss": 1.2093,
+      "step": 492
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5926553672316385e-05,
+      "loss": 1.227,
+      "step": 493
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5954802259887005e-05,
+      "loss": 1.2928,
+      "step": 494
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.598305084745763e-05,
+      "loss": 1.2997,
+      "step": 495
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.601129943502825e-05,
+      "loss": 1.187,
+      "step": 496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6039548022598872e-05,
+      "loss": 1.2326,
+      "step": 497
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6067796610169492e-05,
+      "loss": 1.2544,
+      "step": 498
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6096045197740115e-05,
+      "loss": 1.3099,
+      "step": 499
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6124293785310735e-05,
+      "loss": 1.1836,
+      "step": 500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.615254237288136e-05,
+      "loss": 1.1687,
+      "step": 501
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.618079096045198e-05,
+      "loss": 1.1919,
+      "step": 502
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6209039548022602e-05,
+      "loss": 1.1656,
+      "step": 503
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6237288135593222e-05,
+      "loss": 1.3453,
+      "step": 504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6265536723163842e-05,
+      "loss": 1.1706,
+      "step": 505
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6293785310734466e-05,
+      "loss": 1.1588,
+      "step": 506
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6322033898305086e-05,
+      "loss": 1.1319,
+      "step": 507
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.635028248587571e-05,
+      "loss": 1.2318,
+      "step": 508
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.637853107344633e-05,
+      "loss": 1.2052,
+      "step": 509
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.640677966101695e-05,
+      "loss": 1.1926,
+      "step": 510
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6435028248587573e-05,
+      "loss": 1.2095,
+      "step": 511
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6463276836158193e-05,
+      "loss": 1.2914,
+      "step": 512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6491525423728813e-05,
+      "loss": 1.2039,
+      "step": 513
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6519774011299436e-05,
+      "loss": 1.1502,
+      "step": 514
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.654802259887006e-05,
+      "loss": 1.2482,
+      "step": 515
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.657627118644068e-05,
+      "loss": 1.2557,
+      "step": 516
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6604519774011303e-05,
+      "loss": 1.2006,
+      "step": 517
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6632768361581923e-05,
+      "loss": 1.2077,
+      "step": 518
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6661016949152546e-05,
+      "loss": 1.3086,
+      "step": 519
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6689265536723166e-05,
+      "loss": 1.241,
+      "step": 520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6717514124293786e-05,
+      "loss": 1.2416,
+      "step": 521
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.674576271186441e-05,
+      "loss": 1.2089,
+      "step": 522
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.677401129943503e-05,
+      "loss": 1.2181,
+      "step": 523
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.680225988700565e-05,
+      "loss": 1.2573,
+      "step": 524
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6830508474576273e-05,
+      "loss": 1.1772,
+      "step": 525
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6858757062146893e-05,
+      "loss": 1.2556,
+      "step": 526
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6887005649717516e-05,
+      "loss": 1.1669,
+      "step": 527
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6915254237288136e-05,
+      "loss": 1.2715,
+      "step": 528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6943502824858756e-05,
+      "loss": 1.1193,
+      "step": 529
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.697175141242938e-05,
+      "loss": 1.2474,
+      "step": 530
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7000000000000003e-05,
+      "loss": 1.2445,
+      "step": 531
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7028248587570623e-05,
+      "loss": 1.209,
+      "step": 532
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7056497175141247e-05,
+      "loss": 1.2343,
+      "step": 533
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7084745762711867e-05,
+      "loss": 1.2938,
+      "step": 534
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7112994350282487e-05,
+      "loss": 1.272,
+      "step": 535
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.714124293785311e-05,
+      "loss": 1.1898,
+      "step": 536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.716949152542373e-05,
+      "loss": 1.1886,
+      "step": 537
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7197740112994354e-05,
+      "loss": 1.2472,
+      "step": 538
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7225988700564974e-05,
+      "loss": 1.2339,
+      "step": 539
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7254237288135594e-05,
+      "loss": 1.2226,
+      "step": 540
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7282485875706217e-05,
+      "loss": 1.2387,
+      "step": 541
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7310734463276837e-05,
+      "loss": 1.0801,
+      "step": 542
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.733898305084746e-05,
+      "loss": 1.2953,
+      "step": 543
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.736723163841808e-05,
+      "loss": 1.0647,
+      "step": 544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7395480225988704e-05,
+      "loss": 1.1584,
+      "step": 545
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7423728813559327e-05,
+      "loss": 1.1376,
+      "step": 546
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7451977401129947e-05,
+      "loss": 1.1649,
+      "step": 547
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7480225988700567e-05,
+      "loss": 1.2049,
+      "step": 548
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.750847457627119e-05,
+      "loss": 1.1919,
+      "step": 549
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.753672316384181e-05,
+      "loss": 1.1505,
+      "step": 550
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.756497175141243e-05,
+      "loss": 1.2195,
+      "step": 551
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7593220338983054e-05,
+      "loss": 1.3336,
+      "step": 552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7621468926553674e-05,
+      "loss": 1.126,
+      "step": 553
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7649717514124294e-05,
+      "loss": 1.2666,
+      "step": 554
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7677966101694917e-05,
+      "loss": 1.1996,
+      "step": 555
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7706214689265537e-05,
+      "loss": 1.1214,
+      "step": 556
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.773446327683616e-05,
+      "loss": 1.2596,
+      "step": 557
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.776271186440678e-05,
+      "loss": 1.2539,
+      "step": 558
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7790960451977404e-05,
+      "loss": 1.1429,
+      "step": 559
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7819209039548024e-05,
+      "loss": 1.175,
+      "step": 560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7847457627118648e-05,
+      "loss": 1.1889,
+      "step": 561
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7875706214689268e-05,
+      "loss": 1.1845,
+      "step": 562
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.790395480225989e-05,
+      "loss": 1.2101,
+      "step": 563
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.793220338983051e-05,
+      "loss": 1.2379,
+      "step": 564
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.796045197740113e-05,
+      "loss": 1.1901,
+      "step": 565
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7988700564971754e-05,
+      "loss": 1.1882,
+      "step": 566
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8016949152542374e-05,
+      "loss": 1.0944,
+      "step": 567
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8045197740112998e-05,
+      "loss": 1.2571,
+      "step": 568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8073446327683618e-05,
+      "loss": 1.1504,
+      "step": 569
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8101694915254238e-05,
+      "loss": 1.2277,
+      "step": 570
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.812994350282486e-05,
+      "loss": 1.2223,
+      "step": 571
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.815819209039548e-05,
+      "loss": 1.1753,
+      "step": 572
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8186440677966105e-05,
+      "loss": 1.0824,
+      "step": 573
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8214689265536725e-05,
+      "loss": 1.2414,
+      "step": 574
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8242937853107348e-05,
+      "loss": 1.205,
+      "step": 575
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8271186440677968e-05,
+      "loss": 1.249,
+      "step": 576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.829943502824859e-05,
+      "loss": 1.2565,
+      "step": 577
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.832768361581921e-05,
+      "loss": 1.2552,
+      "step": 578
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8355932203389835e-05,
+      "loss": 1.2511,
+      "step": 579
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8384180790960455e-05,
+      "loss": 1.141,
+      "step": 580
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8412429378531075e-05,
+      "loss": 1.2939,
+      "step": 581
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.84406779661017e-05,
+      "loss": 1.2428,
+      "step": 582
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.846892655367232e-05,
+      "loss": 1.2558,
+      "step": 583
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.849717514124294e-05,
+      "loss": 1.2125,
+      "step": 584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8525423728813562e-05,
+      "loss": 1.2165,
+      "step": 585
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8553672316384182e-05,
+      "loss": 1.1636,
+      "step": 586
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8581920903954805e-05,
+      "loss": 1.1946,
+      "step": 587
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8610169491525425e-05,
+      "loss": 1.16,
+      "step": 588
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.863841807909605e-05,
+      "loss": 1.1937,
+      "step": 589
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.866666666666667e-05,
+      "loss": 1.175,
+      "step": 590
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8694915254237292e-05,
+      "loss": 1.1894,
+      "step": 591
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8723163841807912e-05,
+      "loss": 1.2307,
+      "step": 592
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8751412429378535e-05,
+      "loss": 1.2196,
+      "step": 593
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8779661016949155e-05,
+      "loss": 1.257,
+      "step": 594
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.880790960451978e-05,
+      "loss": 1.2361,
+      "step": 595
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.88361581920904e-05,
+      "loss": 1.2619,
+      "step": 596
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.886440677966102e-05,
+      "loss": 1.1457,
+      "step": 597
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8892655367231642e-05,
+      "loss": 1.2266,
+      "step": 598
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8920903954802262e-05,
+      "loss": 1.1074,
+      "step": 599
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8949152542372882e-05,
+      "loss": 1.2231,
+      "step": 600
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8977401129943506e-05,
+      "loss": 1.2164,
+      "step": 601
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9005649717514126e-05,
+      "loss": 1.1641,
+      "step": 602
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9033898305084746e-05,
+      "loss": 1.2387,
+      "step": 603
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.906214689265537e-05,
+      "loss": 1.2538,
+      "step": 604
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9090395480225993e-05,
+      "loss": 1.1978,
+      "step": 605
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9118644067796613e-05,
+      "loss": 1.1397,
+      "step": 606
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9146892655367236e-05,
+      "loss": 1.278,
+      "step": 607
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9175141242937856e-05,
+      "loss": 1.1987,
+      "step": 608
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.920338983050848e-05,
+      "loss": 1.2797,
+      "step": 609
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.92316384180791e-05,
+      "loss": 1.3074,
+      "step": 610
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.925988700564972e-05,
+      "loss": 1.254,
+      "step": 611
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9288135593220343e-05,
+      "loss": 1.1426,
+      "step": 612
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9316384180790963e-05,
+      "loss": 1.224,
+      "step": 613
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9344632768361583e-05,
+      "loss": 1.1746,
+      "step": 614
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9372881355932206e-05,
+      "loss": 1.1449,
+      "step": 615
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9401129943502826e-05,
+      "loss": 1.126,
+      "step": 616
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.942937853107345e-05,
+      "loss": 1.216,
+      "step": 617
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.945762711864407e-05,
+      "loss": 1.1506,
+      "step": 618
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.948587570621469e-05,
+      "loss": 1.2073,
+      "step": 619
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9514124293785313e-05,
+      "loss": 1.274,
+      "step": 620
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9542372881355933e-05,
+      "loss": 1.1699,
+      "step": 621
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9570621468926556e-05,
+      "loss": 1.2035,
+      "step": 622
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.959887005649718e-05,
+      "loss": 1.1475,
+      "step": 623
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.96271186440678e-05,
+      "loss": 1.131,
+      "step": 624
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9655367231638423e-05,
+      "loss": 1.2577,
+      "step": 625
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9683615819209043e-05,
+      "loss": 1.2101,
+      "step": 626
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9711864406779663e-05,
+      "loss": 1.1142,
+      "step": 627
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9740112994350287e-05,
+      "loss": 1.2876,
+      "step": 628
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9768361581920907e-05,
+      "loss": 1.2596,
+      "step": 629
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9796610169491527e-05,
+      "loss": 1.1956,
+      "step": 630
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.982485875706215e-05,
+      "loss": 1.2957,
+      "step": 631
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.985310734463277e-05,
+      "loss": 1.3016,
+      "step": 632
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.988135593220339e-05,
+      "loss": 1.2351,
+      "step": 633
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9909604519774014e-05,
+      "loss": 1.176,
+      "step": 634
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9937853107344634e-05,
+      "loss": 1.3087,
+      "step": 635
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9966101694915257e-05,
+      "loss": 1.1995,
+      "step": 636
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9994350282485877e-05,
+      "loss": 1.1857,
+      "step": 637
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.00225988700565e-05,
+      "loss": 1.2739,
+      "step": 638
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.005084745762712e-05,
+      "loss": 1.2375,
+      "step": 639
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0079096045197744e-05,
+      "loss": 1.1837,
+      "step": 640
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0107344632768364e-05,
+      "loss": 1.2321,
+      "step": 641
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0135593220338987e-05,
+      "loss": 1.2335,
+      "step": 642
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0163841807909607e-05,
+      "loss": 1.1431,
+      "step": 643
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0192090395480227e-05,
+      "loss": 1.1988,
+      "step": 644
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.022033898305085e-05,
+      "loss": 1.1808,
+      "step": 645
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.024858757062147e-05,
+      "loss": 1.1826,
+      "step": 646
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0276836158192094e-05,
+      "loss": 1.2077,
+      "step": 647
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0305084745762714e-05,
+      "loss": 1.1779,
+      "step": 648
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0333333333333334e-05,
+      "loss": 1.2599,
+      "step": 649
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0361581920903957e-05,
+      "loss": 1.1849,
+      "step": 650
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0389830508474577e-05,
+      "loss": 1.2507,
+      "step": 651
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.04180790960452e-05,
+      "loss": 1.2924,
+      "step": 652
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.044632768361582e-05,
+      "loss": 1.131,
+      "step": 653
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0474576271186444e-05,
+      "loss": 1.2332,
+      "step": 654
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0502824858757064e-05,
+      "loss": 1.1967,
+      "step": 655
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0531073446327688e-05,
+      "loss": 1.1875,
+      "step": 656
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0559322033898308e-05,
+      "loss": 1.1448,
+      "step": 657
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.058757062146893e-05,
+      "loss": 1.2123,
+      "step": 658
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.061581920903955e-05,
+      "loss": 1.2782,
+      "step": 659
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.064406779661017e-05,
+      "loss": 1.2958,
+      "step": 660
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0672316384180795e-05,
+      "loss": 1.2373,
+      "step": 661
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0700564971751415e-05,
+      "loss": 1.2368,
+      "step": 662
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0728813559322035e-05,
+      "loss": 1.1685,
+      "step": 663
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0757062146892658e-05,
+      "loss": 1.2174,
+      "step": 664
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0785310734463278e-05,
+      "loss": 1.2247,
+      "step": 665
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.08135593220339e-05,
+      "loss": 1.1887,
+      "step": 666
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.084180790960452e-05,
+      "loss": 1.167,
+      "step": 667
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0870056497175145e-05,
+      "loss": 1.1729,
+      "step": 668
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0898305084745765e-05,
+      "loss": 1.2172,
+      "step": 669
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0926553672316388e-05,
+      "loss": 1.1623,
+      "step": 670
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0954802259887008e-05,
+      "loss": 1.1685,
+      "step": 671
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.098305084745763e-05,
+      "loss": 1.3016,
+      "step": 672
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.101129943502825e-05,
+      "loss": 1.1956,
+      "step": 673
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1039548022598875e-05,
+      "loss": 1.3237,
+      "step": 674
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1067796610169495e-05,
+      "loss": 1.2465,
+      "step": 675
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1096045197740115e-05,
+      "loss": 1.265,
+      "step": 676
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.112429378531074e-05,
+      "loss": 1.1298,
+      "step": 677
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.115254237288136e-05,
+      "loss": 1.2026,
+      "step": 678
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.118079096045198e-05,
+      "loss": 1.2764,
+      "step": 679
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1209039548022602e-05,
+      "loss": 1.2149,
+      "step": 680
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1237288135593222e-05,
+      "loss": 1.0931,
+      "step": 681
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1265536723163845e-05,
+      "loss": 1.1755,
+      "step": 682
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1293785310734465e-05,
+      "loss": 1.2595,
+      "step": 683
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.132203389830509e-05,
+      "loss": 1.2304,
+      "step": 684
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.135028248587571e-05,
+      "loss": 1.1991,
+      "step": 685
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1378531073446332e-05,
+      "loss": 1.2137,
+      "step": 686
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1406779661016952e-05,
+      "loss": 1.3253,
+      "step": 687
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1435028248587575e-05,
+      "loss": 1.1001,
+      "step": 688
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1463276836158195e-05,
+      "loss": 1.219,
+      "step": 689
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1491525423728816e-05,
+      "loss": 1.1336,
+      "step": 690
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.151977401129944e-05,
+      "loss": 1.141,
+      "step": 691
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.154802259887006e-05,
+      "loss": 1.1158,
+      "step": 692
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.157627118644068e-05,
+      "loss": 1.2134,
+      "step": 693
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1604519774011302e-05,
+      "loss": 1.2662,
+      "step": 694
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1632768361581922e-05,
+      "loss": 1.209,
+      "step": 695
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1661016949152546e-05,
+      "loss": 1.1936,
+      "step": 696
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1689265536723166e-05,
+      "loss": 1.2045,
+      "step": 697
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1717514124293786e-05,
+      "loss": 1.1951,
+      "step": 698
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.174576271186441e-05,
+      "loss": 1.1705,
+      "step": 699
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1774011299435033e-05,
+      "loss": 1.2538,
+      "step": 700
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1802259887005653e-05,
+      "loss": 1.2074,
+      "step": 701
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1830508474576276e-05,
+      "loss": 1.1348,
+      "step": 702
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1858757062146896e-05,
+      "loss": 1.1338,
+      "step": 703
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.188700564971752e-05,
+      "loss": 1.2562,
+      "step": 704
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.191525423728814e-05,
+      "loss": 1.2741,
+      "step": 705
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.194350282485876e-05,
+      "loss": 1.2316,
+      "step": 706
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1971751412429383e-05,
+      "loss": 1.196,
+      "step": 707
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 1.1698,
+      "step": 708
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999999905815777e-05,
+      "loss": 1.2279,
+      "step": 709
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199999962326311e-05,
+      "loss": 1.2393,
+      "step": 710
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999999152341996e-05,
+      "loss": 1.1232,
+      "step": 711
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999998493052455e-05,
+      "loss": 1.1218,
+      "step": 712
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999997645394492e-05,
+      "loss": 1.1987,
+      "step": 713
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999996609368125e-05,
+      "loss": 1.2684,
+      "step": 714
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999995384973377e-05,
+      "loss": 1.3091,
+      "step": 715
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999993972210265e-05,
+      "loss": 1.1622,
+      "step": 716
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199999237107882e-05,
+      "loss": 1.2487,
+      "step": 717
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199999058157907e-05,
+      "loss": 1.214,
+      "step": 718
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199998860371105e-05,
+      "loss": 1.2239,
+      "step": 719
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.19999864374748e-05,
+      "loss": 1.1352,
+      "step": 720
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199998408287035e-05,
+      "loss": 1.1671,
+      "step": 721
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999981539897758e-05,
+      "loss": 1.1311,
+      "step": 722
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999978808557064e-05,
+      "loss": 1.2407,
+      "step": 723
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999975888848322e-05,
+      "loss": 1.214,
+      "step": 724
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999972780771585e-05,
+      "loss": 1.179,
+      "step": 725
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199996948432691e-05,
+      "loss": 1.1573,
+      "step": 726
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199996599951437e-05,
+      "loss": 1.2252,
+      "step": 727
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999962326334017e-05,
+      "loss": 1.1905,
+      "step": 728
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999958464785926e-05,
+      "loss": 1.2939,
+      "step": 729
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199995441487017e-05,
+      "loss": 1.2701,
+      "step": 730
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999950176586826e-05,
+      "loss": 1.1471,
+      "step": 731
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199994574993597e-05,
+      "loss": 1.2098,
+      "step": 732
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999941134917693e-05,
+      "loss": 1.2431,
+      "step": 733
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999936331532068e-05,
+      "loss": 1.177,
+      "step": 734
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999931339779204e-05,
+      "loss": 1.2232,
+      "step": 735
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199992615965918e-05,
+      "loss": 1.2104,
+      "step": 736
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.19999207911721e-05,
+      "loss": 1.2272,
+      "step": 737
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999915234318065e-05,
+      "loss": 1.1485,
+      "step": 738
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999909489097177e-05,
+      "loss": 1.2578,
+      "step": 739
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999903555509547e-05,
+      "loss": 1.2071,
+      "step": 740
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199989743355529e-05,
+      "loss": 1.29,
+      "step": 741
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199989112323451e-05,
+      "loss": 1.2811,
+      "step": 742
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999884624547335e-05,
+      "loss": 1.2696,
+      "step": 743
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999877937493887e-05,
+      "loss": 1.2141,
+      "step": 744
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199987106207429e-05,
+      "loss": 1.1927,
+      "step": 745
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999863998288676e-05,
+      "loss": 1.1206,
+      "step": 746
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999856746137173e-05,
+      "loss": 1.1976,
+      "step": 747
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199984930561992e-05,
+      "loss": 1.1768,
+      "step": 748
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199984167673706e-05,
+      "loss": 1.2322,
+      "step": 749
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999833859488732e-05,
+      "loss": 1.2444,
+      "step": 750
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999825853875087e-05,
+      "loss": 1.3297,
+      "step": 751
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999817659896277e-05,
+      "loss": 1.2007,
+      "step": 752
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999809277552446e-05,
+      "loss": 1.1792,
+      "step": 753
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999800706843766e-05,
+      "loss": 1.2574,
+      "step": 754
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999791947770394e-05,
+      "loss": 1.2575,
+      "step": 755
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999783000332487e-05,
+      "loss": 1.1874,
+      "step": 756
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999773864530227e-05,
+      "loss": 1.2364,
+      "step": 757
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999764540363775e-05,
+      "loss": 1.1674,
+      "step": 758
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199975502783331e-05,
+      "loss": 1.3079,
+      "step": 759
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199974532693901e-05,
+      "loss": 1.2596,
+      "step": 760
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199973543768106e-05,
+      "loss": 1.1792,
+      "step": 761
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199972536005965e-05,
+      "loss": 1.1879,
+      "step": 762
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199971509407496e-05,
+      "loss": 1.253,
+      "step": 763
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199970463972719e-05,
+      "loss": 1.2902,
+      "step": 764
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999693997016536e-05,
+      "loss": 1.1792,
+      "step": 765
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.19996831659432e-05,
+      "loss": 1.2257,
+      "step": 766
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199967214650738e-05,
+      "loss": 1.1361,
+      "step": 767
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199966093870929e-05,
+      "loss": 1.0828,
+      "step": 768
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999649542549142e-05,
+      "loss": 1.1769,
+      "step": 769
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999637958027145e-05,
+      "loss": 1.1758,
+      "step": 770
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199962618514352e-05,
+      "loss": 1.2312,
+      "step": 771
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199961422389849e-05,
+      "loss": 1.1971,
+      "step": 772
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999602074292278e-05,
+      "loss": 1.2244,
+      "step": 773
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999589736325113e-05,
+      "loss": 1.1583,
+      "step": 774
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199957720999723e-05,
+      "loss": 1.24,
+      "step": 775
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999564495308862e-05,
+      "loss": 1.16,
+      "step": 776
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199955159226025e-05,
+      "loss": 1.2146,
+      "step": 777
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999538500851634e-05,
+      "loss": 1.2061,
+      "step": 778
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999525221083267e-05,
+      "loss": 1.1888,
+      "step": 779
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199951175295539e-05,
+      "loss": 1.1912,
+      "step": 780
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999498096468267e-05,
+      "loss": 1.1592,
+      "step": 781
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199948425162215e-05,
+      "loss": 1.1579,
+      "step": 782
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.19994702184173e-05,
+      "loss": 1.1285,
+      "step": 783
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999455996853982e-05,
+      "loss": 1.3773,
+      "step": 784
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199944158693246e-05,
+      "loss": 1.1743,
+      "step": 785
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999426988653013e-05,
+      "loss": 1.2377,
+      "step": 786
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199941220201591e-05,
+      "loss": 1.1013,
+      "step": 787
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199939722702143e-05,
+      "loss": 1.18,
+      "step": 788
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999382063669856e-05,
+      "loss": 1.1752,
+      "step": 789
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999366711961477e-05,
+      "loss": 1.1135,
+      "step": 790
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199935117189658e-05,
+      "loss": 1.2156,
+      "step": 791
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999335443475454e-05,
+      "loss": 1.1334,
+      "step": 792
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.19993195266984e-05,
+      "loss": 1.2129,
+      "step": 793
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999303421565716e-05,
+      "loss": 1.1993,
+      "step": 794
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999287128077703e-05,
+      "loss": 1.2881,
+      "step": 795
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999270646234672e-05,
+      "loss": 1.2329,
+      "step": 796
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999253976036932e-05,
+      "loss": 1.1159,
+      "step": 797
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.19992371174848e-05,
+      "loss": 1.2033,
+      "step": 798
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999220070578582e-05,
+      "loss": 1.2274,
+      "step": 799
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999202835318614e-05,
+      "loss": 1.31,
+      "step": 800
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999185411705216e-05,
+      "loss": 1.1184,
+      "step": 801
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999167799738708e-05,
+      "loss": 1.182,
+      "step": 802
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999149999419434e-05,
+      "loss": 1.2052,
+      "step": 803
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199913201074772e-05,
+      "loss": 1.2054,
+      "step": 804
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999113833723908e-05,
+      "loss": 1.1501,
+      "step": 805
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999095468348342e-05,
+      "loss": 1.1257,
+      "step": 806
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999076914621367e-05,
+      "loss": 1.1368,
+      "step": 807
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199905817254333e-05,
+      "loss": 1.2343,
+      "step": 808
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999039242114587e-05,
+      "loss": 1.1541,
+      "step": 809
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999020123335497e-05,
+      "loss": 1.1463,
+      "step": 810
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999000816206414e-05,
+      "loss": 1.1974,
+      "step": 811
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1998981320727703e-05,
+      "loss": 1.1898,
+      "step": 812
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1998961636899737e-05,
+      "loss": 1.1898,
+      "step": 813
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1998941764722876e-05,
+      "loss": 1.19,
+      "step": 814
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1998921704197505e-05,
+      "loss": 1.1649,
+      "step": 815
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1998901455324e-05,
+      "loss": 1.2276,
+      "step": 816
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1998881018102736e-05,
+      "loss": 1.1869,
+      "step": 817
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1998860392534107e-05,
+      "loss": 1.1955,
+      "step": 818
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1998839578618493e-05,
+      "loss": 1.2026,
+      "step": 819
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1998818576356287e-05,
+      "loss": 1.1217,
+      "step": 820
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1998797385747893e-05,
+      "loss": 1.1164,
+      "step": 821
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1998776006793704e-05,
+      "loss": 1.1637,
+      "step": 822
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199875443949412e-05,
+      "loss": 1.2215,
+      "step": 823
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1998732683849554e-05,
+      "loss": 1.2012,
+      "step": 824
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199871073986041e-05,
+      "loss": 1.128,
+      "step": 825
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1998688607527104e-05,
+      "loss": 1.2481,
+      "step": 826
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1998666286850052e-05,
+      "loss": 1.2241,
+      "step": 827
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1998643777829675e-05,
+      "loss": 1.1854,
+      "step": 828
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.19986210804664e-05,
+      "loss": 1.1001,
+      "step": 829
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1998598194760648e-05,
+      "loss": 1.1635,
+      "step": 830
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1998575120712855e-05,
+      "loss": 1.1002,
+      "step": 831
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1998551858323455e-05,
+      "loss": 1.1653,
+      "step": 832
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1998528407592884e-05,
+      "loss": 1.1858,
+      "step": 833
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199850476852159e-05,
+      "loss": 1.1446,
+      "step": 834
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1998480941110008e-05,
+      "loss": 1.1452,
+      "step": 835
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199845692535859e-05,
+      "loss": 1.1875,
+      "step": 836
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.19984327212678e-05,
+      "loss": 1.2344,
+      "step": 837
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199840832883808e-05,
+      "loss": 1.2391,
+      "step": 838
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1998383748069893e-05,
+      "loss": 1.2621,
+      "step": 839
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1998358978963703e-05,
+      "loss": 1.2178,
+      "step": 840
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199833402151998e-05,
+      "loss": 1.2084,
+      "step": 841
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1998308875739186e-05,
+      "loss": 1.1251,
+      "step": 842
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1998283541621804e-05,
+      "loss": 1.2425,
+      "step": 843
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1998258019168306e-05,
+      "loss": 1.0974,
+      "step": 844
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199823230837917e-05,
+      "loss": 1.1329,
+      "step": 845
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1998206409254887e-05,
+      "loss": 1.161,
+      "step": 846
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199818032179594e-05,
+      "loss": 1.1994,
+      "step": 847
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1998154046002823e-05,
+      "loss": 1.1063,
+      "step": 848
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199812758187603e-05,
+      "loss": 1.2022,
+      "step": 849
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1998100929416058e-05,
+      "loss": 1.247,
+      "step": 850
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199807408862341e-05,
+      "loss": 1.1799,
+      "step": 851
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.19980470594986e-05,
+      "loss": 1.2419,
+      "step": 852
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199801984204212e-05,
+      "loss": 1.1837,
+      "step": 853
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997992436254497e-05,
+      "loss": 1.1804,
+      "step": 854
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199796484213624e-05,
+      "loss": 1.1675,
+      "step": 855
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997937059687874e-05,
+      "loss": 1.2618,
+      "step": 856
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997909088909917e-05,
+      "loss": 1.2115,
+      "step": 857
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997880929802896e-05,
+      "loss": 1.2434,
+      "step": 858
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199785258236735e-05,
+      "loss": 1.2188,
+      "step": 859
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.19978240466038e-05,
+      "loss": 1.2684,
+      "step": 860
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997795322512797e-05,
+      "loss": 1.2392,
+      "step": 861
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199776641009487e-05,
+      "loss": 1.2008,
+      "step": 862
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199773730935057e-05,
+      "loss": 1.2942,
+      "step": 863
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997708020280442e-05,
+      "loss": 1.2919,
+      "step": 864
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997678542885043e-05,
+      "loss": 1.1917,
+      "step": 865
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997648877164924e-05,
+      "loss": 1.3143,
+      "step": 866
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997619023120645e-05,
+      "loss": 1.1598,
+      "step": 867
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199758898075277e-05,
+      "loss": 1.2784,
+      "step": 868
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199755875006186e-05,
+      "loss": 1.1724,
+      "step": 869
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199752833104849e-05,
+      "loss": 1.2239,
+      "step": 870
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997497723713226e-05,
+      "loss": 1.281,
+      "step": 871
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997466928056654e-05,
+      "loss": 1.2524,
+      "step": 872
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997435944079348e-05,
+      "loss": 1.2269,
+      "step": 873
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997404771781895e-05,
+      "loss": 1.2832,
+      "step": 874
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997373411164873e-05,
+      "loss": 1.3111,
+      "step": 875
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997341862228887e-05,
+      "loss": 1.2159,
+      "step": 876
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199731012497452e-05,
+      "loss": 1.368,
+      "step": 877
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997278199402374e-05,
+      "loss": 1.1932,
+      "step": 878
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997246085513054e-05,
+      "loss": 1.4176,
+      "step": 879
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997213783307154e-05,
+      "loss": 1.3706,
+      "step": 880
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997181292785295e-05,
+      "loss": 1.2278,
+      "step": 881
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997148613948082e-05,
+      "loss": 1.258,
+      "step": 882
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997115746796133e-05,
+      "loss": 1.2063,
+      "step": 883
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997082691330067e-05,
+      "loss": 1.2501,
+      "step": 884
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997049447550505e-05,
+      "loss": 1.3057,
+      "step": 885
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997016015458074e-05,
+      "loss": 1.3463,
+      "step": 886
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1996982395053402e-05,
+      "loss": 1.33,
+      "step": 887
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199694858633713e-05,
+      "loss": 1.2297,
+      "step": 888
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1996914589309886e-05,
+      "loss": 1.3214,
+      "step": 889
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199688040397231e-05,
+      "loss": 1.2364,
+      "step": 890
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1996846030325052e-05,
+      "loss": 1.3889,
+      "step": 891
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1996811468368758e-05,
+      "loss": 1.2072,
+      "step": 892
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1996776718104077e-05,
+      "loss": 1.315,
+      "step": 893
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1996741779531666e-05,
+      "loss": 1.3256,
+      "step": 894
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1996706652652183e-05,
+      "loss": 1.3076,
+      "step": 895
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1996671337466284e-05,
+      "loss": 1.2026,
+      "step": 896
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1996635833974638e-05,
+      "loss": 1.1817,
+      "step": 897
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1996600142177918e-05,
+      "loss": 1.2938,
+      "step": 898
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1996564262076792e-05,
+      "loss": 1.2671,
+      "step": 899
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1996528193671937e-05,
+      "loss": 1.3152,
+      "step": 900
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1996491936964028e-05,
+      "loss": 1.4438,
+      "step": 901
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1996455491953756e-05,
+      "loss": 1.2213,
+      "step": 902
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1996418858641806e-05,
+      "loss": 1.3458,
+      "step": 903
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199638203702886e-05,
+      "loss": 1.2108,
+      "step": 904
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199634502711562e-05,
+      "loss": 1.3206,
+      "step": 905
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199630782890278e-05,
+      "loss": 1.2595,
+      "step": 906
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199627044239104e-05,
+      "loss": 1.3052,
+      "step": 907
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1996232867581105e-05,
+      "loss": 1.3186,
+      "step": 908
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199619510447368e-05,
+      "loss": 1.2759,
+      "step": 909
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1996157153069486e-05,
+      "loss": 1.3081,
+      "step": 910
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199611901336923e-05,
+      "loss": 1.2393,
+      "step": 911
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199608068537363e-05,
+      "loss": 1.2687,
+      "step": 912
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1996042169083405e-05,
+      "loss": 1.2321,
+      "step": 913
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199600346449929e-05,
+      "loss": 1.1026,
+      "step": 914
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199596457162201e-05,
+      "loss": 1.2409,
+      "step": 915
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1995925490452292e-05,
+      "loss": 1.1989,
+      "step": 916
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199588622099088e-05,
+      "loss": 1.1519,
+      "step": 917
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1995846763238515e-05,
+      "loss": 1.2189,
+      "step": 918
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1995807117195932e-05,
+      "loss": 1.229,
+      "step": 919
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199576728286388e-05,
+      "loss": 1.1975,
+      "step": 920
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199572726024311e-05,
+      "loss": 1.2937,
+      "step": 921
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1995687049334385e-05,
+      "loss": 1.1936,
+      "step": 922
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1995646650138447e-05,
+      "loss": 1.2705,
+      "step": 923
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199560606265607e-05,
+      "loss": 1.1653,
+      "step": 924
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1995565286888007e-05,
+      "loss": 1.2275,
+      "step": 925
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1995524322835036e-05,
+      "loss": 1.2323,
+      "step": 926
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1995483170497924e-05,
+      "loss": 1.3067,
+      "step": 927
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1995441829877445e-05,
+      "loss": 1.2579,
+      "step": 928
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1995400300974384e-05,
+      "loss": 1.2511,
+      "step": 929
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1995358583789515e-05,
+      "loss": 1.2309,
+      "step": 930
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1995316678323628e-05,
+      "loss": 1.2699,
+      "step": 931
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1995274584577513e-05,
+      "loss": 1.2805,
+      "step": 932
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1995232302551966e-05,
+      "loss": 1.21,
+      "step": 933
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1995189832247773e-05,
+      "loss": 1.1743,
+      "step": 934
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1995147173665743e-05,
+      "loss": 1.2921,
+      "step": 935
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1995104326806676e-05,
+      "loss": 1.1877,
+      "step": 936
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1995061291671383e-05,
+      "loss": 1.1616,
+      "step": 937
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199501806826067e-05,
+      "loss": 1.3721,
+      "step": 938
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199497465657535e-05,
+      "loss": 1.2857,
+      "step": 939
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1994931056616248e-05,
+      "loss": 1.2566,
+      "step": 940
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199488726838418e-05,
+      "loss": 1.2082,
+      "step": 941
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199484329187997e-05,
+      "loss": 1.2133,
+      "step": 942
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199479912710445e-05,
+      "loss": 1.2079,
+      "step": 943
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199475477405845e-05,
+      "loss": 1.2495,
+      "step": 944
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1994710232742806e-05,
+      "loss": 1.2692,
+      "step": 945
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1994665503158352e-05,
+      "loss": 1.2804,
+      "step": 946
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1994620585305942e-05,
+      "loss": 1.2573,
+      "step": 947
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1994575479186412e-05,
+      "loss": 1.3282,
+      "step": 948
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1994530184800617e-05,
+      "loss": 1.2948,
+      "step": 949
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1994484702149405e-05,
+      "loss": 1.1934,
+      "step": 950
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199443903123364e-05,
+      "loss": 1.2147,
+      "step": 951
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1994393172054172e-05,
+      "loss": 1.2341,
+      "step": 952
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1994347124611874e-05,
+      "loss": 1.1826,
+      "step": 953
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1994300888907614e-05,
+      "loss": 1.221,
+      "step": 954
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1994254464942254e-05,
+      "loss": 1.1724,
+      "step": 955
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1994207852716677e-05,
+      "loss": 1.2285,
+      "step": 956
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199416105223176e-05,
+      "loss": 1.1994,
+      "step": 957
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1994114063488377e-05,
+      "loss": 1.2842,
+      "step": 958
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199406688648742e-05,
+      "loss": 1.1688,
+      "step": 959
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199401952122978e-05,
+      "loss": 1.1574,
+      "step": 960
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199397196771634e-05,
+      "loss": 1.2817,
+      "step": 961
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1993924225948006e-05,
+      "loss": 1.2967,
+      "step": 962
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199387629592567e-05,
+      "loss": 1.1698,
+      "step": 963
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199382817765024e-05,
+      "loss": 1.2624,
+      "step": 964
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1993779871122614e-05,
+      "loss": 1.2775,
+      "step": 965
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1993731376343712e-05,
+      "loss": 1.2045,
+      "step": 966
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1993682693314445e-05,
+      "loss": 1.1907,
+      "step": 967
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1993633822035723e-05,
+      "loss": 1.2602,
+      "step": 968
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1993584762508473e-05,
+      "loss": 1.1362,
+      "step": 969
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1993535514733622e-05,
+      "loss": 1.2701,
+      "step": 970
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1993486078712092e-05,
+      "loss": 1.2092,
+      "step": 971
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1993436454444815e-05,
+      "loss": 1.2583,
+      "step": 972
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1993386641932725e-05,
+      "loss": 1.2016,
+      "step": 973
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199333664117676e-05,
+      "loss": 1.1648,
+      "step": 974
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199328645217787e-05,
+      "loss": 1.2166,
+      "step": 975
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199323607493699e-05,
+      "loss": 1.2046,
+      "step": 976
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1993185509455075e-05,
+      "loss": 1.1448,
+      "step": 977
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1993134755733076e-05,
+      "loss": 1.2116,
+      "step": 978
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1993083813771947e-05,
+      "loss": 1.212,
+      "step": 979
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199303268357265e-05,
+      "loss": 1.1234,
+      "step": 980
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1992981365136146e-05,
+      "loss": 1.156,
+      "step": 981
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1992929858463406e-05,
+      "loss": 1.1412,
+      "step": 982
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199287816355539e-05,
+      "loss": 1.2839,
+      "step": 983
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1992826280413088e-05,
+      "loss": 1.1833,
+      "step": 984
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1992774209037465e-05,
+      "loss": 1.1557,
+      "step": 985
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1992721949429504e-05,
+      "loss": 1.1635,
+      "step": 986
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1992669501590193e-05,
+      "loss": 1.2249,
+      "step": 987
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1992616865520516e-05,
+      "loss": 1.2016,
+      "step": 988
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1992564041221467e-05,
+      "loss": 1.2093,
+      "step": 989
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1992511028694037e-05,
+      "loss": 1.1822,
+      "step": 990
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199245782793923e-05,
+      "loss": 1.127,
+      "step": 991
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1992404438958048e-05,
+      "loss": 1.1881,
+      "step": 992
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1992350861751493e-05,
+      "loss": 1.1933,
+      "step": 993
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199229709632057e-05,
+      "loss": 1.1144,
+      "step": 994
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1992243142666305e-05,
+      "loss": 1.201,
+      "step": 995
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.19921890007897e-05,
+      "loss": 1.3689,
+      "step": 996
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199213467069179e-05,
+      "loss": 1.1567,
+      "step": 997
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1992080152373588e-05,
+      "loss": 1.1355,
+      "step": 998
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1992025445836118e-05,
+      "loss": 1.1973,
+      "step": 999
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1991970551080418e-05,
+      "loss": 1.2314,
+      "step": 1000
+    },
+    {
+      "epoch": 0.04,
+      "eval_text_loss": 0.8186610341072083,
+      "eval_text_runtime": 1.3962,
+      "eval_text_samples_per_second": 53.001,
+      "eval_text_steps_per_second": 1.432,
+      "step": 1000
+    },
+    {
+      "epoch": 0.04,
+      "eval_svit_loss": 1.2973212003707886,
+      "eval_svit_runtime": 4.9617,
+      "eval_svit_samples_per_second": 137.251,
+      "eval_svit_steps_per_second": 2.217,
+      "step": 1000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1991915468107523e-05,
+      "loss": 1.1351,
+      "step": 1001
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1991860196918463e-05,
+      "loss": 1.1893,
+      "step": 1002
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1991804737514283e-05,
+      "loss": 1.1263,
+      "step": 1003
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199174908989603e-05,
+      "loss": 1.1291,
+      "step": 1004
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199169325406475e-05,
+      "loss": 1.1624,
+      "step": 1005
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1991637230021493e-05,
+      "loss": 1.1823,
+      "step": 1006
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199158101776732e-05,
+      "loss": 1.3064,
+      "step": 1007
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1991524617303284e-05,
+      "loss": 1.165,
+      "step": 1008
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199146802863045e-05,
+      "loss": 1.1947,
+      "step": 1009
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1991411251749882e-05,
+      "loss": 1.0533,
+      "step": 1010
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1991354286662653e-05,
+      "loss": 1.2468,
+      "step": 1011
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1991297133369833e-05,
+      "loss": 1.1489,
+      "step": 1012
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.19912397918725e-05,
+      "loss": 1.1639,
+      "step": 1013
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1991182262171735e-05,
+      "loss": 1.1842,
+      "step": 1014
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199112454426862e-05,
+      "loss": 1.2609,
+      "step": 1015
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199106663816424e-05,
+      "loss": 1.2202,
+      "step": 1016
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1991008543859688e-05,
+      "loss": 1.1723,
+      "step": 1017
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199095026135606e-05,
+      "loss": 1.2824,
+      "step": 1018
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1990891790654456e-05,
+      "loss": 1.2144,
+      "step": 1019
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199083313175597e-05,
+      "loss": 1.1908,
+      "step": 1020
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1990774284661714e-05,
+      "loss": 1.1979,
+      "step": 1021
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1990715249372792e-05,
+      "loss": 1.2119,
+      "step": 1022
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1990656025890317e-05,
+      "loss": 1.1834,
+      "step": 1023
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1990596614215406e-05,
+      "loss": 1.1623,
+      "step": 1024
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199053701434918e-05,
+      "loss": 1.2003,
+      "step": 1025
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1990477226292752e-05,
+      "loss": 1.2423,
+      "step": 1026
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199041725004726e-05,
+      "loss": 1.1188,
+      "step": 1027
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199035708561383e-05,
+      "loss": 1.2337,
+      "step": 1028
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199029673299359e-05,
+      "loss": 1.2065,
+      "step": 1029
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1990236192187683e-05,
+      "loss": 1.2191,
+      "step": 1030
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1990175463197246e-05,
+      "loss": 1.2014,
+      "step": 1031
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1990114546023424e-05,
+      "loss": 1.2482,
+      "step": 1032
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199005344066737e-05,
+      "loss": 1.2131,
+      "step": 1033
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1989992147130222e-05,
+      "loss": 1.1708,
+      "step": 1034
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.198993066541315e-05,
+      "loss": 1.2217,
+      "step": 1035
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.19898689955173e-05,
+      "loss": 1.249,
+      "step": 1036
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1989807137443842e-05,
+      "loss": 1.2889,
+      "step": 1037
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1989745091193932e-05,
+      "loss": 1.2293,
+      "step": 1038
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.198968285676875e-05,
+      "loss": 1.2133,
+      "step": 1039
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.198962043416946e-05,
+      "loss": 1.1375,
+      "step": 1040
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.198955782339724e-05,
+      "loss": 1.2558,
+      "step": 1041
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1989495024453272e-05,
+      "loss": 1.2376,
+      "step": 1042
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1989432037338734e-05,
+      "loss": 1.1908,
+      "step": 1043
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1989368862054815e-05,
+      "loss": 1.1404,
+      "step": 1044
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1989305498602705e-05,
+      "loss": 1.1282,
+      "step": 1045
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.19892419469836e-05,
+      "loss": 1.1497,
+      "step": 1046
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.198917820719869e-05,
+      "loss": 1.1861,
+      "step": 1047
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1989114279249186e-05,
+      "loss": 1.2149,
+      "step": 1048
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1989050163136285e-05,
+      "loss": 1.2308,
+      "step": 1049
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1988985858861195e-05,
+      "loss": 1.175,
+      "step": 1050
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.198892136642513e-05,
+      "loss": 1.0835,
+      "step": 1051
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1988856685829305e-05,
+      "loss": 1.2011,
+      "step": 1052
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1988791817074934e-05,
+      "loss": 1.1514,
+      "step": 1053
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1988726760163247e-05,
+      "loss": 1.2302,
+      "step": 1054
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.198866151509546e-05,
+      "loss": 1.2275,
+      "step": 1055
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1988596081872806e-05,
+      "loss": 1.2517,
+      "step": 1056
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.198853046049652e-05,
+      "loss": 1.1202,
+      "step": 1057
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1988464650967835e-05,
+      "loss": 1.22,
+      "step": 1058
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.198839865328799e-05,
+      "loss": 1.2367,
+      "step": 1059
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1988332467458234e-05,
+      "loss": 1.2335,
+      "step": 1060
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1988266093479804e-05,
+      "loss": 1.1833,
+      "step": 1061
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1988199531353964e-05,
+      "loss": 1.2561,
+      "step": 1062
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198813278108195e-05,
+      "loss": 1.171,
+      "step": 1063
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1988065842665037e-05,
+      "loss": 1.2107,
+      "step": 1064
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1987998716104474e-05,
+      "loss": 1.2721,
+      "step": 1065
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1987931401401532e-05,
+      "loss": 1.1584,
+      "step": 1066
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1987863898557476e-05,
+      "loss": 1.2284,
+      "step": 1067
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1987796207573574e-05,
+      "loss": 1.2072,
+      "step": 1068
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1987728328451106e-05,
+      "loss": 1.2017,
+      "step": 1069
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1987660261191354e-05,
+      "loss": 1.2092,
+      "step": 1070
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1987592005795593e-05,
+      "loss": 1.1424,
+      "step": 1071
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198752356226511e-05,
+      "loss": 1.2083,
+      "step": 1072
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1987454930601197e-05,
+      "loss": 1.2536,
+      "step": 1073
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1987386110805148e-05,
+      "loss": 1.136,
+      "step": 1074
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1987317102878255e-05,
+      "loss": 1.2083,
+      "step": 1075
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198724790682182e-05,
+      "loss": 1.2374,
+      "step": 1076
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1987178522637145e-05,
+      "loss": 1.2335,
+      "step": 1077
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1987108950325542e-05,
+      "loss": 1.1484,
+      "step": 1078
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1987039189888315e-05,
+      "loss": 1.2185,
+      "step": 1079
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198696924132678e-05,
+      "loss": 1.2074,
+      "step": 1080
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198689910464226e-05,
+      "loss": 1.198,
+      "step": 1081
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1986828779836066e-05,
+      "loss": 1.2528,
+      "step": 1082
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198675826690953e-05,
+      "loss": 1.1341,
+      "step": 1083
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198668756586398e-05,
+      "loss": 1.1194,
+      "step": 1084
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1986616676700745e-05,
+      "loss": 1.3367,
+      "step": 1085
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198654559942116e-05,
+      "loss": 1.1392,
+      "step": 1086
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198647433402657e-05,
+      "loss": 1.2912,
+      "step": 1087
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198640288051831e-05,
+      "loss": 1.2036,
+      "step": 1088
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198633123889773e-05,
+      "loss": 1.1349,
+      "step": 1089
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1986259409166178e-05,
+      "loss": 1.2884,
+      "step": 1090
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1986187391325008e-05,
+      "loss": 1.2568,
+      "step": 1091
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1986115185375577e-05,
+      "loss": 1.1279,
+      "step": 1092
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198604279131924e-05,
+      "loss": 1.1911,
+      "step": 1093
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198597020915737e-05,
+      "loss": 1.1872,
+      "step": 1094
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198589743889133e-05,
+      "loss": 1.2547,
+      "step": 1095
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1985824480522483e-05,
+      "loss": 1.1876,
+      "step": 1096
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1985751334052215e-05,
+      "loss": 1.233,
+      "step": 1097
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.19856779994819e-05,
+      "loss": 1.1906,
+      "step": 1098
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1985604476812914e-05,
+      "loss": 1.2628,
+      "step": 1099
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1985530766046647e-05,
+      "loss": 1.1876,
+      "step": 1100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1985456867184486e-05,
+      "loss": 1.2331,
+      "step": 1101
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1985382780227827e-05,
+      "loss": 1.1656,
+      "step": 1102
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1985308505178062e-05,
+      "loss": 1.2868,
+      "step": 1103
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198523404203659e-05,
+      "loss": 1.1446,
+      "step": 1104
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1985159390804815e-05,
+      "loss": 1.2796,
+      "step": 1105
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198508455148414e-05,
+      "loss": 1.2363,
+      "step": 1106
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198500952407598e-05,
+      "loss": 1.2194,
+      "step": 1107
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1984934308581743e-05,
+      "loss": 1.2344,
+      "step": 1108
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198485890500285e-05,
+      "loss": 1.1145,
+      "step": 1109
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1984783313340716e-05,
+      "loss": 1.1788,
+      "step": 1110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1984707533596773e-05,
+      "loss": 1.2678,
+      "step": 1111
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1984631565772443e-05,
+      "loss": 1.2492,
+      "step": 1112
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1984555409869153e-05,
+      "loss": 1.096,
+      "step": 1113
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198447906588835e-05,
+      "loss": 1.1807,
+      "step": 1114
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1984402533831457e-05,
+      "loss": 1.2286,
+      "step": 1115
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198432581369993e-05,
+      "loss": 1.1907,
+      "step": 1116
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1984248905495205e-05,
+      "loss": 1.1787,
+      "step": 1117
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1984171809218733e-05,
+      "loss": 1.2283,
+      "step": 1118
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1984094524871964e-05,
+      "loss": 1.1789,
+      "step": 1119
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198401705245636e-05,
+      "loss": 1.2607,
+      "step": 1120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198393939197337e-05,
+      "loss": 1.3062,
+      "step": 1121
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1983861543424468e-05,
+      "loss": 1.1761,
+      "step": 1122
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1983783506811117e-05,
+      "loss": 1.0939,
+      "step": 1123
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198370528213478e-05,
+      "loss": 1.179,
+      "step": 1124
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1983626869396943e-05,
+      "loss": 1.2044,
+      "step": 1125
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1983548268599072e-05,
+      "loss": 1.2272,
+      "step": 1126
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1983469479742652e-05,
+      "loss": 1.2115,
+      "step": 1127
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198339050282917e-05,
+      "loss": 1.2338,
+      "step": 1128
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1983311337860108e-05,
+      "loss": 1.1877,
+      "step": 1129
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198323198483696e-05,
+      "loss": 1.1569,
+      "step": 1130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198315244376122e-05,
+      "loss": 1.2576,
+      "step": 1131
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1983072714634386e-05,
+      "loss": 1.1083,
+      "step": 1132
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1982992797457963e-05,
+      "loss": 1.2278,
+      "step": 1133
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1982912692233457e-05,
+      "loss": 1.256,
+      "step": 1134
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1982832398962367e-05,
+      "loss": 1.1161,
+      "step": 1135
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198275191764622e-05,
+      "loss": 1.2541,
+      "step": 1136
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198267124828652e-05,
+      "loss": 1.236,
+      "step": 1137
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1982590390884788e-05,
+      "loss": 1.2304,
+      "step": 1138
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1982509345442556e-05,
+      "loss": 1.1926,
+      "step": 1139
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198242811196134e-05,
+      "loss": 1.2887,
+      "step": 1140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1982346690442678e-05,
+      "loss": 1.1859,
+      "step": 1141
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.19822650808881e-05,
+      "loss": 1.1368,
+      "step": 1142
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1982183283299146e-05,
+      "loss": 1.1689,
+      "step": 1143
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1982101297677353e-05,
+      "loss": 1.0758,
+      "step": 1144
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1982019124024268e-05,
+      "loss": 1.1988,
+      "step": 1145
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198193676234144e-05,
+      "loss": 1.1836,
+      "step": 1146
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1981854212630414e-05,
+      "loss": 1.2282,
+      "step": 1147
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1981771474892753e-05,
+      "loss": 1.1185,
+      "step": 1148
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198168854913001e-05,
+      "loss": 1.2484,
+      "step": 1149
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198160543534375e-05,
+      "loss": 1.2425,
+      "step": 1150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1981522133535533e-05,
+      "loss": 1.2162,
+      "step": 1151
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198143864370694e-05,
+      "loss": 1.2373,
+      "step": 1152
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1981354965859534e-05,
+      "loss": 1.1228,
+      "step": 1153
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1981271099994892e-05,
+      "loss": 1.169,
+      "step": 1154
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1981187046114598e-05,
+      "loss": 1.23,
+      "step": 1155
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1981102804220234e-05,
+      "loss": 1.2103,
+      "step": 1156
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198101837431338e-05,
+      "loss": 1.2012,
+      "step": 1157
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1980933756395636e-05,
+      "loss": 1.0792,
+      "step": 1158
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1980848950468592e-05,
+      "loss": 1.1561,
+      "step": 1159
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1980763956533843e-05,
+      "loss": 1.1077,
+      "step": 1160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1980678774592997e-05,
+      "loss": 1.2087,
+      "step": 1161
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1980593404647648e-05,
+      "loss": 1.141,
+      "step": 1162
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1980507846699413e-05,
+      "loss": 1.1182,
+      "step": 1163
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.19804221007499e-05,
+      "loss": 1.1889,
+      "step": 1164
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1980336166800725e-05,
+      "loss": 1.1529,
+      "step": 1165
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198025004485351e-05,
+      "loss": 1.1323,
+      "step": 1166
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198016373490987e-05,
+      "loss": 1.2424,
+      "step": 1167
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1980077236971437e-05,
+      "loss": 1.2512,
+      "step": 1168
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1979990551039837e-05,
+      "loss": 1.0717,
+      "step": 1169
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1979903677116706e-05,
+      "loss": 1.2466,
+      "step": 1170
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1979816615203675e-05,
+      "loss": 1.2526,
+      "step": 1171
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1979729365302388e-05,
+      "loss": 1.1365,
+      "step": 1172
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1979641927414488e-05,
+      "loss": 1.1556,
+      "step": 1173
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1979554301541625e-05,
+      "loss": 1.2473,
+      "step": 1174
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1979466487685442e-05,
+      "loss": 1.2247,
+      "step": 1175
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.19793784858476e-05,
+      "loss": 1.2556,
+      "step": 1176
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1979290296029756e-05,
+      "loss": 1.1429,
+      "step": 1177
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197920191823357e-05,
+      "loss": 1.1572,
+      "step": 1178
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1979113352460702e-05,
+      "loss": 1.2683,
+      "step": 1179
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1979024598712828e-05,
+      "loss": 1.1838,
+      "step": 1180
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1978935656991615e-05,
+      "loss": 1.1979,
+      "step": 1181
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197884652729874e-05,
+      "loss": 1.2916,
+      "step": 1182
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1978757209635883e-05,
+      "loss": 1.2206,
+      "step": 1183
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1978667704004722e-05,
+      "loss": 1.2964,
+      "step": 1184
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1978578010406948e-05,
+      "loss": 1.2451,
+      "step": 1185
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1978488128844252e-05,
+      "loss": 1.2393,
+      "step": 1186
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197839805931832e-05,
+      "loss": 1.1753,
+      "step": 1187
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1978307801830856e-05,
+      "loss": 1.1719,
+      "step": 1188
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197821735638355e-05,
+      "loss": 1.1544,
+      "step": 1189
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1978126722978118e-05,
+      "loss": 1.1919,
+      "step": 1190
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197803590161626e-05,
+      "loss": 1.2097,
+      "step": 1191
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197794489229969e-05,
+      "loss": 1.2335,
+      "step": 1192
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197785369503012e-05,
+      "loss": 1.3169,
+      "step": 1193
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1977762309809267e-05,
+      "loss": 1.2389,
+      "step": 1194
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1977670736638855e-05,
+      "loss": 1.1299,
+      "step": 1195
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1977578975520606e-05,
+      "loss": 1.1751,
+      "step": 1196
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1977487026456253e-05,
+      "loss": 1.151,
+      "step": 1197
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1977394889447527e-05,
+      "loss": 1.2209,
+      "step": 1198
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197730256449616e-05,
+      "loss": 1.1725,
+      "step": 1199
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197721005160389e-05,
+      "loss": 1.2403,
+      "step": 1200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1977117350772467e-05,
+      "loss": 1.1593,
+      "step": 1201
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1977024462003634e-05,
+      "loss": 1.2415,
+      "step": 1202
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1976931385299134e-05,
+      "loss": 1.197,
+      "step": 1203
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197683812066073e-05,
+      "loss": 1.1669,
+      "step": 1204
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1976744668090176e-05,
+      "loss": 1.2021,
+      "step": 1205
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197665102758923e-05,
+      "loss": 1.2884,
+      "step": 1206
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1976557199159656e-05,
+      "loss": 1.0997,
+      "step": 1207
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1976463182803223e-05,
+      "loss": 1.1807,
+      "step": 1208
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.19763689785217e-05,
+      "loss": 1.2563,
+      "step": 1209
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1976274586316867e-05,
+      "loss": 1.2656,
+      "step": 1210
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1976180006190496e-05,
+      "loss": 1.1298,
+      "step": 1211
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197608523814437e-05,
+      "loss": 1.1751,
+      "step": 1212
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1975990282180275e-05,
+      "loss": 1.2854,
+      "step": 1213
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.19758951383e-05,
+      "loss": 1.2115,
+      "step": 1214
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1975799806505336e-05,
+      "loss": 1.2554,
+      "step": 1215
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197570428679808e-05,
+      "loss": 1.3042,
+      "step": 1216
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197560857918003e-05,
+      "loss": 1.2212,
+      "step": 1217
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1975512683652986e-05,
+      "loss": 1.1659,
+      "step": 1218
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1975416600218765e-05,
+      "loss": 1.2336,
+      "step": 1219
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1975320328879164e-05,
+      "loss": 1.1011,
+      "step": 1220
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1975223869636007e-05,
+      "loss": 1.0475,
+      "step": 1221
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1975127222491103e-05,
+      "loss": 1.2076,
+      "step": 1222
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1975030387446274e-05,
+      "loss": 1.2548,
+      "step": 1223
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1974933364503348e-05,
+      "loss": 1.1803,
+      "step": 1224
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1974836153664154e-05,
+      "loss": 1.2583,
+      "step": 1225
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1974738754930514e-05,
+      "loss": 1.1354,
+      "step": 1226
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1974641168304274e-05,
+      "loss": 1.2403,
+      "step": 1227
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1974543393787263e-05,
+      "loss": 1.1496,
+      "step": 1228
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1974445431381324e-05,
+      "loss": 1.206,
+      "step": 1229
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1974347281088306e-05,
+      "loss": 1.2181,
+      "step": 1230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1974248942910055e-05,
+      "loss": 1.3339,
+      "step": 1231
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197415041684843e-05,
+      "loss": 1.1841,
+      "step": 1232
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197405170290528e-05,
+      "loss": 1.1266,
+      "step": 1233
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1973952801082462e-05,
+      "loss": 1.153,
+      "step": 1234
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1973853711381845e-05,
+      "loss": 1.0528,
+      "step": 1235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1973754433805295e-05,
+      "loss": 1.1962,
+      "step": 1236
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197365496835468e-05,
+      "loss": 1.1548,
+      "step": 1237
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1973555315031874e-05,
+      "loss": 1.1917,
+      "step": 1238
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1973455473838754e-05,
+      "loss": 1.1837,
+      "step": 1239
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1973355444777206e-05,
+      "loss": 1.2037,
+      "step": 1240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1973255227849104e-05,
+      "loss": 1.1706,
+      "step": 1241
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1973154823056344e-05,
+      "loss": 1.1553,
+      "step": 1242
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1973054230400814e-05,
+      "loss": 1.1728,
+      "step": 1243
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197295344988441e-05,
+      "loss": 1.2511,
+      "step": 1244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1972852481509024e-05,
+      "loss": 1.1949,
+      "step": 1245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197275132527657e-05,
+      "loss": 1.2615,
+      "step": 1246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1972649981188947e-05,
+      "loss": 1.1854,
+      "step": 1247
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1972548449248062e-05,
+      "loss": 1.1937,
+      "step": 1248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197244672945583e-05,
+      "loss": 1.2041,
+      "step": 1249
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1972344821814164e-05,
+      "loss": 1.1969,
+      "step": 1250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1972242726324993e-05,
+      "loss": 1.2736,
+      "step": 1251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197214044299023e-05,
+      "loss": 1.1583,
+      "step": 1252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1972037971811804e-05,
+      "loss": 1.1668,
+      "step": 1253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1971935312791647e-05,
+      "loss": 1.1732,
+      "step": 1254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1971832465931695e-05,
+      "loss": 1.2229,
+      "step": 1255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197172943123388e-05,
+      "loss": 1.1019,
+      "step": 1256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1971626208700146e-05,
+      "loss": 1.2951,
+      "step": 1257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1971522798332437e-05,
+      "loss": 1.2656,
+      "step": 1258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.19714192001327e-05,
+      "loss": 1.1114,
+      "step": 1259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1971315414102887e-05,
+      "loss": 1.1579,
+      "step": 1260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197121144024495e-05,
+      "loss": 1.1771,
+      "step": 1261
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197110727856086e-05,
+      "loss": 1.2398,
+      "step": 1262
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1971002929052563e-05,
+      "loss": 1.2283,
+      "step": 1263
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1970898391722033e-05,
+      "loss": 1.1978,
+      "step": 1264
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1970793666571237e-05,
+      "loss": 1.2181,
+      "step": 1265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197068875360215e-05,
+      "loss": 1.2166,
+      "step": 1266
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1970583652816742e-05,
+      "loss": 1.1535,
+      "step": 1267
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1970478364217e-05,
+      "loss": 1.1897,
+      "step": 1268
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1970372887804908e-05,
+      "loss": 1.236,
+      "step": 1269
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1970267223582442e-05,
+      "loss": 1.1706,
+      "step": 1270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1970161371551607e-05,
+      "loss": 1.2257,
+      "step": 1271
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1970055331714384e-05,
+      "loss": 1.0999,
+      "step": 1272
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1969949104072782e-05,
+      "loss": 1.1351,
+      "step": 1273
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1969842688628793e-05,
+      "loss": 1.2378,
+      "step": 1274
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1969736085384424e-05,
+      "loss": 1.0403,
+      "step": 1275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1969629294341686e-05,
+      "loss": 1.2419,
+      "step": 1276
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1969522315502585e-05,
+      "loss": 1.0881,
+      "step": 1277
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1969415148869142e-05,
+      "loss": 1.284,
+      "step": 1278
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.196930779444337e-05,
+      "loss": 1.1667,
+      "step": 1279
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.19692002522273e-05,
+      "loss": 1.2128,
+      "step": 1280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.196909252222295e-05,
+      "loss": 1.1316,
+      "step": 1281
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.196898460443235e-05,
+      "loss": 1.1488,
+      "step": 1282
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.196887649885753e-05,
+      "loss": 1.2242,
+      "step": 1283
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1968768205500538e-05,
+      "loss": 1.1139,
+      "step": 1284
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1968659724363403e-05,
+      "loss": 1.2731,
+      "step": 1285
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1968551055448174e-05,
+      "loss": 1.163,
+      "step": 1286
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1968442198756897e-05,
+      "loss": 1.1298,
+      "step": 1287
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1968333154291617e-05,
+      "loss": 1.1461,
+      "step": 1288
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1968223922054395e-05,
+      "loss": 1.2225,
+      "step": 1289
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1968114502047286e-05,
+      "loss": 1.1698,
+      "step": 1290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1968004894272352e-05,
+      "loss": 1.1535,
+      "step": 1291
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1967895098731658e-05,
+      "loss": 1.1881,
+      "step": 1292
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.196778511542727e-05,
+      "loss": 1.2106,
+      "step": 1293
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.196767494436126e-05,
+      "loss": 1.121,
+      "step": 1294
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1967564585535704e-05,
+      "loss": 1.1867,
+      "step": 1295
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.196745403895268e-05,
+      "loss": 1.0915,
+      "step": 1296
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1967343304614274e-05,
+      "loss": 1.2603,
+      "step": 1297
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1967232382522565e-05,
+      "loss": 1.1618,
+      "step": 1298
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.196712127267965e-05,
+      "loss": 1.161,
+      "step": 1299
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.196700997508762e-05,
+      "loss": 1.2137,
+      "step": 1300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1966898489748568e-05,
+      "loss": 1.1367,
+      "step": 1301
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1966786816664596e-05,
+      "loss": 1.1104,
+      "step": 1302
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.196667495583781e-05,
+      "loss": 1.212,
+      "step": 1303
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1966562907270312e-05,
+      "loss": 1.1856,
+      "step": 1304
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1966450670964217e-05,
+      "loss": 1.2771,
+      "step": 1305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1966338246921635e-05,
+      "loss": 1.2597,
+      "step": 1306
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.196622563514469e-05,
+      "loss": 1.287,
+      "step": 1307
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1966112835635495e-05,
+      "loss": 1.1161,
+      "step": 1308
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1965999848396182e-05,
+      "loss": 1.2148,
+      "step": 1309
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.196588667342888e-05,
+      "loss": 1.1495,
+      "step": 1310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1965773310735712e-05,
+      "loss": 1.2212,
+      "step": 1311
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1965659760318823e-05,
+      "loss": 1.1977,
+      "step": 1312
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1965546022180345e-05,
+      "loss": 1.1197,
+      "step": 1313
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1965432096322424e-05,
+      "loss": 1.1679,
+      "step": 1314
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1965317982747203e-05,
+      "loss": 1.2261,
+      "step": 1315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.196520368145684e-05,
+      "loss": 1.098,
+      "step": 1316
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1965089192453476e-05,
+      "loss": 1.1509,
+      "step": 1317
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1964974515739276e-05,
+      "loss": 1.195,
+      "step": 1318
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1964859651316398e-05,
+      "loss": 1.1985,
+      "step": 1319
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1964744599187008e-05,
+      "loss": 1.2065,
+      "step": 1320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1964629359353266e-05,
+      "loss": 1.1843,
+      "step": 1321
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.196451393181735e-05,
+      "loss": 1.1702,
+      "step": 1322
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.196439831658143e-05,
+      "loss": 1.1978,
+      "step": 1323
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1964282513647687e-05,
+      "loss": 1.138,
+      "step": 1324
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.19641665230183e-05,
+      "loss": 1.1981,
+      "step": 1325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1964050344695455e-05,
+      "loss": 1.2559,
+      "step": 1326
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1963933978681337e-05,
+      "loss": 1.1457,
+      "step": 1327
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1963817424978145e-05,
+      "loss": 1.1065,
+      "step": 1328
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.196370068358807e-05,
+      "loss": 1.1126,
+      "step": 1329
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.196358375451331e-05,
+      "loss": 1.1945,
+      "step": 1330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.196346663775607e-05,
+      "loss": 1.2206,
+      "step": 1331
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.196334933331856e-05,
+      "loss": 1.0639,
+      "step": 1332
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1963231841202977e-05,
+      "loss": 1.1634,
+      "step": 1333
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1963114161411545e-05,
+      "loss": 1.1235,
+      "step": 1334
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1962996293946483e-05,
+      "loss": 1.1523,
+      "step": 1335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1962878238810002e-05,
+      "loss": 1.1841,
+      "step": 1336
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.196275999600433e-05,
+      "loss": 1.1658,
+      "step": 1337
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1962641565531696e-05,
+      "loss": 1.0583,
+      "step": 1338
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1962522947394326e-05,
+      "loss": 1.0358,
+      "step": 1339
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.196240414159446e-05,
+      "loss": 1.1428,
+      "step": 1340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1962285148134333e-05,
+      "loss": 1.1067,
+      "step": 1341
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1962165967016186e-05,
+      "loss": 1.1991,
+      "step": 1342
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1962046598242264e-05,
+      "loss": 1.0641,
+      "step": 1343
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.196192704181482e-05,
+      "loss": 1.1267,
+      "step": 1344
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1961807297736095e-05,
+      "loss": 1.152,
+      "step": 1345
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1961687366008362e-05,
+      "loss": 1.0938,
+      "step": 1346
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1961567246633866e-05,
+      "loss": 1.1165,
+      "step": 1347
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.196144693961487e-05,
+      "loss": 1.1871,
+      "step": 1348
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1961326444953648e-05,
+      "loss": 1.1466,
+      "step": 1349
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.196120576265247e-05,
+      "loss": 1.0995,
+      "step": 1350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.19610848927136e-05,
+      "loss": 1.1319,
+      "step": 1351
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1960963835139318e-05,
+      "loss": 1.124,
+      "step": 1352
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1960842589931908e-05,
+      "loss": 1.2039,
+      "step": 1353
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1960721157093653e-05,
+      "loss": 1.0759,
+      "step": 1354
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.196059953662684e-05,
+      "loss": 1.1937,
+      "step": 1355
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1960477728533758e-05,
+      "loss": 1.1191,
+      "step": 1356
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1960355732816705e-05,
+      "loss": 1.1263,
+      "step": 1357
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1960233549477975e-05,
+      "loss": 1.223,
+      "step": 1358
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1960111178519875e-05,
+      "loss": 1.1656,
+      "step": 1359
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1959988619944702e-05,
+      "loss": 1.1758,
+      "step": 1360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1959865873754774e-05,
+      "loss": 1.1845,
+      "step": 1361
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1959742939952395e-05,
+      "loss": 1.1743,
+      "step": 1362
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.195961981853988e-05,
+      "loss": 1.2046,
+      "step": 1363
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1959496509519562e-05,
+      "loss": 1.1774,
+      "step": 1364
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1959373012893747e-05,
+      "loss": 1.1408,
+      "step": 1365
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.195924932866477e-05,
+      "loss": 1.161,
+      "step": 1366
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.195912545683496e-05,
+      "loss": 1.1002,
+      "step": 1367
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.195900139740665e-05,
+      "loss": 1.2785,
+      "step": 1368
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1958877150382176e-05,
+      "loss": 1.1112,
+      "step": 1369
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1958752715763877e-05,
+      "loss": 1.1335,
+      "step": 1370
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.19586280935541e-05,
+      "loss": 1.0589,
+      "step": 1371
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1958503283755195e-05,
+      "loss": 1.1674,
+      "step": 1372
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1958378286369503e-05,
+      "loss": 1.1207,
+      "step": 1373
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.195825310139939e-05,
+      "loss": 1.21,
+      "step": 1374
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1958127728847205e-05,
+      "loss": 1.2143,
+      "step": 1375
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1958002168715313e-05,
+      "loss": 1.2347,
+      "step": 1376
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1957876421006084e-05,
+      "loss": 1.1824,
+      "step": 1377
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.195775048572188e-05,
+      "loss": 1.1646,
+      "step": 1378
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1957624362865076e-05,
+      "loss": 1.1572,
+      "step": 1379
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1957498052438047e-05,
+      "loss": 1.1714,
+      "step": 1380
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1957371554443174e-05,
+      "loss": 1.1625,
+      "step": 1381
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1957244868882837e-05,
+      "loss": 1.1669,
+      "step": 1382
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1957117995759424e-05,
+      "loss": 1.1906,
+      "step": 1383
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1956990935075323e-05,
+      "loss": 1.1996,
+      "step": 1384
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1956863686832934e-05,
+      "loss": 1.1601,
+      "step": 1385
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1956736251034644e-05,
+      "loss": 1.1315,
+      "step": 1386
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1956608627682864e-05,
+      "loss": 1.2199,
+      "step": 1387
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.195648081677999e-05,
+      "loss": 1.0727,
+      "step": 1388
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1956352818328428e-05,
+      "loss": 1.1711,
+      "step": 1389
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1956224632330597e-05,
+      "loss": 1.2158,
+      "step": 1390
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1956096258788908e-05,
+      "loss": 1.2071,
+      "step": 1391
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1955967697705783e-05,
+      "loss": 1.098,
+      "step": 1392
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1955838949083634e-05,
+      "loss": 1.134,
+      "step": 1393
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1955710012924894e-05,
+      "loss": 1.1125,
+      "step": 1394
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.195558088923199e-05,
+      "loss": 1.1838,
+      "step": 1395
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1955451578007354e-05,
+      "loss": 1.0768,
+      "step": 1396
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.195532207925342e-05,
+      "loss": 1.1544,
+      "step": 1397
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.195519239297263e-05,
+      "loss": 1.1887,
+      "step": 1398
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1955062519167427e-05,
+      "loss": 1.2493,
+      "step": 1399
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1954932457840257e-05,
+      "loss": 1.1062,
+      "step": 1400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1954802208993566e-05,
+      "loss": 1.1668,
+      "step": 1401
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1954671772629813e-05,
+      "loss": 1.2214,
+      "step": 1402
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1954541148751454e-05,
+      "loss": 1.0847,
+      "step": 1403
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1954410337360946e-05,
+      "loss": 1.253,
+      "step": 1404
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1954279338460756e-05,
+      "loss": 1.1758,
+      "step": 1405
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1954148152053353e-05,
+      "loss": 1.2012,
+      "step": 1406
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1954016778141207e-05,
+      "loss": 1.1233,
+      "step": 1407
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.195388521672679e-05,
+      "loss": 1.1294,
+      "step": 1408
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.195375346781258e-05,
+      "loss": 1.0923,
+      "step": 1409
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.195362153140106e-05,
+      "loss": 1.1626,
+      "step": 1410
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1953489407494722e-05,
+      "loss": 1.1809,
+      "step": 1411
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.195335709609605e-05,
+      "loss": 1.2264,
+      "step": 1412
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.195322459720753e-05,
+      "loss": 1.1158,
+      "step": 1413
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1953091910831662e-05,
+      "loss": 1.1505,
+      "step": 1414
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.195295903697095e-05,
+      "loss": 1.2044,
+      "step": 1415
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1952825975627893e-05,
+      "loss": 1.1377,
+      "step": 1416
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1952692726804998e-05,
+      "loss": 1.0516,
+      "step": 1417
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1952559290504774e-05,
+      "loss": 1.1779,
+      "step": 1418
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1952425666729735e-05,
+      "loss": 1.1934,
+      "step": 1419
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1952291855482398e-05,
+      "loss": 1.1654,
+      "step": 1420
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1952157856765287e-05,
+      "loss": 1.1917,
+      "step": 1421
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1952023670580916e-05,
+      "loss": 1.1606,
+      "step": 1422
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1951889296931827e-05,
+      "loss": 1.1846,
+      "step": 1423
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1951754735820543e-05,
+      "loss": 1.1423,
+      "step": 1424
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1951619987249596e-05,
+      "loss": 1.1768,
+      "step": 1425
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1951485051221533e-05,
+      "loss": 1.1404,
+      "step": 1426
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1951349927738885e-05,
+      "loss": 1.1906,
+      "step": 1427
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1951214616804204e-05,
+      "loss": 1.1238,
+      "step": 1428
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1951079118420044e-05,
+      "loss": 1.1497,
+      "step": 1429
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1950943432588943e-05,
+      "loss": 1.1305,
+      "step": 1430
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1950807559313472e-05,
+      "loss": 1.2187,
+      "step": 1431
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1950671498596177e-05,
+      "loss": 1.1739,
+      "step": 1432
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.195053525043963e-05,
+      "loss": 1.2149,
+      "step": 1433
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1950398814846397e-05,
+      "loss": 1.1304,
+      "step": 1434
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1950262191819046e-05,
+      "loss": 1.0978,
+      "step": 1435
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1950125381360145e-05,
+      "loss": 1.1858,
+      "step": 1436
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194998838347228e-05,
+      "loss": 1.1856,
+      "step": 1437
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1949851198158033e-05,
+      "loss": 1.1697,
+      "step": 1438
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1949713825419976e-05,
+      "loss": 1.1299,
+      "step": 1439
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194957626526071e-05,
+      "loss": 1.1746,
+      "step": 1440
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1949438517682818e-05,
+      "loss": 1.2196,
+      "step": 1441
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1949300582688893e-05,
+      "loss": 1.1866,
+      "step": 1442
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194916246028154e-05,
+      "loss": 1.1085,
+      "step": 1443
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194902415046336e-05,
+      "loss": 1.1896,
+      "step": 1444
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1948885653236954e-05,
+      "loss": 1.2292,
+      "step": 1445
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194874696860493e-05,
+      "loss": 1.1732,
+      "step": 1446
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1948608096569908e-05,
+      "loss": 1.123,
+      "step": 1447
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1948469037134497e-05,
+      "loss": 1.1352,
+      "step": 1448
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194832979030132e-05,
+      "loss": 1.1823,
+      "step": 1449
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1948190356073e-05,
+      "loss": 1.1644,
+      "step": 1450
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1948050734452162e-05,
+      "loss": 1.1413,
+      "step": 1451
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1947910925441436e-05,
+      "loss": 1.1608,
+      "step": 1452
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1947770929043455e-05,
+      "loss": 1.2213,
+      "step": 1453
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1947630745260856e-05,
+      "loss": 1.1811,
+      "step": 1454
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194749037409628e-05,
+      "loss": 1.2327,
+      "step": 1455
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1947349815552375e-05,
+      "loss": 1.2462,
+      "step": 1456
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1947209069631787e-05,
+      "loss": 1.1119,
+      "step": 1457
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194706813633716e-05,
+      "loss": 1.2209,
+      "step": 1458
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194692701567116e-05,
+      "loss": 1.2606,
+      "step": 1459
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1946785707636437e-05,
+      "loss": 1.1299,
+      "step": 1460
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1946644212235653e-05,
+      "loss": 1.1702,
+      "step": 1461
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194650252947148e-05,
+      "loss": 1.1514,
+      "step": 1462
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1946360659346578e-05,
+      "loss": 1.0686,
+      "step": 1463
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1946218601863627e-05,
+      "loss": 1.2392,
+      "step": 1464
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1946076357025303e-05,
+      "loss": 1.1674,
+      "step": 1465
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194593392483428e-05,
+      "loss": 1.2537,
+      "step": 1466
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194579130529324e-05,
+      "loss": 1.1845,
+      "step": 1467
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1945648498404876e-05,
+      "loss": 1.2174,
+      "step": 1468
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1945505504171875e-05,
+      "loss": 1.0458,
+      "step": 1469
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1945362322596927e-05,
+      "loss": 1.2295,
+      "step": 1470
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1945218953682737e-05,
+      "loss": 1.1193,
+      "step": 1471
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1945075397432e-05,
+      "loss": 1.0809,
+      "step": 1472
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194493165384742e-05,
+      "loss": 1.1779,
+      "step": 1473
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1944787722931705e-05,
+      "loss": 1.117,
+      "step": 1474
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194464360468757e-05,
+      "loss": 1.3051,
+      "step": 1475
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1944499299117726e-05,
+      "loss": 1.1366,
+      "step": 1476
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1944354806224892e-05,
+      "loss": 1.1697,
+      "step": 1477
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194421012601179e-05,
+      "loss": 1.1127,
+      "step": 1478
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1944065258481143e-05,
+      "loss": 1.2425,
+      "step": 1479
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1943920203635685e-05,
+      "loss": 1.1804,
+      "step": 1480
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1943774961478144e-05,
+      "loss": 1.138,
+      "step": 1481
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194362953201126e-05,
+      "loss": 1.15,
+      "step": 1482
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194348391523777e-05,
+      "loss": 1.1819,
+      "step": 1483
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1943338111160414e-05,
+      "loss": 1.1774,
+      "step": 1484
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1943192119781943e-05,
+      "loss": 1.1231,
+      "step": 1485
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1943045941105104e-05,
+      "loss": 1.0778,
+      "step": 1486
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194289957513265e-05,
+      "loss": 1.219,
+      "step": 1487
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1942753021867342e-05,
+      "loss": 1.1706,
+      "step": 1488
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1942606281311943e-05,
+      "loss": 1.146,
+      "step": 1489
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1942459353469204e-05,
+      "loss": 1.1368,
+      "step": 1490
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1942312238341906e-05,
+      "loss": 1.1704,
+      "step": 1491
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1942164935932815e-05,
+      "loss": 1.1159,
+      "step": 1492
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1942017446244706e-05,
+      "loss": 1.184,
+      "step": 1493
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194186976928036e-05,
+      "loss": 1.1728,
+      "step": 1494
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1941721905042555e-05,
+      "loss": 1.1925,
+      "step": 1495
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1941573853534077e-05,
+      "loss": 1.0608,
+      "step": 1496
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1941425614757717e-05,
+      "loss": 1.1757,
+      "step": 1497
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194127718871626e-05,
+      "loss": 1.1616,
+      "step": 1498
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194112857541251e-05,
+      "loss": 1.2099,
+      "step": 1499
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1940979774849265e-05,
+      "loss": 1.135,
+      "step": 1500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194083078702933e-05,
+      "loss": 1.176,
+      "step": 1501
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1940681611955503e-05,
+      "loss": 1.113,
+      "step": 1502
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1940532249630607e-05,
+      "loss": 1.159,
+      "step": 1503
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194038270005744e-05,
+      "loss": 1.2546,
+      "step": 1504
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194023296323883e-05,
+      "loss": 1.1434,
+      "step": 1505
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1940083039177595e-05,
+      "loss": 1.1755,
+      "step": 1506
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1939932927876557e-05,
+      "loss": 1.192,
+      "step": 1507
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1939782629338545e-05,
+      "loss": 1.1784,
+      "step": 1508
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1939632143566393e-05,
+      "loss": 1.212,
+      "step": 1509
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.193948147056293e-05,
+      "loss": 1.1987,
+      "step": 1510
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1939330610331e-05,
+      "loss": 1.1381,
+      "step": 1511
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.193917956287344e-05,
+      "loss": 1.1806,
+      "step": 1512
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1939028328193098e-05,
+      "loss": 1.1388,
+      "step": 1513
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.193887690629282e-05,
+      "loss": 1.2043,
+      "step": 1514
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1938725297175464e-05,
+      "loss": 1.141,
+      "step": 1515
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.193857350084388e-05,
+      "loss": 1.2809,
+      "step": 1516
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.193842151730093e-05,
+      "loss": 1.1966,
+      "step": 1517
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1938269346549475e-05,
+      "loss": 1.2042,
+      "step": 1518
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1938116988592386e-05,
+      "loss": 1.2639,
+      "step": 1519
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.193796444343253e-05,
+      "loss": 1.1912,
+      "step": 1520
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1937811711072774e-05,
+      "loss": 1.1591,
+      "step": 1521
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1937658791516006e-05,
+      "loss": 1.1427,
+      "step": 1522
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.19375056847651e-05,
+      "loss": 1.1063,
+      "step": 1523
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1937352390822946e-05,
+      "loss": 1.1768,
+      "step": 1524
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.193719890969242e-05,
+      "loss": 1.1289,
+      "step": 1525
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1937045241376428e-05,
+      "loss": 1.1603,
+      "step": 1526
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1936891385877855e-05,
+      "loss": 1.1424,
+      "step": 1527
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1936737343199597e-05,
+      "loss": 1.1727,
+      "step": 1528
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1936583113344565e-05,
+      "loss": 1.1458,
+      "step": 1529
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1936428696315657e-05,
+      "loss": 1.2061,
+      "step": 1530
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1936274092115783e-05,
+      "loss": 1.1963,
+      "step": 1531
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.193611930074786e-05,
+      "loss": 1.2371,
+      "step": 1532
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1935964322214795e-05,
+      "loss": 1.2835,
+      "step": 1533
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1935809156519514e-05,
+      "loss": 1.1494,
+      "step": 1534
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.193565380366494e-05,
+      "loss": 1.1455,
+      "step": 1535
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1935498263653995e-05,
+      "loss": 1.2366,
+      "step": 1536
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1935342536489613e-05,
+      "loss": 1.1944,
+      "step": 1537
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1935186622174724e-05,
+      "loss": 1.1869,
+      "step": 1538
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.193503052071227e-05,
+      "loss": 1.208,
+      "step": 1539
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1934874232105182e-05,
+      "loss": 1.16,
+      "step": 1540
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1934717756356412e-05,
+      "loss": 1.193,
+      "step": 1541
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1934561093468912e-05,
+      "loss": 1.2595,
+      "step": 1542
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.193440424344562e-05,
+      "loss": 1.1302,
+      "step": 1543
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1934247206289497e-05,
+      "loss": 1.1947,
+      "step": 1544
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1934089982003504e-05,
+      "loss": 1.1368,
+      "step": 1545
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.19339325705906e-05,
+      "loss": 1.2438,
+      "step": 1546
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1933774972053745e-05,
+      "loss": 1.2766,
+      "step": 1547
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1933617186395918e-05,
+      "loss": 1.1572,
+      "step": 1548
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1933459213620087e-05,
+      "loss": 1.2088,
+      "step": 1549
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1933301053729226e-05,
+      "loss": 1.1474,
+      "step": 1550
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.193314270672631e-05,
+      "loss": 1.275,
+      "step": 1551
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.193298417261433e-05,
+      "loss": 1.1918,
+      "step": 1552
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1932825451396268e-05,
+      "loss": 1.2244,
+      "step": 1553
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1932666543075114e-05,
+      "loss": 1.1787,
+      "step": 1554
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1932507447653863e-05,
+      "loss": 1.1778,
+      "step": 1555
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.193234816513551e-05,
+      "loss": 1.1648,
+      "step": 1556
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1932188695523056e-05,
+      "loss": 1.2282,
+      "step": 1557
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.193202903881951e-05,
+      "loss": 1.1483,
+      "step": 1558
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1931869195027867e-05,
+      "loss": 1.2276,
+      "step": 1559
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.193170916415115e-05,
+      "loss": 1.1848,
+      "step": 1560
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.193154894619237e-05,
+      "loss": 1.2364,
+      "step": 1561
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1931388541154542e-05,
+      "loss": 1.1771,
+      "step": 1562
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.193122794904069e-05,
+      "loss": 1.1572,
+      "step": 1563
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1931067169853837e-05,
+      "loss": 1.2764,
+      "step": 1564
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1930906203597014e-05,
+      "loss": 1.2624,
+      "step": 1565
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.193074505027325e-05,
+      "loss": 1.1796,
+      "step": 1566
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1930583709885587e-05,
+      "loss": 1.2121,
+      "step": 1567
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.193042218243706e-05,
+      "loss": 1.2106,
+      "step": 1568
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1930260467930707e-05,
+      "loss": 1.1902,
+      "step": 1569
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1930098566369583e-05,
+      "loss": 1.1745,
+      "step": 1570
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1929936477756733e-05,
+      "loss": 1.167,
+      "step": 1571
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.192977420209521e-05,
+      "loss": 1.1273,
+      "step": 1572
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.192961173938807e-05,
+      "loss": 1.1715,
+      "step": 1573
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1929449089638378e-05,
+      "loss": 1.1884,
+      "step": 1574
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1929286252849193e-05,
+      "loss": 1.1833,
+      "step": 1575
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1929123229023583e-05,
+      "loss": 1.1418,
+      "step": 1576
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.192896001816462e-05,
+      "loss": 1.1621,
+      "step": 1577
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1928796620275378e-05,
+      "loss": 1.2193,
+      "step": 1578
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.192863303535894e-05,
+      "loss": 1.1259,
+      "step": 1579
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1928469263418377e-05,
+      "loss": 1.1298,
+      "step": 1580
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.192830530445678e-05,
+      "loss": 1.1732,
+      "step": 1581
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.192814115847724e-05,
+      "loss": 1.1955,
+      "step": 1582
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1927976825482842e-05,
+      "loss": 1.1735,
+      "step": 1583
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1927812305476686e-05,
+      "loss": 1.2138,
+      "step": 1584
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1927647598461874e-05,
+      "loss": 1.1871,
+      "step": 1585
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1927482704441504e-05,
+      "loss": 1.2601,
+      "step": 1586
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.192731762341868e-05,
+      "loss": 1.1534,
+      "step": 1587
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1927152355396516e-05,
+      "loss": 1.1291,
+      "step": 1588
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1926986900378127e-05,
+      "loss": 1.0656,
+      "step": 1589
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1926821258366623e-05,
+      "loss": 1.1709,
+      "step": 1590
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.192665542936513e-05,
+      "loss": 1.108,
+      "step": 1591
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.192648941337677e-05,
+      "loss": 1.1341,
+      "step": 1592
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1926323210404665e-05,
+      "loss": 1.1477,
+      "step": 1593
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1926156820451952e-05,
+      "loss": 1.1283,
+      "step": 1594
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1925990243521765e-05,
+      "loss": 1.1987,
+      "step": 1595
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1925823479617243e-05,
+      "loss": 1.124,
+      "step": 1596
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.192565652874152e-05,
+      "loss": 1.1943,
+      "step": 1597
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1925489390897745e-05,
+      "loss": 1.2276,
+      "step": 1598
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.192532206608907e-05,
+      "loss": 1.1527,
+      "step": 1599
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1925154554318644e-05,
+      "loss": 1.1745,
+      "step": 1600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1924986855589623e-05,
+      "loss": 1.1635,
+      "step": 1601
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1924818969905163e-05,
+      "loss": 1.1379,
+      "step": 1602
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.192465089726843e-05,
+      "loss": 1.2067,
+      "step": 1603
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1924482637682585e-05,
+      "loss": 1.2449,
+      "step": 1604
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.19243141911508e-05,
+      "loss": 1.2003,
+      "step": 1605
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1924145557676256e-05,
+      "loss": 1.1309,
+      "step": 1606
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1923976737262115e-05,
+      "loss": 1.23,
+      "step": 1607
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1923807729911568e-05,
+      "loss": 1.2076,
+      "step": 1608
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1923638535627795e-05,
+      "loss": 1.1531,
+      "step": 1609
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1923469154413984e-05,
+      "loss": 1.1419,
+      "step": 1610
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1923299586273324e-05,
+      "loss": 1.029,
+      "step": 1611
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.192312983120901e-05,
+      "loss": 1.1854,
+      "step": 1612
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.192295988922424e-05,
+      "loss": 1.2632,
+      "step": 1613
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1922789760322214e-05,
+      "loss": 1.1707,
+      "step": 1614
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1922619444506138e-05,
+      "loss": 1.12,
+      "step": 1615
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1922448941779217e-05,
+      "loss": 1.1572,
+      "step": 1616
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1922278252144668e-05,
+      "loss": 1.0954,
+      "step": 1617
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1922107375605702e-05,
+      "loss": 1.1509,
+      "step": 1618
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1921936312165542e-05,
+      "loss": 1.1187,
+      "step": 1619
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1921765061827406e-05,
+      "loss": 1.1122,
+      "step": 1620
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.192159362459452e-05,
+      "loss": 1.2491,
+      "step": 1621
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1921422000470117e-05,
+      "loss": 1.2169,
+      "step": 1622
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1921250189457428e-05,
+      "loss": 1.1212,
+      "step": 1623
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1921078191559685e-05,
+      "loss": 1.2064,
+      "step": 1624
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1920906006780135e-05,
+      "loss": 1.1648,
+      "step": 1625
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.192073363512202e-05,
+      "loss": 1.1085,
+      "step": 1626
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1920561076588585e-05,
+      "loss": 1.1718,
+      "step": 1627
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1920388331183077e-05,
+      "loss": 1.2046,
+      "step": 1628
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1920215398908755e-05,
+      "loss": 1.1967,
+      "step": 1629
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.192004227976888e-05,
+      "loss": 1.1507,
+      "step": 1630
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.19198689737667e-05,
+      "loss": 1.1174,
+      "step": 1631
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1919695480905493e-05,
+      "loss": 1.2281,
+      "step": 1632
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1919521801188522e-05,
+      "loss": 1.2207,
+      "step": 1633
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1919347934619058e-05,
+      "loss": 1.1205,
+      "step": 1634
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1919173881200377e-05,
+      "loss": 1.1346,
+      "step": 1635
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1918999640935755e-05,
+      "loss": 1.1092,
+      "step": 1636
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1918825213828476e-05,
+      "loss": 1.1061,
+      "step": 1637
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.191865059988183e-05,
+      "loss": 1.1048,
+      "step": 1638
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.19184757990991e-05,
+      "loss": 1.1973,
+      "step": 1639
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1918300811483577e-05,
+      "loss": 1.2002,
+      "step": 1640
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1918125637038566e-05,
+      "loss": 1.006,
+      "step": 1641
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.191795027576736e-05,
+      "loss": 1.2153,
+      "step": 1642
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1917774727673267e-05,
+      "loss": 1.2525,
+      "step": 1643
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.191759899275959e-05,
+      "loss": 1.2009,
+      "step": 1644
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1917423071029636e-05,
+      "loss": 1.1394,
+      "step": 1645
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1917246962486723e-05,
+      "loss": 1.2192,
+      "step": 1646
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1917070667134173e-05,
+      "loss": 1.1629,
+      "step": 1647
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.19168941849753e-05,
+      "loss": 1.1712,
+      "step": 1648
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1916717516013426e-05,
+      "loss": 1.0975,
+      "step": 1649
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1916540660251888e-05,
+      "loss": 1.2074,
+      "step": 1650
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.191636361769401e-05,
+      "loss": 1.1592,
+      "step": 1651
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1916186388343133e-05,
+      "loss": 1.1187,
+      "step": 1652
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1916008972202587e-05,
+      "loss": 1.0816,
+      "step": 1653
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1915831369275725e-05,
+      "loss": 1.2122,
+      "step": 1654
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1915653579565884e-05,
+      "loss": 1.1543,
+      "step": 1655
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1915475603076415e-05,
+      "loss": 1.1595,
+      "step": 1656
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1915297439810672e-05,
+      "loss": 1.1912,
+      "step": 1657
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.191511908977201e-05,
+      "loss": 1.1263,
+      "step": 1658
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1914940552963788e-05,
+      "loss": 1.2022,
+      "step": 1659
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1914761829389372e-05,
+      "loss": 1.2332,
+      "step": 1660
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1914582919052123e-05,
+      "loss": 1.1206,
+      "step": 1661
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1914403821955415e-05,
+      "loss": 1.2043,
+      "step": 1662
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.191422453810262e-05,
+      "loss": 1.2163,
+      "step": 1663
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.191404506749712e-05,
+      "loss": 1.1306,
+      "step": 1664
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.191386541014229e-05,
+      "loss": 1.1694,
+      "step": 1665
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1913685566041516e-05,
+      "loss": 1.1354,
+      "step": 1666
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1913505535198184e-05,
+      "loss": 1.2027,
+      "step": 1667
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1913325317615686e-05,
+      "loss": 1.1415,
+      "step": 1668
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.191314491329742e-05,
+      "loss": 1.0428,
+      "step": 1669
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1912964322246783e-05,
+      "loss": 1.0863,
+      "step": 1670
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1912783544467176e-05,
+      "loss": 1.1404,
+      "step": 1671
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1912602579962002e-05,
+      "loss": 1.1504,
+      "step": 1672
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1912421428734672e-05,
+      "loss": 1.2272,
+      "step": 1673
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1912240090788596e-05,
+      "loss": 1.1812,
+      "step": 1674
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1912058566127194e-05,
+      "loss": 1.1292,
+      "step": 1675
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1911876854753883e-05,
+      "loss": 1.1357,
+      "step": 1676
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1911694956672085e-05,
+      "loss": 1.0771,
+      "step": 1677
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.191151287188523e-05,
+      "loss": 1.152,
+      "step": 1678
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1911330600396743e-05,
+      "loss": 1.1488,
+      "step": 1679
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.191114814221006e-05,
+      "loss": 1.2145,
+      "step": 1680
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.191096549732862e-05,
+      "loss": 1.1064,
+      "step": 1681
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1910782665755863e-05,
+      "loss": 1.133,
+      "step": 1682
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1910599647495228e-05,
+      "loss": 1.1936,
+      "step": 1683
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1910416442550164e-05,
+      "loss": 1.1004,
+      "step": 1684
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1910233050924125e-05,
+      "loss": 1.1701,
+      "step": 1685
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1910049472620565e-05,
+      "loss": 1.2015,
+      "step": 1686
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.190986570764294e-05,
+      "loss": 1.1128,
+      "step": 1687
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1909681755994713e-05,
+      "loss": 1.1706,
+      "step": 1688
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1909497617679352e-05,
+      "loss": 1.1689,
+      "step": 1689
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.190931329270032e-05,
+      "loss": 1.154,
+      "step": 1690
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1909128781061088e-05,
+      "loss": 1.1031,
+      "step": 1691
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.190894408276514e-05,
+      "loss": 1.1568,
+      "step": 1692
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1908759197815946e-05,
+      "loss": 1.1614,
+      "step": 1693
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1908574126216997e-05,
+      "loss": 1.2332,
+      "step": 1694
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1908388867971773e-05,
+      "loss": 1.127,
+      "step": 1695
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1908203423083768e-05,
+      "loss": 1.1562,
+      "step": 1696
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.190801779155647e-05,
+      "loss": 1.1375,
+      "step": 1697
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.190783197339338e-05,
+      "loss": 1.2119,
+      "step": 1698
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1907645968597998e-05,
+      "loss": 1.1586,
+      "step": 1699
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1907459777173828e-05,
+      "loss": 1.1732,
+      "step": 1700
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1907273399124373e-05,
+      "loss": 1.2089,
+      "step": 1701
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1907086834453147e-05,
+      "loss": 1.1988,
+      "step": 1702
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1906900083163665e-05,
+      "loss": 1.1266,
+      "step": 1703
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1906713145259442e-05,
+      "loss": 1.1345,
+      "step": 1704
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1906526020744002e-05,
+      "loss": 1.1575,
+      "step": 1705
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1906338709620867e-05,
+      "loss": 1.1855,
+      "step": 1706
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1906151211893572e-05,
+      "loss": 1.083,
+      "step": 1707
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1905963527565637e-05,
+      "loss": 1.1318,
+      "step": 1708
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1905775656640613e-05,
+      "loss": 1.2026,
+      "step": 1709
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1905587599122023e-05,
+      "loss": 1.1278,
+      "step": 1710
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1905399355013422e-05,
+      "loss": 1.1233,
+      "step": 1711
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.190521092431835e-05,
+      "loss": 1.1616,
+      "step": 1712
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1905022307040354e-05,
+      "loss": 1.1225,
+      "step": 1713
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1904833503182993e-05,
+      "loss": 1.2365,
+      "step": 1714
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.190464451274982e-05,
+      "loss": 1.1086,
+      "step": 1715
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1904455335744396e-05,
+      "loss": 1.1692,
+      "step": 1716
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1904265972170285e-05,
+      "loss": 1.129,
+      "step": 1717
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.190407642203105e-05,
+      "loss": 1.0771,
+      "step": 1718
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1903886685330265e-05,
+      "loss": 1.1313,
+      "step": 1719
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1903696762071505e-05,
+      "loss": 1.1985,
+      "step": 1720
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.190350665225835e-05,
+      "loss": 1.1533,
+      "step": 1721
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1903316355894372e-05,
+      "loss": 1.1005,
+      "step": 1722
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1903125872983162e-05,
+      "loss": 1.094,
+      "step": 1723
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1902935203528308e-05,
+      "loss": 1.1625,
+      "step": 1724
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.19027443475334e-05,
+      "loss": 1.0892,
+      "step": 1725
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1902553305002033e-05,
+      "loss": 1.1679,
+      "step": 1726
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1902362075937806e-05,
+      "loss": 1.1611,
+      "step": 1727
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1902170660344324e-05,
+      "loss": 1.1659,
+      "step": 1728
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1901979058225188e-05,
+      "loss": 1.1775,
+      "step": 1729
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.190178726958401e-05,
+      "loss": 1.1737,
+      "step": 1730
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.19015952944244e-05,
+      "loss": 1.1438,
+      "step": 1731
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.190140313274998e-05,
+      "loss": 1.1915,
+      "step": 1732
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1901210784564366e-05,
+      "loss": 1.0572,
+      "step": 1733
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.190101824987118e-05,
+      "loss": 1.2148,
+      "step": 1734
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.190082552867405e-05,
+      "loss": 1.0944,
+      "step": 1735
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1900632620976604e-05,
+      "loss": 1.2167,
+      "step": 1736
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.190043952678248e-05,
+      "loss": 1.1378,
+      "step": 1737
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1900246246095316e-05,
+      "loss": 1.1848,
+      "step": 1738
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1900052778918745e-05,
+      "loss": 1.1523,
+      "step": 1739
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1899859125256418e-05,
+      "loss": 1.087,
+      "step": 1740
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1899665285111982e-05,
+      "loss": 1.1233,
+      "step": 1741
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.189947125848909e-05,
+      "loss": 1.1498,
+      "step": 1742
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.189927704539139e-05,
+      "loss": 1.1934,
+      "step": 1743
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1899082645822547e-05,
+      "loss": 1.1418,
+      "step": 1744
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.189888805978622e-05,
+      "loss": 1.1159,
+      "step": 1745
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1898693287286073e-05,
+      "loss": 1.1923,
+      "step": 1746
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1898498328325778e-05,
+      "loss": 1.2276,
+      "step": 1747
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.189830318290901e-05,
+      "loss": 1.1867,
+      "step": 1748
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.189810785103944e-05,
+      "loss": 1.1646,
+      "step": 1749
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1897912332720744e-05,
+      "loss": 1.2647,
+      "step": 1750
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1897716627956616e-05,
+      "loss": 1.1627,
+      "step": 1751
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.189752073675073e-05,
+      "loss": 1.183,
+      "step": 1752
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1897324659106786e-05,
+      "loss": 1.1424,
+      "step": 1753
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1897128395028473e-05,
+      "loss": 1.1135,
+      "step": 1754
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.189693194451949e-05,
+      "loss": 1.1716,
+      "step": 1755
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1896735307583534e-05,
+      "loss": 1.129,
+      "step": 1756
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1896538484224308e-05,
+      "loss": 1.1126,
+      "step": 1757
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1896341474445527e-05,
+      "loss": 1.1376,
+      "step": 1758
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1896144278250895e-05,
+      "loss": 1.0459,
+      "step": 1759
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.189594689564413e-05,
+      "loss": 1.1763,
+      "step": 1760
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1895749326628954e-05,
+      "loss": 1.1661,
+      "step": 1761
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.189555157120908e-05,
+      "loss": 1.078,
+      "step": 1762
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1895353629388233e-05,
+      "loss": 1.1462,
+      "step": 1763
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1895155501170154e-05,
+      "loss": 1.1542,
+      "step": 1764
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1894957186558558e-05,
+      "loss": 1.1683,
+      "step": 1765
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1894758685557193e-05,
+      "loss": 1.1355,
+      "step": 1766
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1894559998169798e-05,
+      "loss": 1.1429,
+      "step": 1767
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1894361124400105e-05,
+      "loss": 1.0825,
+      "step": 1768
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1894162064251875e-05,
+      "loss": 1.2525,
+      "step": 1769
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1893962817728843e-05,
+      "loss": 1.0895,
+      "step": 1770
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1893763384834772e-05,
+      "loss": 1.1098,
+      "step": 1771
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1893563765573415e-05,
+      "loss": 0.9976,
+      "step": 1772
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1893363959948533e-05,
+      "loss": 1.1622,
+      "step": 1773
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.189316396796389e-05,
+      "loss": 1.1982,
+      "step": 1774
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1892963789623254e-05,
+      "loss": 1.2104,
+      "step": 1775
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1892763424930392e-05,
+      "loss": 1.0978,
+      "step": 1776
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1892562873889083e-05,
+      "loss": 1.2226,
+      "step": 1777
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1892362136503103e-05,
+      "loss": 1.1586,
+      "step": 1778
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1892161212776232e-05,
+      "loss": 1.1595,
+      "step": 1779
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1891960102712252e-05,
+      "loss": 1.118,
+      "step": 1780
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.189175880631496e-05,
+      "loss": 1.1537,
+      "step": 1781
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1891557323588142e-05,
+      "loss": 1.0755,
+      "step": 1782
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.189135565453559e-05,
+      "loss": 1.1209,
+      "step": 1783
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.189115379916111e-05,
+      "loss": 1.2982,
+      "step": 1784
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.18909517574685e-05,
+      "loss": 1.0682,
+      "step": 1785
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.189074952946157e-05,
+      "loss": 1.2347,
+      "step": 1786
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1890547115144122e-05,
+      "loss": 1.1487,
+      "step": 1787
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1890344514519976e-05,
+      "loss": 1.1192,
+      "step": 1788
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1890141727592944e-05,
+      "loss": 1.13,
+      "step": 1789
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1889938754366844e-05,
+      "loss": 1.1203,
+      "step": 1790
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.188973559484551e-05,
+      "loss": 1.1953,
+      "step": 1791
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1889532249032757e-05,
+      "loss": 1.1409,
+      "step": 1792
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.188932871693242e-05,
+      "loss": 1.0668,
+      "step": 1793
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1889124998548333e-05,
+      "loss": 1.1977,
+      "step": 1794
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1888921093884336e-05,
+      "loss": 1.1639,
+      "step": 1795
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1888717002944262e-05,
+      "loss": 1.0581,
+      "step": 1796
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1888512725731966e-05,
+      "loss": 1.1837,
+      "step": 1797
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1888308262251287e-05,
+      "loss": 1.165,
+      "step": 1798
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1888103612506085e-05,
+      "loss": 1.0864,
+      "step": 1799
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1887898776500204e-05,
+      "loss": 1.0938,
+      "step": 1800
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.188769375423751e-05,
+      "loss": 1.1158,
+      "step": 1801
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1887488545721865e-05,
+      "loss": 1.0788,
+      "step": 1802
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.188728315095713e-05,
+      "loss": 1.1843,
+      "step": 1803
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1887077569947182e-05,
+      "loss": 1.0895,
+      "step": 1804
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1886871802695885e-05,
+      "loss": 1.1252,
+      "step": 1805
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1886665849207117e-05,
+      "loss": 1.0794,
+      "step": 1806
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.188645970948476e-05,
+      "loss": 1.038,
+      "step": 1807
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.18862533835327e-05,
+      "loss": 1.1039,
+      "step": 1808
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1886046871354813e-05,
+      "loss": 1.1412,
+      "step": 1809
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1885840172955e-05,
+      "loss": 1.2747,
+      "step": 1810
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1885633288337147e-05,
+      "loss": 1.2177,
+      "step": 1811
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1885426217505155e-05,
+      "loss": 1.1394,
+      "step": 1812
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1885218960462925e-05,
+      "loss": 1.1896,
+      "step": 1813
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.188501151721436e-05,
+      "loss": 1.1465,
+      "step": 1814
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.188480388776337e-05,
+      "loss": 1.1418,
+      "step": 1815
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1884596072113857e-05,
+      "loss": 1.1263,
+      "step": 1816
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1884388070269742e-05,
+      "loss": 1.1728,
+      "step": 1817
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1884179882234948e-05,
+      "loss": 1.1061,
+      "step": 1818
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1883971508013387e-05,
+      "loss": 1.1639,
+      "step": 1819
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1883762947608994e-05,
+      "loss": 1.176,
+      "step": 1820
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1883554201025688e-05,
+      "loss": 1.1739,
+      "step": 1821
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1883345268267407e-05,
+      "loss": 1.1272,
+      "step": 1822
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1883136149338084e-05,
+      "loss": 1.1567,
+      "step": 1823
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1882926844241663e-05,
+      "loss": 1.1068,
+      "step": 1824
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.188271735298208e-05,
+      "loss": 1.1311,
+      "step": 1825
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1882507675563286e-05,
+      "loss": 1.2169,
+      "step": 1826
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1882297811989226e-05,
+      "loss": 1.1691,
+      "step": 1827
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1882087762263858e-05,
+      "loss": 1.1489,
+      "step": 1828
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1881877526391134e-05,
+      "loss": 1.1589,
+      "step": 1829
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.188166710437502e-05,
+      "loss": 1.0629,
+      "step": 1830
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1881456496219473e-05,
+      "loss": 1.0789,
+      "step": 1831
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1881245701928468e-05,
+      "loss": 1.1504,
+      "step": 1832
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1881034721505967e-05,
+      "loss": 1.1347,
+      "step": 1833
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1880823554955955e-05,
+      "loss": 1.1746,
+      "step": 1834
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1880612202282396e-05,
+      "loss": 1.0759,
+      "step": 1835
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.188040066348928e-05,
+      "loss": 1.2016,
+      "step": 1836
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.188018893858059e-05,
+      "loss": 1.1885,
+      "step": 1837
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.187997702756032e-05,
+      "loss": 1.2529,
+      "step": 1838
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1879764930432453e-05,
+      "loss": 1.1072,
+      "step": 1839
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1879552647200984e-05,
+      "loss": 1.0596,
+      "step": 1840
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.187934017786992e-05,
+      "loss": 1.1201,
+      "step": 1841
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1879127522443257e-05,
+      "loss": 1.1488,
+      "step": 1842
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1878914680925002e-05,
+      "loss": 1.1668,
+      "step": 1843
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1878701653319166e-05,
+      "loss": 1.0697,
+      "step": 1844
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1878488439629755e-05,
+      "loss": 1.128,
+      "step": 1845
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1878275039860796e-05,
+      "loss": 1.1711,
+      "step": 1846
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.18780614540163e-05,
+      "loss": 1.1587,
+      "step": 1847
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1877847682100295e-05,
+      "loss": 1.1515,
+      "step": 1848
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1877633724116807e-05,
+      "loss": 1.1918,
+      "step": 1849
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1877419580069864e-05,
+      "loss": 1.0656,
+      "step": 1850
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.18772052499635e-05,
+      "loss": 1.1798,
+      "step": 1851
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1876990733801756e-05,
+      "loss": 1.2076,
+      "step": 1852
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1876776031588672e-05,
+      "loss": 1.1001,
+      "step": 1853
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1876561143328288e-05,
+      "loss": 1.1035,
+      "step": 1854
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.187634606902466e-05,
+      "loss": 1.0837,
+      "step": 1855
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1876130808681827e-05,
+      "loss": 1.1563,
+      "step": 1856
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1875915362303855e-05,
+      "loss": 1.0608,
+      "step": 1857
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1875699729894798e-05,
+      "loss": 1.1459,
+      "step": 1858
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1875483911458715e-05,
+      "loss": 1.1703,
+      "step": 1859
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1875267906999675e-05,
+      "loss": 1.1219,
+      "step": 1860
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1875051716521747e-05,
+      "loss": 1.1052,
+      "step": 1861
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1874835340029005e-05,
+      "loss": 1.0807,
+      "step": 1862
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.187461877752552e-05,
+      "loss": 1.1158,
+      "step": 1863
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1874402029015375e-05,
+      "loss": 1.014,
+      "step": 1864
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.187418509450265e-05,
+      "loss": 1.0928,
+      "step": 1865
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1873967973991433e-05,
+      "loss": 1.1976,
+      "step": 1866
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1873750667485812e-05,
+      "loss": 1.1411,
+      "step": 1867
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1873533174989885e-05,
+      "loss": 1.1655,
+      "step": 1868
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1873315496507746e-05,
+      "loss": 1.1861,
+      "step": 1869
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1873097632043494e-05,
+      "loss": 1.187,
+      "step": 1870
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.187287958160124e-05,
+      "loss": 1.231,
+      "step": 1871
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1872661345185077e-05,
+      "loss": 1.1295,
+      "step": 1872
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.187244292279913e-05,
+      "loss": 1.0718,
+      "step": 1873
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1872224314447503e-05,
+      "loss": 1.0763,
+      "step": 1874
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1872005520134327e-05,
+      "loss": 1.0419,
+      "step": 1875
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1871786539863708e-05,
+      "loss": 1.1092,
+      "step": 1876
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1871567373639784e-05,
+      "loss": 1.105,
+      "step": 1877
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1871348021466674e-05,
+      "loss": 1.0699,
+      "step": 1878
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1871128483348514e-05,
+      "loss": 1.0104,
+      "step": 1879
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1870908759289437e-05,
+      "loss": 1.0955,
+      "step": 1880
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1870688849293584e-05,
+      "loss": 1.112,
+      "step": 1881
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.18704687533651e-05,
+      "loss": 1.1024,
+      "step": 1882
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1870248471508124e-05,
+      "loss": 1.1091,
+      "step": 1883
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1870028003726812e-05,
+      "loss": 1.1067,
+      "step": 1884
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1869807350025314e-05,
+      "loss": 1.1314,
+      "step": 1885
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.186958651040779e-05,
+      "loss": 1.187,
+      "step": 1886
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1869365484878392e-05,
+      "loss": 1.0674,
+      "step": 1887
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.186914427344129e-05,
+      "loss": 1.2017,
+      "step": 1888
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.186892287610065e-05,
+      "loss": 1.1318,
+      "step": 1889
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.186870129286064e-05,
+      "loss": 1.1449,
+      "step": 1890
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1868479523725437e-05,
+      "loss": 1.221,
+      "step": 1891
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1868257568699216e-05,
+      "loss": 1.0836,
+      "step": 1892
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.186803542778616e-05,
+      "loss": 1.1203,
+      "step": 1893
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.186781310099045e-05,
+      "loss": 1.0086,
+      "step": 1894
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1867590588316278e-05,
+      "loss": 1.1005,
+      "step": 1895
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1867367889767833e-05,
+      "loss": 1.0662,
+      "step": 1896
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1867145005349314e-05,
+      "loss": 1.0747,
+      "step": 1897
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.186692193506491e-05,
+      "loss": 1.1402,
+      "step": 1898
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.186669867891883e-05,
+      "loss": 1.0969,
+      "step": 1899
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.186647523691528e-05,
+      "loss": 1.1767,
+      "step": 1900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.186625160905847e-05,
+      "loss": 1.1718,
+      "step": 1901
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1866027795352605e-05,
+      "loss": 1.0982,
+      "step": 1902
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1865803795801906e-05,
+      "loss": 1.1662,
+      "step": 1903
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1865579610410596e-05,
+      "loss": 1.1003,
+      "step": 1904
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1865355239182893e-05,
+      "loss": 1.1035,
+      "step": 1905
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1865130682123022e-05,
+      "loss": 1.1791,
+      "step": 1906
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1864905939235216e-05,
+      "loss": 1.1458,
+      "step": 1907
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1864681010523712e-05,
+      "loss": 1.1244,
+      "step": 1908
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.186445589599274e-05,
+      "loss": 1.228,
+      "step": 1909
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1864230595646544e-05,
+      "loss": 1.1172,
+      "step": 1910
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1864005109489366e-05,
+      "loss": 1.1199,
+      "step": 1911
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1863779437525455e-05,
+      "loss": 1.0373,
+      "step": 1912
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1863553579759063e-05,
+      "loss": 1.1261,
+      "step": 1913
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.186332753619444e-05,
+      "loss": 1.1323,
+      "step": 1914
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1863101306835846e-05,
+      "loss": 1.1156,
+      "step": 1915
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.186287489168755e-05,
+      "loss": 1.1328,
+      "step": 1916
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1862648290753808e-05,
+      "loss": 1.1907,
+      "step": 1917
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.186242150403889e-05,
+      "loss": 1.1396,
+      "step": 1918
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1862194531547065e-05,
+      "loss": 1.2005,
+      "step": 1919
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.186196737328262e-05,
+      "loss": 1.0389,
+      "step": 1920
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.186174002924982e-05,
+      "loss": 1.1385,
+      "step": 1921
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1861512499452957e-05,
+      "loss": 1.2017,
+      "step": 1922
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.186128478389631e-05,
+      "loss": 1.1115,
+      "step": 1923
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.186105688258418e-05,
+      "loss": 1.1185,
+      "step": 1924
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1860828795520844e-05,
+      "loss": 1.1476,
+      "step": 1925
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.186060052271061e-05,
+      "loss": 1.1003,
+      "step": 1926
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1860372064157774e-05,
+      "loss": 1.1,
+      "step": 1927
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.186014341986664e-05,
+      "loss": 1.0112,
+      "step": 1928
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1859914589841515e-05,
+      "loss": 1.1585,
+      "step": 1929
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.185968557408671e-05,
+      "loss": 1.13,
+      "step": 1930
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1859456372606533e-05,
+      "loss": 1.0911,
+      "step": 1931
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1859226985405314e-05,
+      "loss": 1.0821,
+      "step": 1932
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1858997412487363e-05,
+      "loss": 1.1106,
+      "step": 1933
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1858767653857005e-05,
+      "loss": 1.0918,
+      "step": 1934
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1858537709518573e-05,
+      "loss": 1.2586,
+      "step": 1935
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.18583075794764e-05,
+      "loss": 1.2229,
+      "step": 1936
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.185807726373481e-05,
+      "loss": 1.1744,
+      "step": 1937
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.185784676229816e-05,
+      "loss": 1.1277,
+      "step": 1938
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1857616075170772e-05,
+      "loss": 1.234,
+      "step": 1939
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1857385202357e-05,
+      "loss": 1.1485,
+      "step": 1940
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1857154143861193e-05,
+      "loss": 1.1733,
+      "step": 1941
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1856922899687704e-05,
+      "loss": 1.1411,
+      "step": 1942
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.185669146984089e-05,
+      "loss": 1.1624,
+      "step": 1943
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.185645985432511e-05,
+      "loss": 1.0859,
+      "step": 1944
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.185622805314472e-05,
+      "loss": 1.2074,
+      "step": 1945
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1855996066304097e-05,
+      "loss": 1.1081,
+      "step": 1946
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1855763893807605e-05,
+      "loss": 1.0686,
+      "step": 1947
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1855531535659613e-05,
+      "loss": 1.125,
+      "step": 1948
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.185529899186451e-05,
+      "loss": 1.0788,
+      "step": 1949
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1855066262426664e-05,
+      "loss": 1.1628,
+      "step": 1950
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.185483334735047e-05,
+      "loss": 1.0693,
+      "step": 1951
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1854600246640306e-05,
+      "loss": 1.0352,
+      "step": 1952
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1854366960300567e-05,
+      "loss": 1.1078,
+      "step": 1953
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1854133488335648e-05,
+      "loss": 1.0941,
+      "step": 1954
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1853899830749947e-05,
+      "loss": 1.1449,
+      "step": 1955
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.185366598754786e-05,
+      "loss": 1.1009,
+      "step": 1956
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1853431958733802e-05,
+      "loss": 1.2041,
+      "step": 1957
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1853197744312172e-05,
+      "loss": 1.1417,
+      "step": 1958
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1852963344287388e-05,
+      "loss": 1.0737,
+      "step": 1959
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1852728758663858e-05,
+      "loss": 1.0588,
+      "step": 1960
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1852493987446008e-05,
+      "loss": 1.0461,
+      "step": 1961
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.185225903063826e-05,
+      "loss": 1.144,
+      "step": 1962
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1852023888245037e-05,
+      "loss": 1.1842,
+      "step": 1963
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1851788560270767e-05,
+      "loss": 1.1211,
+      "step": 1964
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1851553046719885e-05,
+      "loss": 1.178,
+      "step": 1965
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1851317347596828e-05,
+      "loss": 1.1665,
+      "step": 1966
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1851081462906036e-05,
+      "loss": 1.1211,
+      "step": 1967
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1850845392651952e-05,
+      "loss": 1.161,
+      "step": 1968
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.185060913683902e-05,
+      "loss": 1.1913,
+      "step": 1969
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1850372695471695e-05,
+      "loss": 1.136,
+      "step": 1970
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.185013606855443e-05,
+      "loss": 1.1984,
+      "step": 1971
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1849899256091673e-05,
+      "loss": 1.0275,
+      "step": 1972
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.18496622580879e-05,
+      "loss": 1.0302,
+      "step": 1973
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1849425074547566e-05,
+      "loss": 1.1449,
+      "step": 1974
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.184918770547514e-05,
+      "loss": 1.0565,
+      "step": 1975
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1848950150875094e-05,
+      "loss": 1.1293,
+      "step": 1976
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1848712410751903e-05,
+      "loss": 1.0711,
+      "step": 1977
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1848474485110045e-05,
+      "loss": 1.1466,
+      "step": 1978
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1848236373954e-05,
+      "loss": 1.1728,
+      "step": 1979
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1847998077288256e-05,
+      "loss": 1.0817,
+      "step": 1980
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1847759595117304e-05,
+      "loss": 1.1825,
+      "step": 1981
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.184752092744563e-05,
+      "loss": 1.1623,
+      "step": 1982
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.184728207427773e-05,
+      "loss": 1.1342,
+      "step": 1983
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.184704303561811e-05,
+      "loss": 1.1054,
+      "step": 1984
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1846803811471266e-05,
+      "loss": 1.1511,
+      "step": 1985
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.184656440184171e-05,
+      "loss": 1.1672,
+      "step": 1986
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1846324806733946e-05,
+      "loss": 1.1079,
+      "step": 1987
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.184608502615249e-05,
+      "loss": 1.199,
+      "step": 1988
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.184584506010186e-05,
+      "loss": 1.1084,
+      "step": 1989
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.184560490858657e-05,
+      "loss": 1.1093,
+      "step": 1990
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1845364571611152e-05,
+      "loss": 1.1365,
+      "step": 1991
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1845124049180133e-05,
+      "loss": 1.1785,
+      "step": 1992
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1844883341298036e-05,
+      "loss": 1.1338,
+      "step": 1993
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1844642447969402e-05,
+      "loss": 1.2396,
+      "step": 1994
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1844401369198765e-05,
+      "loss": 1.1646,
+      "step": 1995
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1844160104990666e-05,
+      "loss": 1.1537,
+      "step": 1996
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1843918655349654e-05,
+      "loss": 1.1334,
+      "step": 1997
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1843677020280272e-05,
+      "loss": 1.1561,
+      "step": 1998
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1843435199787073e-05,
+      "loss": 1.1658,
+      "step": 1999
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1843193193874615e-05,
+      "loss": 1.1475,
+      "step": 2000
+    },
+    {
+      "epoch": 0.08,
+      "eval_text_loss": 0.8094040155410767,
+      "eval_text_runtime": 1.3903,
+      "eval_text_samples_per_second": 53.224,
+      "eval_text_steps_per_second": 1.438,
+      "step": 2000
+    },
+    {
+      "epoch": 0.08,
+      "eval_svit_loss": 1.2210115194320679,
+      "eval_svit_runtime": 4.9805,
+      "eval_svit_samples_per_second": 136.734,
+      "eval_svit_steps_per_second": 2.209,
+      "step": 2000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1842951002547455e-05,
+      "loss": 1.1087,
+      "step": 2001
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1842708625810153e-05,
+      "loss": 1.0956,
+      "step": 2002
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.184246606366728e-05,
+      "loss": 1.0957,
+      "step": 2003
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1842223316123394e-05,
+      "loss": 1.1135,
+      "step": 2004
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.184198038318308e-05,
+      "loss": 1.1131,
+      "step": 2005
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.184173726485091e-05,
+      "loss": 1.0328,
+      "step": 2006
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.184149396113146e-05,
+      "loss": 1.0905,
+      "step": 2007
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1841250472029317e-05,
+      "loss": 1.0478,
+      "step": 2008
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1841006797549065e-05,
+      "loss": 1.2312,
+      "step": 2009
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1840762937695297e-05,
+      "loss": 1.0304,
+      "step": 2010
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1840518892472607e-05,
+      "loss": 1.1506,
+      "step": 2011
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1840274661885586e-05,
+      "loss": 1.1673,
+      "step": 2012
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.184003024593884e-05,
+      "loss": 1.1034,
+      "step": 2013
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.183978564463697e-05,
+      "loss": 1.0703,
+      "step": 2014
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1839540857984584e-05,
+      "loss": 1.0481,
+      "step": 2015
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.18392958859863e-05,
+      "loss": 1.1459,
+      "step": 2016
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1839050728646722e-05,
+      "loss": 1.0377,
+      "step": 2017
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1838805385970473e-05,
+      "loss": 1.1403,
+      "step": 2018
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1838559857962175e-05,
+      "loss": 1.1059,
+      "step": 2019
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1838314144626453e-05,
+      "loss": 1.1329,
+      "step": 2020
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1838068245967932e-05,
+      "loss": 1.1283,
+      "step": 2021
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.183782216199125e-05,
+      "loss": 1.0612,
+      "step": 2022
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1837575892701034e-05,
+      "loss": 1.176,
+      "step": 2023
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1837329438101928e-05,
+      "loss": 1.1465,
+      "step": 2024
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.183708279819858e-05,
+      "loss": 1.1435,
+      "step": 2025
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1836835972995623e-05,
+      "loss": 1.2979,
+      "step": 2026
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1836588962497718e-05,
+      "loss": 1.0706,
+      "step": 2027
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1836341766709513e-05,
+      "loss": 1.1471,
+      "step": 2028
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1836094385635662e-05,
+      "loss": 1.1477,
+      "step": 2029
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1835846819280832e-05,
+      "loss": 1.0773,
+      "step": 2030
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1835599067649678e-05,
+      "loss": 1.1432,
+      "step": 2031
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1835351130746874e-05,
+      "loss": 1.1279,
+      "step": 2032
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1835103008577088e-05,
+      "loss": 1.1153,
+      "step": 2033
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1834854701144987e-05,
+      "loss": 1.1176,
+      "step": 2034
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1834606208455258e-05,
+      "loss": 1.1431,
+      "step": 2035
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1834357530512582e-05,
+      "loss": 1.1407,
+      "step": 2036
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1834108667321636e-05,
+      "loss": 1.0734,
+      "step": 2037
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1833859618887112e-05,
+      "loss": 1.1291,
+      "step": 2038
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.18336103852137e-05,
+      "loss": 1.1658,
+      "step": 2039
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1833360966306097e-05,
+      "loss": 1.1208,
+      "step": 2040
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1833111362168997e-05,
+      "loss": 1.1714,
+      "step": 2041
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1832861572807107e-05,
+      "loss": 1.1065,
+      "step": 2042
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.183261159822513e-05,
+      "loss": 1.1468,
+      "step": 2043
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.183236143842777e-05,
+      "loss": 1.0811,
+      "step": 2044
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1832111093419752e-05,
+      "loss": 1.1503,
+      "step": 2045
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1831860563205777e-05,
+      "loss": 1.161,
+      "step": 2046
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.183160984779057e-05,
+      "loss": 1.2847,
+      "step": 2047
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1831358947178856e-05,
+      "loss": 1.1293,
+      "step": 2048
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.183110786137536e-05,
+      "loss": 1.2779,
+      "step": 2049
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1830856590384815e-05,
+      "loss": 1.0971,
+      "step": 2050
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1830605134211945e-05,
+      "loss": 1.1539,
+      "step": 2051
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1830353492861494e-05,
+      "loss": 1.1595,
+      "step": 2052
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1830101666338196e-05,
+      "loss": 1.1797,
+      "step": 2053
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.18298496546468e-05,
+      "loss": 1.1845,
+      "step": 2054
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1829597457792055e-05,
+      "loss": 1.1583,
+      "step": 2055
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1829345075778706e-05,
+      "loss": 1.092,
+      "step": 2056
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.182909250861151e-05,
+      "loss": 1.1772,
+      "step": 2057
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1828839756295224e-05,
+      "loss": 1.1609,
+      "step": 2058
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.182858681883461e-05,
+      "loss": 1.1095,
+      "step": 2059
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.182833369623443e-05,
+      "loss": 1.09,
+      "step": 2060
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1828080388499455e-05,
+      "loss": 1.2386,
+      "step": 2061
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1827826895634453e-05,
+      "loss": 1.203,
+      "step": 2062
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.18275732176442e-05,
+      "loss": 1.1263,
+      "step": 2063
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1827319354533482e-05,
+      "loss": 1.1164,
+      "step": 2064
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.182706530630707e-05,
+      "loss": 1.243,
+      "step": 2065
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1826811072969756e-05,
+      "loss": 1.1367,
+      "step": 2066
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1826556654526324e-05,
+      "loss": 1.0906,
+      "step": 2067
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.182630205098157e-05,
+      "loss": 1.1098,
+      "step": 2068
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1826047262340293e-05,
+      "loss": 1.1026,
+      "step": 2069
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1825792288607285e-05,
+      "loss": 1.1673,
+      "step": 2070
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1825537129787356e-05,
+      "loss": 1.1525,
+      "step": 2071
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1825281785885304e-05,
+      "loss": 1.0754,
+      "step": 2072
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.182502625690595e-05,
+      "loss": 1.0978,
+      "step": 2073
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1824770542854095e-05,
+      "loss": 1.1445,
+      "step": 2074
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.182451464373457e-05,
+      "loss": 1.1348,
+      "step": 2075
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1824258559552183e-05,
+      "loss": 1.1039,
+      "step": 2076
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1824002290311763e-05,
+      "loss": 1.187,
+      "step": 2077
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1823745836018133e-05,
+      "loss": 1.2196,
+      "step": 2078
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1823489196676133e-05,
+      "loss": 1.2107,
+      "step": 2079
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.182323237229059e-05,
+      "loss": 1.0595,
+      "step": 2080
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.182297536286634e-05,
+      "loss": 1.1232,
+      "step": 2081
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1822718168408232e-05,
+      "loss": 1.1027,
+      "step": 2082
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1822460788921104e-05,
+      "loss": 1.1046,
+      "step": 2083
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1822203224409806e-05,
+      "loss": 1.1877,
+      "step": 2084
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1821945474879195e-05,
+      "loss": 1.1502,
+      "step": 2085
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1821687540334117e-05,
+      "loss": 1.1392,
+      "step": 2086
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.182142942077944e-05,
+      "loss": 1.0862,
+      "step": 2087
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1821171116220013e-05,
+      "loss": 1.1113,
+      "step": 2088
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.182091262666072e-05,
+      "loss": 1.0865,
+      "step": 2089
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1820653952106413e-05,
+      "loss": 1.1462,
+      "step": 2090
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1820395092561973e-05,
+      "loss": 1.1249,
+      "step": 2091
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1820136048032277e-05,
+      "loss": 1.0707,
+      "step": 2092
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.18198768185222e-05,
+      "loss": 1.1331,
+      "step": 2093
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.181961740403663e-05,
+      "loss": 1.1483,
+      "step": 2094
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1819357804580448e-05,
+      "loss": 1.1942,
+      "step": 2095
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.181909802015855e-05,
+      "loss": 1.1325,
+      "step": 2096
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1818838050775826e-05,
+      "loss": 1.2256,
+      "step": 2097
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1818577896437172e-05,
+      "loss": 1.0193,
+      "step": 2098
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1818317557147496e-05,
+      "loss": 1.1643,
+      "step": 2099
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1818057032911688e-05,
+      "loss": 1.1021,
+      "step": 2100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1817796323734666e-05,
+      "loss": 1.1408,
+      "step": 2101
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.181753542962134e-05,
+      "loss": 1.0296,
+      "step": 2102
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.181727435057662e-05,
+      "loss": 1.1407,
+      "step": 2103
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1817013086605433e-05,
+      "loss": 1.1396,
+      "step": 2104
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1816751637712688e-05,
+      "loss": 1.1808,
+      "step": 2105
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1816490003903317e-05,
+      "loss": 1.1472,
+      "step": 2106
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.181622818518225e-05,
+      "loss": 1.1073,
+      "step": 2107
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1815966181554414e-05,
+      "loss": 1.163,
+      "step": 2108
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.181570399302475e-05,
+      "loss": 1.1417,
+      "step": 2109
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.181544161959819e-05,
+      "loss": 1.1571,
+      "step": 2110
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.181517906127968e-05,
+      "loss": 1.0958,
+      "step": 2111
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1814916318074164e-05,
+      "loss": 1.0863,
+      "step": 2112
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1814653389986598e-05,
+      "loss": 1.1533,
+      "step": 2113
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1814390277021924e-05,
+      "loss": 1.1054,
+      "step": 2114
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1814126979185106e-05,
+      "loss": 1.1144,
+      "step": 2115
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1813863496481098e-05,
+      "loss": 1.1376,
+      "step": 2116
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1813599828914873e-05,
+      "loss": 1.117,
+      "step": 2117
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.181333597649139e-05,
+      "loss": 1.1492,
+      "step": 2118
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1813071939215614e-05,
+      "loss": 1.1545,
+      "step": 2119
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1812807717092528e-05,
+      "loss": 1.1376,
+      "step": 2120
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1812543310127106e-05,
+      "loss": 1.0953,
+      "step": 2121
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1812278718324334e-05,
+      "loss": 1.1081,
+      "step": 2122
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1812013941689183e-05,
+      "loss": 1.1804,
+      "step": 2123
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.181174898022665e-05,
+      "loss": 1.0213,
+      "step": 2124
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1811483833941727e-05,
+      "loss": 1.1153,
+      "step": 2125
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1811218502839406e-05,
+      "loss": 1.1685,
+      "step": 2126
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1810952986924684e-05,
+      "loss": 1.1877,
+      "step": 2127
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1810687286202565e-05,
+      "loss": 1.1629,
+      "step": 2128
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.181042140067805e-05,
+      "loss": 1.0995,
+      "step": 2129
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1810155330356148e-05,
+      "loss": 1.0824,
+      "step": 2130
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1809889075241875e-05,
+      "loss": 1.1586,
+      "step": 2131
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1809622635340246e-05,
+      "loss": 1.1353,
+      "step": 2132
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1809356010656275e-05,
+      "loss": 1.1944,
+      "step": 2133
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1809089201194988e-05,
+      "loss": 1.2152,
+      "step": 2134
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.180882220696141e-05,
+      "loss": 1.0754,
+      "step": 2135
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.180855502796057e-05,
+      "loss": 1.0879,
+      "step": 2136
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1808287664197505e-05,
+      "loss": 1.0997,
+      "step": 2137
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.180802011567724e-05,
+      "loss": 1.1292,
+      "step": 2138
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.180775238240483e-05,
+      "loss": 1.0478,
+      "step": 2139
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1807484464385306e-05,
+      "loss": 1.1428,
+      "step": 2140
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1807216361623723e-05,
+      "loss": 1.082,
+      "step": 2141
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1806948074125125e-05,
+      "loss": 1.1605,
+      "step": 2142
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.180667960189457e-05,
+      "loss": 1.0596,
+      "step": 2143
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.180641094493711e-05,
+      "loss": 1.2047,
+      "step": 2144
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.180614210325781e-05,
+      "loss": 1.0881,
+      "step": 2145
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1805873076861735e-05,
+      "loss": 1.1433,
+      "step": 2146
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1805603865753953e-05,
+      "loss": 1.1393,
+      "step": 2147
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.180533446993953e-05,
+      "loss": 1.1412,
+      "step": 2148
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1805064889423542e-05,
+      "loss": 1.1689,
+      "step": 2149
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.180479512421107e-05,
+      "loss": 1.1785,
+      "step": 2150
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1804525174307196e-05,
+      "loss": 1.1933,
+      "step": 2151
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1804255039717e-05,
+      "loss": 1.1082,
+      "step": 2152
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1803984720445577e-05,
+      "loss": 1.1328,
+      "step": 2153
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1803714216498014e-05,
+      "loss": 1.0907,
+      "step": 2154
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1803443527879406e-05,
+      "loss": 1.1138,
+      "step": 2155
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1803172654594856e-05,
+      "loss": 1.1521,
+      "step": 2156
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1802901596649463e-05,
+      "loss": 1.1962,
+      "step": 2157
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1802630354048336e-05,
+      "loss": 1.1115,
+      "step": 2158
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.180235892679658e-05,
+      "loss": 1.0371,
+      "step": 2159
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.180208731489931e-05,
+      "loss": 1.1787,
+      "step": 2160
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1801815518361646e-05,
+      "loss": 1.0348,
+      "step": 2161
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.18015435371887e-05,
+      "loss": 1.1464,
+      "step": 2162
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1801271371385602e-05,
+      "loss": 1.1259,
+      "step": 2163
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1800999020957476e-05,
+      "loss": 1.1382,
+      "step": 2164
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1800726485909456e-05,
+      "loss": 1.09,
+      "step": 2165
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.180045376624667e-05,
+      "loss": 1.0506,
+      "step": 2166
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1800180861974256e-05,
+      "loss": 1.2187,
+      "step": 2167
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1799907773097356e-05,
+      "loss": 1.1036,
+      "step": 2168
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1799634499621117e-05,
+      "loss": 1.1518,
+      "step": 2169
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.179936104155068e-05,
+      "loss": 1.1726,
+      "step": 2170
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1799087398891204e-05,
+      "loss": 1.1304,
+      "step": 2171
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1798813571647836e-05,
+      "loss": 1.1128,
+      "step": 2172
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.179853955982574e-05,
+      "loss": 1.183,
+      "step": 2173
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1798265363430076e-05,
+      "loss": 1.0863,
+      "step": 2174
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1797990982466007e-05,
+      "loss": 1.2109,
+      "step": 2175
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1797716416938704e-05,
+      "loss": 1.1534,
+      "step": 2176
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1797441666853333e-05,
+      "loss": 1.1042,
+      "step": 2177
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.179716673221508e-05,
+      "loss": 1.1262,
+      "step": 2178
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1796891613029115e-05,
+      "loss": 1.1173,
+      "step": 2179
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.179661630930062e-05,
+      "loss": 1.1731,
+      "step": 2180
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1796340821034788e-05,
+      "loss": 1.1955,
+      "step": 2181
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1796065148236804e-05,
+      "loss": 1.2396,
+      "step": 2182
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1795789290911862e-05,
+      "loss": 1.1049,
+      "step": 2183
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1795513249065156e-05,
+      "loss": 1.0185,
+      "step": 2184
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1795237022701888e-05,
+      "loss": 1.0525,
+      "step": 2185
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.179496061182726e-05,
+      "loss": 1.173,
+      "step": 2186
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.179468401644648e-05,
+      "loss": 1.1318,
+      "step": 2187
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1794407236564754e-05,
+      "loss": 1.1813,
+      "step": 2188
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1794130272187303e-05,
+      "loss": 1.0833,
+      "step": 2189
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.179385312331934e-05,
+      "loss": 1.1243,
+      "step": 2190
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1793575789966084e-05,
+      "loss": 1.1746,
+      "step": 2191
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.179329827213276e-05,
+      "loss": 1.1191,
+      "step": 2192
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.17930205698246e-05,
+      "loss": 1.1918,
+      "step": 2193
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.179274268304683e-05,
+      "loss": 1.0464,
+      "step": 2194
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1792464611804684e-05,
+      "loss": 1.1634,
+      "step": 2195
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1792186356103404e-05,
+      "loss": 1.1437,
+      "step": 2196
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.179190791594823e-05,
+      "loss": 1.1394,
+      "step": 2197
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1791629291344405e-05,
+      "loss": 1.1648,
+      "step": 2198
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1791350482297178e-05,
+      "loss": 1.1229,
+      "step": 2199
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1791071488811804e-05,
+      "loss": 1.1721,
+      "step": 2200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1790792310893536e-05,
+      "loss": 1.1387,
+      "step": 2201
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1790512948547634e-05,
+      "loss": 1.1576,
+      "step": 2202
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1790233401779356e-05,
+      "loss": 1.206,
+      "step": 2203
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1789953670593973e-05,
+      "loss": 1.0956,
+      "step": 2204
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1789673754996748e-05,
+      "loss": 1.0217,
+      "step": 2205
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1789393654992963e-05,
+      "loss": 1.1514,
+      "step": 2206
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1789113370587885e-05,
+      "loss": 1.1629,
+      "step": 2207
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.17888329017868e-05,
+      "loss": 1.1382,
+      "step": 2208
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.178855224859499e-05,
+      "loss": 1.1196,
+      "step": 2209
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.178827141101774e-05,
+      "loss": 1.0607,
+      "step": 2210
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.178799038906034e-05,
+      "loss": 1.1023,
+      "step": 2211
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1787709182728083e-05,
+      "loss": 1.1674,
+      "step": 2212
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.178742779202627e-05,
+      "loss": 1.0849,
+      "step": 2213
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1787146216960197e-05,
+      "loss": 1.0967,
+      "step": 2214
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.178686445753517e-05,
+      "loss": 1.0965,
+      "step": 2215
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1786582513756497e-05,
+      "loss": 1.15,
+      "step": 2216
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1786300385629488e-05,
+      "loss": 1.1817,
+      "step": 2217
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.178601807315946e-05,
+      "loss": 1.2055,
+      "step": 2218
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1785735576351724e-05,
+      "loss": 1.0901,
+      "step": 2219
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1785452895211607e-05,
+      "loss": 1.1835,
+      "step": 2220
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1785170029744433e-05,
+      "loss": 1.1997,
+      "step": 2221
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1784886979955528e-05,
+      "loss": 1.1465,
+      "step": 2222
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.178460374585023e-05,
+      "loss": 1.0469,
+      "step": 2223
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1784320327433866e-05,
+      "loss": 1.1036,
+      "step": 2224
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.178403672471178e-05,
+      "loss": 1.1411,
+      "step": 2225
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1783752937689313e-05,
+      "loss": 1.1332,
+      "step": 2226
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.178346896637181e-05,
+      "loss": 1.0847,
+      "step": 2227
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.178318481076462e-05,
+      "loss": 1.1461,
+      "step": 2228
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1782900470873097e-05,
+      "loss": 1.1782,
+      "step": 2229
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1782615946702597e-05,
+      "loss": 1.1216,
+      "step": 2230
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.178233123825848e-05,
+      "loss": 1.2262,
+      "step": 2231
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1782046345546103e-05,
+      "loss": 1.1683,
+      "step": 2232
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1781761268570843e-05,
+      "loss": 1.0357,
+      "step": 2233
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.178147600733806e-05,
+      "loss": 1.0965,
+      "step": 2234
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1781190561853132e-05,
+      "loss": 1.0744,
+      "step": 2235
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1780904932121436e-05,
+      "loss": 1.1806,
+      "step": 2236
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1780619118148355e-05,
+      "loss": 1.0935,
+      "step": 2237
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1780333119939265e-05,
+      "loss": 1.0935,
+      "step": 2238
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.178004693749956e-05,
+      "loss": 1.1544,
+      "step": 2239
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1779760570834627e-05,
+      "loss": 1.088,
+      "step": 2240
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1779474019949865e-05,
+      "loss": 1.176,
+      "step": 2241
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1779187284850666e-05,
+      "loss": 1.1237,
+      "step": 2242
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1778900365542438e-05,
+      "loss": 1.1187,
+      "step": 2243
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1778613262030578e-05,
+      "loss": 1.1011,
+      "step": 2244
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1778325974320498e-05,
+      "loss": 1.1707,
+      "step": 2245
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.177803850241761e-05,
+      "loss": 1.1186,
+      "step": 2246
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1777750846327328e-05,
+      "loss": 1.2018,
+      "step": 2247
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.177746300605507e-05,
+      "loss": 1.1397,
+      "step": 2248
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1777174981606263e-05,
+      "loss": 1.1652,
+      "step": 2249
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1776886772986326e-05,
+      "loss": 1.149,
+      "step": 2250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.177659838020069e-05,
+      "loss": 1.2002,
+      "step": 2251
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.177630980325479e-05,
+      "loss": 1.0789,
+      "step": 2252
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1776021042154055e-05,
+      "loss": 1.1392,
+      "step": 2253
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.177573209690393e-05,
+      "loss": 1.1098,
+      "step": 2254
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.177544296750986e-05,
+      "loss": 1.1551,
+      "step": 2255
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1775153653977285e-05,
+      "loss": 1.1876,
+      "step": 2256
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.177486415631166e-05,
+      "loss": 1.1352,
+      "step": 2257
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1774574474518434e-05,
+      "loss": 1.0835,
+      "step": 2258
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1774284608603065e-05,
+      "loss": 1.0315,
+      "step": 2259
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1773994558571017e-05,
+      "loss": 1.2105,
+      "step": 2260
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.177370432442775e-05,
+      "loss": 1.1452,
+      "step": 2261
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.177341390617873e-05,
+      "loss": 1.1563,
+      "step": 2262
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.177312330382943e-05,
+      "loss": 1.1065,
+      "step": 2263
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1772832517385322e-05,
+      "loss": 1.1828,
+      "step": 2264
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1772541546851888e-05,
+      "loss": 1.1333,
+      "step": 2265
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.17722503922346e-05,
+      "loss": 1.099,
+      "step": 2266
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1771959053538952e-05,
+      "loss": 1.1943,
+      "step": 2267
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1771667530770428e-05,
+      "loss": 1.2387,
+      "step": 2268
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.177137582393452e-05,
+      "loss": 1.2312,
+      "step": 2269
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.177108393303672e-05,
+      "loss": 1.0959,
+      "step": 2270
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.177079185808253e-05,
+      "loss": 1.1725,
+      "step": 2271
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.177049959907745e-05,
+      "loss": 1.1155,
+      "step": 2272
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.177020715602699e-05,
+      "loss": 1.1307,
+      "step": 2273
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1769914528936647e-05,
+      "loss": 1.1927,
+      "step": 2274
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1769621717811942e-05,
+      "loss": 1.1627,
+      "step": 2275
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1769328722658392e-05,
+      "loss": 1.1295,
+      "step": 2276
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.176903554348151e-05,
+      "loss": 1.1693,
+      "step": 2277
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.176874218028682e-05,
+      "loss": 1.0893,
+      "step": 2278
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1768448633079853e-05,
+      "loss": 1.2312,
+      "step": 2279
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1768154901866137e-05,
+      "loss": 1.2366,
+      "step": 2280
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.17678609866512e-05,
+      "loss": 1.0988,
+      "step": 2281
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.176756688744058e-05,
+      "loss": 1.1677,
+      "step": 2282
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1767272604239824e-05,
+      "loss": 1.1299,
+      "step": 2283
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1766978137054468e-05,
+      "loss": 1.1152,
+      "step": 2284
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1766683485890062e-05,
+      "loss": 1.1934,
+      "step": 2285
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1766388650752153e-05,
+      "loss": 1.1151,
+      "step": 2286
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1766093631646295e-05,
+      "loss": 1.1358,
+      "step": 2287
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1765798428578046e-05,
+      "loss": 1.1365,
+      "step": 2288
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.176550304155297e-05,
+      "loss": 1.1087,
+      "step": 2289
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1765207470576634e-05,
+      "loss": 1.1985,
+      "step": 2290
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1764911715654596e-05,
+      "loss": 1.2038,
+      "step": 2291
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.176461577679243e-05,
+      "loss": 1.1375,
+      "step": 2292
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1764319653995714e-05,
+      "loss": 1.0288,
+      "step": 2293
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1764023347270027e-05,
+      "loss": 1.1659,
+      "step": 2294
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.176372685662094e-05,
+      "loss": 1.1949,
+      "step": 2295
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1763430182054053e-05,
+      "loss": 1.081,
+      "step": 2296
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1763133323574943e-05,
+      "loss": 1.243,
+      "step": 2297
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.176283628118921e-05,
+      "loss": 1.2495,
+      "step": 2298
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1762539054902443e-05,
+      "loss": 1.1668,
+      "step": 2299
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.176224164472024e-05,
+      "loss": 1.1045,
+      "step": 2300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.176194405064821e-05,
+      "loss": 1.1179,
+      "step": 2301
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1761646272691955e-05,
+      "loss": 1.1823,
+      "step": 2302
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1761348310857083e-05,
+      "loss": 1.1805,
+      "step": 2303
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.176105016514921e-05,
+      "loss": 1.1337,
+      "step": 2304
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1760751835573946e-05,
+      "loss": 1.189,
+      "step": 2305
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1760453322136917e-05,
+      "loss": 1.149,
+      "step": 2306
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1760154624843743e-05,
+      "loss": 1.0863,
+      "step": 2307
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1759855743700052e-05,
+      "loss": 1.1777,
+      "step": 2308
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1759556678711472e-05,
+      "loss": 1.08,
+      "step": 2309
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.175925742988364e-05,
+      "loss": 1.1559,
+      "step": 2310
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.175895799722219e-05,
+      "loss": 1.1074,
+      "step": 2311
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.175865838073276e-05,
+      "loss": 1.1779,
+      "step": 2312
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1758358580421e-05,
+      "loss": 1.2478,
+      "step": 2313
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.175805859629255e-05,
+      "loss": 1.1316,
+      "step": 2314
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1757758428353068e-05,
+      "loss": 1.1553,
+      "step": 2315
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1757458076608206e-05,
+      "loss": 1.2408,
+      "step": 2316
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1757157541063618e-05,
+      "loss": 1.1937,
+      "step": 2317
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.175685682172497e-05,
+      "loss": 1.128,
+      "step": 2318
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1756555918597922e-05,
+      "loss": 1.1878,
+      "step": 2319
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1756254831688144e-05,
+      "loss": 1.2345,
+      "step": 2320
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1755953561001313e-05,
+      "loss": 1.1311,
+      "step": 2321
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1755652106543095e-05,
+      "loss": 1.2238,
+      "step": 2322
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1755350468319172e-05,
+      "loss": 1.1628,
+      "step": 2323
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1755048646335228e-05,
+      "loss": 1.1411,
+      "step": 2324
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1754746640596947e-05,
+      "loss": 1.173,
+      "step": 2325
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1754444451110015e-05,
+      "loss": 1.0896,
+      "step": 2326
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.175414207788013e-05,
+      "loss": 1.0742,
+      "step": 2327
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1753839520912986e-05,
+      "loss": 1.1051,
+      "step": 2328
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1753536780214275e-05,
+      "loss": 1.1653,
+      "step": 2329
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1753233855789707e-05,
+      "loss": 1.0675,
+      "step": 2330
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.175293074764499e-05,
+      "loss": 1.1024,
+      "step": 2331
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1752627455785826e-05,
+      "loss": 1.2508,
+      "step": 2332
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1752323980217936e-05,
+      "loss": 1.286,
+      "step": 2333
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.175202032094703e-05,
+      "loss": 1.1419,
+      "step": 2334
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1751716477978833e-05,
+      "loss": 1.1791,
+      "step": 2335
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.175141245131907e-05,
+      "loss": 1.1368,
+      "step": 2336
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1751108240973458e-05,
+      "loss": 1.1061,
+      "step": 2337
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1750803846947732e-05,
+      "loss": 1.2206,
+      "step": 2338
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1750499269247632e-05,
+      "loss": 1.2358,
+      "step": 2339
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1750194507878893e-05,
+      "loss": 1.2423,
+      "step": 2340
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1749889562847247e-05,
+      "loss": 1.2141,
+      "step": 2341
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.174958443415845e-05,
+      "loss": 1.2244,
+      "step": 2342
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1749279121818237e-05,
+      "loss": 1.0505,
+      "step": 2343
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1748973625832374e-05,
+      "loss": 1.1954,
+      "step": 2344
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.17486679462066e-05,
+      "loss": 1.1505,
+      "step": 2345
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.174836208294669e-05,
+      "loss": 1.2108,
+      "step": 2346
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.174805603605839e-05,
+      "loss": 1.1762,
+      "step": 2347
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1747749805547475e-05,
+      "loss": 1.1985,
+      "step": 2348
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1747443391419707e-05,
+      "loss": 1.1799,
+      "step": 2349
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1747136793680865e-05,
+      "loss": 1.1746,
+      "step": 2350
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1746830012336715e-05,
+      "loss": 1.1919,
+      "step": 2351
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1746523047393047e-05,
+      "loss": 1.1022,
+      "step": 2352
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1746215898855632e-05,
+      "loss": 1.2373,
+      "step": 2353
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1745908566730263e-05,
+      "loss": 1.1583,
+      "step": 2354
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1745601051022724e-05,
+      "loss": 1.192,
+      "step": 2355
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1745293351738814e-05,
+      "loss": 1.2372,
+      "step": 2356
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1744985468884322e-05,
+      "loss": 1.1923,
+      "step": 2357
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1744677402465054e-05,
+      "loss": 1.1407,
+      "step": 2358
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.174436915248681e-05,
+      "loss": 1.2309,
+      "step": 2359
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1744060718955397e-05,
+      "loss": 1.1604,
+      "step": 2360
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1743752101876626e-05,
+      "loss": 1.1802,
+      "step": 2361
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1743443301256304e-05,
+      "loss": 1.193,
+      "step": 2362
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1743134317100257e-05,
+      "loss": 1.17,
+      "step": 2363
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.17428251494143e-05,
+      "loss": 1.1213,
+      "step": 2364
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.174251579820426e-05,
+      "loss": 1.1494,
+      "step": 2365
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1742206263475957e-05,
+      "loss": 1.1551,
+      "step": 2366
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1741896545235232e-05,
+      "loss": 1.1837,
+      "step": 2367
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.174158664348791e-05,
+      "loss": 1.1401,
+      "step": 2368
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1741276558239834e-05,
+      "loss": 1.1683,
+      "step": 2369
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1740966289496845e-05,
+      "loss": 1.1334,
+      "step": 2370
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1740655837264784e-05,
+      "loss": 1.2487,
+      "step": 2371
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.17403452015495e-05,
+      "loss": 1.1969,
+      "step": 2372
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1740034382356847e-05,
+      "loss": 1.1783,
+      "step": 2373
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1739723379692677e-05,
+      "loss": 1.1787,
+      "step": 2374
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1739412193562848e-05,
+      "loss": 1.098,
+      "step": 2375
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1739100823973227e-05,
+      "loss": 1.1449,
+      "step": 2376
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.173878927092967e-05,
+      "loss": 1.0626,
+      "step": 2377
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1738477534438056e-05,
+      "loss": 1.1951,
+      "step": 2378
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.173816561450425e-05,
+      "loss": 1.1263,
+      "step": 2379
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.173785351113413e-05,
+      "loss": 1.1251,
+      "step": 2380
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1737541224333575e-05,
+      "loss": 1.1737,
+      "step": 2381
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.173722875410847e-05,
+      "loss": 1.0297,
+      "step": 2382
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1736916100464693e-05,
+      "loss": 1.1421,
+      "step": 2383
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.173660326340814e-05,
+      "loss": 1.1957,
+      "step": 2384
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1736290242944706e-05,
+      "loss": 1.1377,
+      "step": 2385
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.173597703908028e-05,
+      "loss": 1.0891,
+      "step": 2386
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.173566365182077e-05,
+      "loss": 1.1486,
+      "step": 2387
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.173535008117207e-05,
+      "loss": 1.1839,
+      "step": 2388
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1735036327140095e-05,
+      "loss": 1.1192,
+      "step": 2389
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.173472238973075e-05,
+      "loss": 1.2496,
+      "step": 2390
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.173440826894995e-05,
+      "loss": 1.0876,
+      "step": 2391
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1734093964803612e-05,
+      "loss": 1.1728,
+      "step": 2392
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1733779477297656e-05,
+      "loss": 1.1075,
+      "step": 2393
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.173346480643801e-05,
+      "loss": 1.1775,
+      "step": 2394
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1733149952230595e-05,
+      "loss": 1.1566,
+      "step": 2395
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1732834914681348e-05,
+      "loss": 1.2145,
+      "step": 2396
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1732519693796197e-05,
+      "loss": 1.1629,
+      "step": 2397
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.173220428958109e-05,
+      "loss": 1.1299,
+      "step": 2398
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1731888702041955e-05,
+      "loss": 1.0794,
+      "step": 2399
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1731572931184743e-05,
+      "loss": 1.0319,
+      "step": 2400
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1731256977015404e-05,
+      "loss": 1.1501,
+      "step": 2401
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1730940839539885e-05,
+      "loss": 1.1822,
+      "step": 2402
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.173062451876415e-05,
+      "loss": 1.1399,
+      "step": 2403
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1730308014694144e-05,
+      "loss": 1.1463,
+      "step": 2404
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.172999132733584e-05,
+      "loss": 1.0815,
+      "step": 2405
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.17296744566952e-05,
+      "loss": 1.1379,
+      "step": 2406
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.172935740277819e-05,
+      "loss": 1.2162,
+      "step": 2407
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1729040165590785e-05,
+      "loss": 1.1496,
+      "step": 2408
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.172872274513896e-05,
+      "loss": 1.1528,
+      "step": 2409
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.17284051414287e-05,
+      "loss": 1.14,
+      "step": 2410
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.172808735446598e-05,
+      "loss": 1.1703,
+      "step": 2411
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1727769384256785e-05,
+      "loss": 1.0949,
+      "step": 2412
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1727451230807107e-05,
+      "loss": 1.1491,
+      "step": 2413
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1727132894122943e-05,
+      "loss": 1.18,
+      "step": 2414
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1726814374210288e-05,
+      "loss": 1.1001,
+      "step": 2415
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1726495671075136e-05,
+      "loss": 1.2347,
+      "step": 2416
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.17261767847235e-05,
+      "loss": 1.1636,
+      "step": 2417
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1725857715161377e-05,
+      "loss": 1.134,
+      "step": 2418
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1725538462394782e-05,
+      "loss": 1.1988,
+      "step": 2419
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.172521902642973e-05,
+      "loss": 1.2329,
+      "step": 2420
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1724899407272238e-05,
+      "loss": 1.0887,
+      "step": 2421
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.172457960492832e-05,
+      "loss": 1.2616,
+      "step": 2422
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1724259619404013e-05,
+      "loss": 1.1671,
+      "step": 2423
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.172393945070533e-05,
+      "loss": 1.1175,
+      "step": 2424
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.172361909883831e-05,
+      "loss": 1.0761,
+      "step": 2425
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1723298563808993e-05,
+      "loss": 1.2282,
+      "step": 2426
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1722977845623402e-05,
+      "loss": 1.0837,
+      "step": 2427
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.172265694428759e-05,
+      "loss": 1.2459,
+      "step": 2428
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.17223358598076e-05,
+      "loss": 1.0765,
+      "step": 2429
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1722014592189473e-05,
+      "loss": 1.0455,
+      "step": 2430
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.172169314143927e-05,
+      "loss": 1.1784,
+      "step": 2431
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1721371507563042e-05,
+      "loss": 1.1752,
+      "step": 2432
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1721049690566844e-05,
+      "loss": 1.0898,
+      "step": 2433
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1720727690456746e-05,
+      "loss": 1.1484,
+      "step": 2434
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.172040550723881e-05,
+      "loss": 1.0934,
+      "step": 2435
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.17200831409191e-05,
+      "loss": 1.1498,
+      "step": 2436
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1719760591503696e-05,
+      "loss": 1.2284,
+      "step": 2437
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.171943785899867e-05,
+      "loss": 1.1058,
+      "step": 2438
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1719114943410096e-05,
+      "loss": 1.1926,
+      "step": 2439
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.171879184474407e-05,
+      "loss": 1.1587,
+      "step": 2440
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1718468563006667e-05,
+      "loss": 1.1942,
+      "step": 2441
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1718145098203978e-05,
+      "loss": 1.2085,
+      "step": 2442
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1717821450342103e-05,
+      "loss": 1.1544,
+      "step": 2443
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1717497619427132e-05,
+      "loss": 1.1196,
+      "step": 2444
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1717173605465163e-05,
+      "loss": 1.1293,
+      "step": 2445
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.171684940846231e-05,
+      "loss": 1.2322,
+      "step": 2446
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1716525028424667e-05,
+      "loss": 1.1575,
+      "step": 2447
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1716200465358353e-05,
+      "loss": 1.2231,
+      "step": 2448
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1715875719269474e-05,
+      "loss": 1.1239,
+      "step": 2449
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1715550790164157e-05,
+      "loss": 1.1316,
+      "step": 2450
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1715225678048516e-05,
+      "loss": 1.0955,
+      "step": 2451
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1714900382928675e-05,
+      "loss": 1.1646,
+      "step": 2452
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1714574904810768e-05,
+      "loss": 1.1187,
+      "step": 2453
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1714249243700917e-05,
+      "loss": 1.2268,
+      "step": 2454
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1713923399605264e-05,
+      "loss": 1.1895,
+      "step": 2455
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1713597372529942e-05,
+      "loss": 1.2135,
+      "step": 2456
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1713271162481094e-05,
+      "loss": 1.2089,
+      "step": 2457
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1712944769464865e-05,
+      "loss": 1.1542,
+      "step": 2458
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1712618193487403e-05,
+      "loss": 1.0774,
+      "step": 2459
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.171229143455486e-05,
+      "loss": 1.1978,
+      "step": 2460
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.171196449267339e-05,
+      "loss": 1.127,
+      "step": 2461
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1711637367849152e-05,
+      "loss": 1.0557,
+      "step": 2462
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1711310060088308e-05,
+      "loss": 1.1098,
+      "step": 2463
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.171098256939702e-05,
+      "loss": 1.1147,
+      "step": 2464
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1710654895781468e-05,
+      "loss": 1.0984,
+      "step": 2465
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1710327039247815e-05,
+      "loss": 1.1321,
+      "step": 2466
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1709998999802236e-05,
+      "loss": 1.1249,
+      "step": 2467
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1709670777450913e-05,
+      "loss": 1.1301,
+      "step": 2468
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1709342372200033e-05,
+      "loss": 1.1211,
+      "step": 2469
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1709013784055774e-05,
+      "loss": 1.1693,
+      "step": 2470
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1708685013024328e-05,
+      "loss": 1.1463,
+      "step": 2471
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1708356059111892e-05,
+      "loss": 1.2479,
+      "step": 2472
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1708026922324656e-05,
+      "loss": 1.1496,
+      "step": 2473
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.170769760266883e-05,
+      "loss": 1.1128,
+      "step": 2474
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1707368100150608e-05,
+      "loss": 1.1458,
+      "step": 2475
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1707038414776196e-05,
+      "loss": 1.1438,
+      "step": 2476
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1706708546551813e-05,
+      "loss": 1.1164,
+      "step": 2477
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1706378495483665e-05,
+      "loss": 1.2115,
+      "step": 2478
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1706048261577974e-05,
+      "loss": 1.1786,
+      "step": 2479
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1705717844840956e-05,
+      "loss": 1.1951,
+      "step": 2480
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.170538724527884e-05,
+      "loss": 1.1876,
+      "step": 2481
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1705056462897847e-05,
+      "loss": 1.2017,
+      "step": 2482
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1704725497704214e-05,
+      "loss": 1.1369,
+      "step": 2483
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1704394349704175e-05,
+      "loss": 1.1545,
+      "step": 2484
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1704063018903963e-05,
+      "loss": 1.1688,
+      "step": 2485
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1703731505309822e-05,
+      "loss": 1.1734,
+      "step": 2486
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1703399808927998e-05,
+      "loss": 1.146,
+      "step": 2487
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1703067929764738e-05,
+      "loss": 1.0849,
+      "step": 2488
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1702735867826292e-05,
+      "loss": 1.1691,
+      "step": 2489
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.170240362311892e-05,
+      "loss": 1.1506,
+      "step": 2490
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.170207119564887e-05,
+      "loss": 1.0801,
+      "step": 2491
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1701738585422417e-05,
+      "loss": 1.1649,
+      "step": 2492
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1701405792445816e-05,
+      "loss": 1.1216,
+      "step": 2493
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1701072816725344e-05,
+      "loss": 1.0754,
+      "step": 2494
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1700739658267268e-05,
+      "loss": 1.1397,
+      "step": 2495
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1700406317077862e-05,
+      "loss": 1.1449,
+      "step": 2496
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.170007279316341e-05,
+      "loss": 1.0848,
+      "step": 2497
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.169973908653019e-05,
+      "loss": 1.1983,
+      "step": 2498
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.169940519718449e-05,
+      "loss": 1.1835,
+      "step": 2499
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1699071125132605e-05,
+      "loss": 1.0774,
+      "step": 2500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.169873687038082e-05,
+      "loss": 1.2221,
+      "step": 2501
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1698402432935434e-05,
+      "loss": 1.1316,
+      "step": 2502
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1698067812802745e-05,
+      "loss": 1.1187,
+      "step": 2503
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1697733009989057e-05,
+      "loss": 1.1736,
+      "step": 2504
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.169739802450068e-05,
+      "loss": 1.1309,
+      "step": 2505
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1697062856343922e-05,
+      "loss": 1.1932,
+      "step": 2506
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1696727505525095e-05,
+      "loss": 1.0959,
+      "step": 2507
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1696391972050517e-05,
+      "loss": 1.1626,
+      "step": 2508
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1696056255926506e-05,
+      "loss": 1.1476,
+      "step": 2509
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1695720357159392e-05,
+      "loss": 1.2062,
+      "step": 2510
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1695384275755495e-05,
+      "loss": 1.1794,
+      "step": 2511
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1695048011721153e-05,
+      "loss": 1.0762,
+      "step": 2512
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1694711565062697e-05,
+      "loss": 1.1088,
+      "step": 2513
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1694374935786458e-05,
+      "loss": 1.0749,
+      "step": 2514
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1694038123898785e-05,
+      "loss": 1.2629,
+      "step": 2515
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1693701129406022e-05,
+      "loss": 1.1648,
+      "step": 2516
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1693363952314515e-05,
+      "loss": 1.1733,
+      "step": 2517
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1693026592630615e-05,
+      "loss": 1.1498,
+      "step": 2518
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1692689050360682e-05,
+      "loss": 1.1524,
+      "step": 2519
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1692351325511067e-05,
+      "loss": 1.1252,
+      "step": 2520
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.169201341808813e-05,
+      "loss": 1.1681,
+      "step": 2521
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1691675328098245e-05,
+      "loss": 1.131,
+      "step": 2522
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.169133705554778e-05,
+      "loss": 1.1276,
+      "step": 2523
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1690998600443097e-05,
+      "loss": 1.1956,
+      "step": 2524
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1690659962790577e-05,
+      "loss": 1.1797,
+      "step": 2525
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1690321142596603e-05,
+      "loss": 1.162,
+      "step": 2526
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1689982139867547e-05,
+      "loss": 1.1493,
+      "step": 2527
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.168964295460981e-05,
+      "loss": 1.1898,
+      "step": 2528
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1689303586829766e-05,
+      "loss": 1.1545,
+      "step": 2529
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.168896403653382e-05,
+      "loss": 1.1282,
+      "step": 2530
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.168862430372836e-05,
+      "loss": 1.0945,
+      "step": 2531
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1688284388419785e-05,
+      "loss": 1.1864,
+      "step": 2532
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1687944290614504e-05,
+      "loss": 1.1512,
+      "step": 2533
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1687604010318918e-05,
+      "loss": 1.1786,
+      "step": 2534
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.168726354753944e-05,
+      "loss": 1.2449,
+      "step": 2535
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.168692290228248e-05,
+      "loss": 1.1794,
+      "step": 2536
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.168658207455446e-05,
+      "loss": 1.1266,
+      "step": 2537
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1686241064361793e-05,
+      "loss": 1.0517,
+      "step": 2538
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.168589987171091e-05,
+      "loss": 1.2564,
+      "step": 2539
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1685558496608237e-05,
+      "loss": 1.2014,
+      "step": 2540
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1685216939060194e-05,
+      "loss": 1.0332,
+      "step": 2541
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.168487519907323e-05,
+      "loss": 1.1232,
+      "step": 2542
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.168453327665377e-05,
+      "loss": 1.0817,
+      "step": 2543
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1684191171808264e-05,
+      "loss": 1.1633,
+      "step": 2544
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.168384888454315e-05,
+      "loss": 1.2382,
+      "step": 2545
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.168350641486488e-05,
+      "loss": 1.0874,
+      "step": 2546
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1683163762779896e-05,
+      "loss": 1.2123,
+      "step": 2547
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1682820928294666e-05,
+      "loss": 1.2239,
+      "step": 2548
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1682477911415635e-05,
+      "loss": 1.1532,
+      "step": 2549
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1682134712149273e-05,
+      "loss": 1.1539,
+      "step": 2550
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1681791330502046e-05,
+      "loss": 1.2487,
+      "step": 2551
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1681447766480413e-05,
+      "loss": 1.2781,
+      "step": 2552
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1681104020090855e-05,
+      "loss": 1.2147,
+      "step": 2553
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.168076009133984e-05,
+      "loss": 1.2403,
+      "step": 2554
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1680415980233853e-05,
+      "loss": 1.5613,
+      "step": 2555
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.168007168677937e-05,
+      "loss": 1.668,
+      "step": 2556
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1679727210982882e-05,
+      "loss": 1.5763,
+      "step": 2557
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1679382552850868e-05,
+      "loss": 1.5246,
+      "step": 2558
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1679037712389835e-05,
+      "loss": 1.59,
+      "step": 2559
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1678692689606268e-05,
+      "loss": 1.5115,
+      "step": 2560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.167834748450667e-05,
+      "loss": 1.5842,
+      "step": 2561
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.167800209709754e-05,
+      "loss": 1.591,
+      "step": 2562
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1677656527385387e-05,
+      "loss": 1.6643,
+      "step": 2563
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1677310775376722e-05,
+      "loss": 1.6554,
+      "step": 2564
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1676964841078052e-05,
+      "loss": 1.5471,
+      "step": 2565
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1676618724495904e-05,
+      "loss": 1.4502,
+      "step": 2566
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1676272425636785e-05,
+      "loss": 1.4945,
+      "step": 2567
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1675925944507227e-05,
+      "loss": 1.4606,
+      "step": 2568
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1675579281113754e-05,
+      "loss": 1.3779,
+      "step": 2569
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1675232435462894e-05,
+      "loss": 1.452,
+      "step": 2570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1674885407561182e-05,
+      "loss": 1.4783,
+      "step": 2571
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1674538197415158e-05,
+      "loss": 1.3649,
+      "step": 2572
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.167419080503136e-05,
+      "loss": 1.3321,
+      "step": 2573
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.167384323041633e-05,
+      "loss": 1.4035,
+      "step": 2574
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1673495473576614e-05,
+      "loss": 1.4403,
+      "step": 2575
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.167314753451877e-05,
+      "loss": 1.3977,
+      "step": 2576
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1672799413249344e-05,
+      "loss": 1.3672,
+      "step": 2577
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1672451109774893e-05,
+      "loss": 1.3576,
+      "step": 2578
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1672102624101988e-05,
+      "loss": 1.2758,
+      "step": 2579
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1671753956237188e-05,
+      "loss": 1.2076,
+      "step": 2580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1671405106187056e-05,
+      "loss": 1.3107,
+      "step": 2581
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1671056073958166e-05,
+      "loss": 1.3907,
+      "step": 2582
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1670706859557094e-05,
+      "loss": 1.3006,
+      "step": 2583
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.167035746299042e-05,
+      "loss": 1.3034,
+      "step": 2584
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1670007884264724e-05,
+      "loss": 1.3247,
+      "step": 2585
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.166965812338659e-05,
+      "loss": 1.3168,
+      "step": 2586
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1669308180362603e-05,
+      "loss": 1.2725,
+      "step": 2587
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1668958055199366e-05,
+      "loss": 1.2809,
+      "step": 2588
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.166860774790346e-05,
+      "loss": 1.29,
+      "step": 2589
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1668257258481492e-05,
+      "loss": 1.2663,
+      "step": 2590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1667906586940062e-05,
+      "loss": 1.2543,
+      "step": 2591
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.166755573328578e-05,
+      "loss": 1.4481,
+      "step": 2592
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1667204697525248e-05,
+      "loss": 1.36,
+      "step": 2593
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1666853479665084e-05,
+      "loss": 1.3493,
+      "step": 2594
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1666502079711898e-05,
+      "loss": 1.4361,
+      "step": 2595
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1666150497672313e-05,
+      "loss": 1.4059,
+      "step": 2596
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1665798733552955e-05,
+      "loss": 1.3718,
+      "step": 2597
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1665446787360445e-05,
+      "loss": 1.4738,
+      "step": 2598
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1665094659101412e-05,
+      "loss": 1.34,
+      "step": 2599
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1664742348782493e-05,
+      "loss": 1.3719,
+      "step": 2600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1664389856410322e-05,
+      "loss": 1.4808,
+      "step": 2601
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.166403718199154e-05,
+      "loss": 1.4587,
+      "step": 2602
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.166368432553279e-05,
+      "loss": 1.4508,
+      "step": 2603
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1663331287040714e-05,
+      "loss": 1.5255,
+      "step": 2604
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1662978066521972e-05,
+      "loss": 1.4518,
+      "step": 2605
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1662624663983205e-05,
+      "loss": 1.5476,
+      "step": 2606
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1662271079431085e-05,
+      "loss": 1.3337,
+      "step": 2607
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.166191731287226e-05,
+      "loss": 1.4273,
+      "step": 2608
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.16615633643134e-05,
+      "loss": 1.4964,
+      "step": 2609
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1661209233761168e-05,
+      "loss": 1.3352,
+      "step": 2610
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1660854921222238e-05,
+      "loss": 1.3999,
+      "step": 2611
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1660500426703283e-05,
+      "loss": 1.3585,
+      "step": 2612
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1660145750210982e-05,
+      "loss": 1.4734,
+      "step": 2613
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1659790891752013e-05,
+      "loss": 1.4082,
+      "step": 2614
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1659435851333063e-05,
+      "loss": 1.3484,
+      "step": 2615
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.165908062896082e-05,
+      "loss": 1.3743,
+      "step": 2616
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1658725224641975e-05,
+      "loss": 1.3375,
+      "step": 2617
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.165836963838322e-05,
+      "loss": 1.2424,
+      "step": 2618
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.165801387019126e-05,
+      "loss": 1.3945,
+      "step": 2619
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1657657920072785e-05,
+      "loss": 1.3425,
+      "step": 2620
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.165730178803451e-05,
+      "loss": 1.3355,
+      "step": 2621
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.165694547408314e-05,
+      "loss": 1.3746,
+      "step": 2622
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1656588978225387e-05,
+      "loss": 1.364,
+      "step": 2623
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1656232300467966e-05,
+      "loss": 1.2633,
+      "step": 2624
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1655875440817595e-05,
+      "loss": 1.3728,
+      "step": 2625
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1655518399281e-05,
+      "loss": 1.2069,
+      "step": 2626
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.16551611758649e-05,
+      "loss": 1.3166,
+      "step": 2627
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.165480377057603e-05,
+      "loss": 1.3399,
+      "step": 2628
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.165444618342112e-05,
+      "loss": 1.387,
+      "step": 2629
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1654088414406902e-05,
+      "loss": 1.3765,
+      "step": 2630
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.165373046354012e-05,
+      "loss": 1.35,
+      "step": 2631
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.165337233082752e-05,
+      "loss": 1.3523,
+      "step": 2632
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.165301401627584e-05,
+      "loss": 1.2975,
+      "step": 2633
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1652655519891832e-05,
+      "loss": 1.253,
+      "step": 2634
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1652296841682252e-05,
+      "loss": 1.3438,
+      "step": 2635
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1651937981653855e-05,
+      "loss": 1.3536,
+      "step": 2636
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.16515789398134e-05,
+      "loss": 1.2367,
+      "step": 2637
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.165121971616765e-05,
+      "loss": 1.3109,
+      "step": 2638
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1650860310723375e-05,
+      "loss": 1.2734,
+      "step": 2639
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1650500723487336e-05,
+      "loss": 1.341,
+      "step": 2640
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1650140954466318e-05,
+      "loss": 1.287,
+      "step": 2641
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1649781003667092e-05,
+      "loss": 1.3199,
+      "step": 2642
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.164942087109644e-05,
+      "loss": 1.266,
+      "step": 2643
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1649060556761143e-05,
+      "loss": 1.3501,
+      "step": 2644
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1648700060667988e-05,
+      "loss": 1.3651,
+      "step": 2645
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.164833938282377e-05,
+      "loss": 1.2999,
+      "step": 2646
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.164797852323528e-05,
+      "loss": 1.4179,
+      "step": 2647
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1647617481909315e-05,
+      "loss": 1.3325,
+      "step": 2648
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.164725625885268e-05,
+      "loss": 1.3231,
+      "step": 2649
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1646894854072172e-05,
+      "loss": 1.2533,
+      "step": 2650
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1646533267574605e-05,
+      "loss": 1.3049,
+      "step": 2651
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.164617149936679e-05,
+      "loss": 1.333,
+      "step": 2652
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1645809549455534e-05,
+      "loss": 1.3431,
+      "step": 2653
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1645447417847665e-05,
+      "loss": 1.2383,
+      "step": 2654
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.164508510455e-05,
+      "loss": 1.2717,
+      "step": 2655
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1644722609569364e-05,
+      "loss": 1.3626,
+      "step": 2656
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1644359932912585e-05,
+      "loss": 1.3171,
+      "step": 2657
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1643997074586492e-05,
+      "loss": 1.4115,
+      "step": 2658
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1643634034597925e-05,
+      "loss": 1.2587,
+      "step": 2659
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1643270812953723e-05,
+      "loss": 1.2529,
+      "step": 2660
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.164290740966072e-05,
+      "loss": 1.2508,
+      "step": 2661
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.164254382472577e-05,
+      "loss": 1.3492,
+      "step": 2662
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.164218005815572e-05,
+      "loss": 1.2664,
+      "step": 2663
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1641816109957417e-05,
+      "loss": 1.2016,
+      "step": 2664
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1641451980137726e-05,
+      "loss": 1.2077,
+      "step": 2665
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1641087668703496e-05,
+      "loss": 1.2622,
+      "step": 2666
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1640723175661596e-05,
+      "loss": 1.3057,
+      "step": 2667
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1640358501018886e-05,
+      "loss": 1.2295,
+      "step": 2668
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1639993644782242e-05,
+      "loss": 1.2722,
+      "step": 2669
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1639628606958536e-05,
+      "loss": 1.2657,
+      "step": 2670
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.163926338755464e-05,
+      "loss": 1.3001,
+      "step": 2671
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1638897986577436e-05,
+      "loss": 1.3056,
+      "step": 2672
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1638532404033808e-05,
+      "loss": 1.2212,
+      "step": 2673
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.163816663993064e-05,
+      "loss": 1.3285,
+      "step": 2674
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1637800694274822e-05,
+      "loss": 1.3732,
+      "step": 2675
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1637434567073247e-05,
+      "loss": 1.3794,
+      "step": 2676
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1637068258332813e-05,
+      "loss": 1.2166,
+      "step": 2677
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1636701768060422e-05,
+      "loss": 1.35,
+      "step": 2678
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1636335096262978e-05,
+      "loss": 1.3037,
+      "step": 2679
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.163596824294738e-05,
+      "loss": 1.2497,
+      "step": 2680
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1635601208120545e-05,
+      "loss": 1.3282,
+      "step": 2681
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.163523399178939e-05,
+      "loss": 1.2882,
+      "step": 2682
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1634866593960823e-05,
+      "loss": 1.301,
+      "step": 2683
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1634499014641772e-05,
+      "loss": 1.3695,
+      "step": 2684
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1634131253839157e-05,
+      "loss": 1.2659,
+      "step": 2685
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1633763311559906e-05,
+      "loss": 1.2646,
+      "step": 2686
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1633395187810953e-05,
+      "loss": 1.3102,
+      "step": 2687
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.163302688259923e-05,
+      "loss": 1.2212,
+      "step": 2688
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1632658395931672e-05,
+      "loss": 1.2449,
+      "step": 2689
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1632289727815228e-05,
+      "loss": 1.2222,
+      "step": 2690
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.163192087825683e-05,
+      "loss": 1.219,
+      "step": 2691
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.163155184726344e-05,
+      "loss": 1.3123,
+      "step": 2692
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1631182634842e-05,
+      "loss": 1.2596,
+      "step": 2693
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.163081324099947e-05,
+      "loss": 1.2975,
+      "step": 2694
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.16304436657428e-05,
+      "loss": 1.3607,
+      "step": 2695
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1630073909078962e-05,
+      "loss": 1.2328,
+      "step": 2696
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1629703971014914e-05,
+      "loss": 1.3083,
+      "step": 2697
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.162933385155763e-05,
+      "loss": 1.3177,
+      "step": 2698
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1628963550714078e-05,
+      "loss": 1.2925,
+      "step": 2699
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1628593068491232e-05,
+      "loss": 1.2686,
+      "step": 2700
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1628222404896075e-05,
+      "loss": 1.2782,
+      "step": 2701
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1627851559935585e-05,
+      "loss": 1.2242,
+      "step": 2702
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.162748053361675e-05,
+      "loss": 1.3565,
+      "step": 2703
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.162710932594656e-05,
+      "loss": 1.2523,
+      "step": 2704
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1626737936932005e-05,
+      "loss": 1.2855,
+      "step": 2705
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.162636636658008e-05,
+      "loss": 1.2799,
+      "step": 2706
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1625994614897787e-05,
+      "loss": 1.1899,
+      "step": 2707
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.162562268189213e-05,
+      "loss": 1.1727,
+      "step": 2708
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.162525056757011e-05,
+      "loss": 1.2435,
+      "step": 2709
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.162487827193874e-05,
+      "loss": 1.2428,
+      "step": 2710
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.162450579500503e-05,
+      "loss": 1.2825,
+      "step": 2711
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1624133136776e-05,
+      "loss": 1.2306,
+      "step": 2712
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.162376029725867e-05,
+      "loss": 1.2718,
+      "step": 2713
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1623387276460053e-05,
+      "loss": 1.216,
+      "step": 2714
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.162301407438719e-05,
+      "loss": 1.1699,
+      "step": 2715
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1622640691047105e-05,
+      "loss": 1.3284,
+      "step": 2716
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1622267126446828e-05,
+      "loss": 1.253,
+      "step": 2717
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.16218933805934e-05,
+      "loss": 1.2239,
+      "step": 2718
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1621519453493856e-05,
+      "loss": 1.1999,
+      "step": 2719
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.162114534515525e-05,
+      "loss": 1.1937,
+      "step": 2720
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1620771055584618e-05,
+      "loss": 1.2114,
+      "step": 2721
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.162039658478902e-05,
+      "loss": 1.1906,
+      "step": 2722
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1620021932775498e-05,
+      "loss": 1.1989,
+      "step": 2723
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.161964709955112e-05,
+      "loss": 1.2429,
+      "step": 2724
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.161927208512294e-05,
+      "loss": 1.2211,
+      "step": 2725
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1618896889498026e-05,
+      "loss": 1.2106,
+      "step": 2726
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1618521512683443e-05,
+      "loss": 1.2031,
+      "step": 2727
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1618145954686266e-05,
+      "loss": 1.2179,
+      "step": 2728
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1617770215513564e-05,
+      "loss": 1.261,
+      "step": 2729
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1617394295172416e-05,
+      "loss": 1.2494,
+      "step": 2730
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.161701819366991e-05,
+      "loss": 1.203,
+      "step": 2731
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.161664191101312e-05,
+      "loss": 1.1984,
+      "step": 2732
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.161626544720914e-05,
+      "loss": 1.2422,
+      "step": 2733
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1615888802265062e-05,
+      "loss": 1.2417,
+      "step": 2734
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1615511976187975e-05,
+      "loss": 1.1969,
+      "step": 2735
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1615134968984986e-05,
+      "loss": 1.2335,
+      "step": 2736
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1614757780663187e-05,
+      "loss": 1.1516,
+      "step": 2737
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1614380411229694e-05,
+      "loss": 1.2885,
+      "step": 2738
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1614002860691603e-05,
+      "loss": 1.2636,
+      "step": 2739
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1613625129056038e-05,
+      "loss": 1.2598,
+      "step": 2740
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1613247216330108e-05,
+      "loss": 1.1657,
+      "step": 2741
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.161286912252093e-05,
+      "loss": 1.1882,
+      "step": 2742
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1612490847635628e-05,
+      "loss": 1.2442,
+      "step": 2743
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.161211239168133e-05,
+      "loss": 1.206,
+      "step": 2744
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.161173375466516e-05,
+      "loss": 1.1586,
+      "step": 2745
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1611354936594256e-05,
+      "loss": 1.1772,
+      "step": 2746
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.161097593747575e-05,
+      "loss": 1.2612,
+      "step": 2747
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.161059675731678e-05,
+      "loss": 1.2817,
+      "step": 2748
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.161021739612449e-05,
+      "loss": 1.2031,
+      "step": 2749
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.160983785390603e-05,
+      "loss": 1.2365,
+      "step": 2750
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1609458130668543e-05,
+      "loss": 1.1987,
+      "step": 2751
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1609078226419187e-05,
+      "loss": 1.1344,
+      "step": 2752
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1608698141165115e-05,
+      "loss": 1.2133,
+      "step": 2753
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1608317874913485e-05,
+      "loss": 1.2233,
+      "step": 2754
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1607937427671466e-05,
+      "loss": 1.1847,
+      "step": 2755
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.160755679944622e-05,
+      "loss": 1.2882,
+      "step": 2756
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1607175990244918e-05,
+      "loss": 1.413,
+      "step": 2757
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1606795000074733e-05,
+      "loss": 1.2205,
+      "step": 2758
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.160641382894284e-05,
+      "loss": 1.2169,
+      "step": 2759
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1606032476856422e-05,
+      "loss": 1.2403,
+      "step": 2760
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1605650943822662e-05,
+      "loss": 1.1796,
+      "step": 2761
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1605269229848743e-05,
+      "loss": 1.2409,
+      "step": 2762
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1604887334941864e-05,
+      "loss": 1.2261,
+      "step": 2763
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1604505259109208e-05,
+      "loss": 1.3155,
+      "step": 2764
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1604123002357975e-05,
+      "loss": 1.1159,
+      "step": 2765
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1603740564695372e-05,
+      "loss": 1.1889,
+      "step": 2766
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.16033579461286e-05,
+      "loss": 1.0708,
+      "step": 2767
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1602975146664864e-05,
+      "loss": 1.2464,
+      "step": 2768
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1602592166311372e-05,
+      "loss": 1.1767,
+      "step": 2769
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1602209005075346e-05,
+      "loss": 1.2258,
+      "step": 2770
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1601825662963996e-05,
+      "loss": 1.1509,
+      "step": 2771
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.160144213998455e-05,
+      "loss": 1.2215,
+      "step": 2772
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1601058436144224e-05,
+      "loss": 1.2749,
+      "step": 2773
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1600674551450253e-05,
+      "loss": 1.1327,
+      "step": 2774
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1600290485909866e-05,
+      "loss": 1.2364,
+      "step": 2775
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1599906239530295e-05,
+      "loss": 1.167,
+      "step": 2776
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1599521812318782e-05,
+      "loss": 1.2333,
+      "step": 2777
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1599137204282567e-05,
+      "loss": 1.1206,
+      "step": 2778
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1598752415428894e-05,
+      "loss": 1.2495,
+      "step": 2779
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.159836744576501e-05,
+      "loss": 1.2034,
+      "step": 2780
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.159798229529817e-05,
+      "loss": 1.3019,
+      "step": 2781
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1597596964035628e-05,
+      "loss": 1.2053,
+      "step": 2782
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.159721145198464e-05,
+      "loss": 1.2611,
+      "step": 2783
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1596825759152467e-05,
+      "loss": 1.1777,
+      "step": 2784
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.159643988554638e-05,
+      "loss": 1.2467,
+      "step": 2785
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1596053831173644e-05,
+      "loss": 1.2527,
+      "step": 2786
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.159566759604153e-05,
+      "loss": 1.1905,
+      "step": 2787
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1595281180157316e-05,
+      "loss": 1.2382,
+      "step": 2788
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.159489458352828e-05,
+      "loss": 1.1786,
+      "step": 2789
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1594507806161704e-05,
+      "loss": 1.1841,
+      "step": 2790
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1594120848064872e-05,
+      "loss": 1.2506,
+      "step": 2791
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1593733709245078e-05,
+      "loss": 1.3385,
+      "step": 2792
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1593346389709604e-05,
+      "loss": 1.1882,
+      "step": 2793
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.159295888946576e-05,
+      "loss": 1.3056,
+      "step": 2794
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1592571208520837e-05,
+      "loss": 1.198,
+      "step": 2795
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1592183346882137e-05,
+      "loss": 1.2668,
+      "step": 2796
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.159179530455697e-05,
+      "loss": 1.2714,
+      "step": 2797
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.159140708155264e-05,
+      "loss": 1.2914,
+      "step": 2798
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.159101867787647e-05,
+      "loss": 1.3562,
+      "step": 2799
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1590630093535763e-05,
+      "loss": 1.2709,
+      "step": 2800
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.159024132853785e-05,
+      "loss": 1.3349,
+      "step": 2801
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.158985238289005e-05,
+      "loss": 1.3336,
+      "step": 2802
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1589463256599686e-05,
+      "loss": 1.3472,
+      "step": 2803
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.158907394967409e-05,
+      "loss": 1.34,
+      "step": 2804
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.15886844621206e-05,
+      "loss": 1.3433,
+      "step": 2805
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1588294793946548e-05,
+      "loss": 1.2884,
+      "step": 2806
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1587904945159276e-05,
+      "loss": 1.3709,
+      "step": 2807
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1587514915766125e-05,
+      "loss": 1.3043,
+      "step": 2808
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1587124705774443e-05,
+      "loss": 1.2637,
+      "step": 2809
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1586734315191582e-05,
+      "loss": 1.2423,
+      "step": 2810
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1586343744024896e-05,
+      "loss": 1.2373,
+      "step": 2811
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.158595299228174e-05,
+      "loss": 1.2706,
+      "step": 2812
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1585562059969477e-05,
+      "loss": 1.3198,
+      "step": 2813
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1585170947095463e-05,
+      "loss": 1.2303,
+      "step": 2814
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1584779653667075e-05,
+      "loss": 1.256,
+      "step": 2815
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1584388179691684e-05,
+      "loss": 1.2604,
+      "step": 2816
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1583996525176654e-05,
+      "loss": 1.2301,
+      "step": 2817
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1583604690129376e-05,
+      "loss": 1.266,
+      "step": 2818
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.158321267455722e-05,
+      "loss": 1.1473,
+      "step": 2819
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.158282047846757e-05,
+      "loss": 1.2923,
+      "step": 2820
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1582428101867826e-05,
+      "loss": 1.2062,
+      "step": 2821
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1582035544765368e-05,
+      "loss": 1.2651,
+      "step": 2822
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1581642807167594e-05,
+      "loss": 1.2983,
+      "step": 2823
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.15812498890819e-05,
+      "loss": 1.3101,
+      "step": 2824
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.158085679051569e-05,
+      "loss": 1.161,
+      "step": 2825
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1580463511476366e-05,
+      "loss": 1.2102,
+      "step": 2826
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.158007005197134e-05,
+      "loss": 1.2578,
+      "step": 2827
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.157967641200802e-05,
+      "loss": 1.2204,
+      "step": 2828
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.157928259159382e-05,
+      "loss": 1.268,
+      "step": 2829
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1578888590736164e-05,
+      "loss": 1.2648,
+      "step": 2830
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1578494409442473e-05,
+      "loss": 1.2504,
+      "step": 2831
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1578100047720165e-05,
+      "loss": 1.2324,
+      "step": 2832
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1577705505576675e-05,
+      "loss": 1.3007,
+      "step": 2833
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1577310783019433e-05,
+      "loss": 1.2378,
+      "step": 2834
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1576915880055873e-05,
+      "loss": 1.2707,
+      "step": 2835
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1576520796693434e-05,
+      "loss": 1.2059,
+      "step": 2836
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.157612553293956e-05,
+      "loss": 1.3014,
+      "step": 2837
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1575730088801697e-05,
+      "loss": 1.2772,
+      "step": 2838
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1575334464287293e-05,
+      "loss": 1.2667,
+      "step": 2839
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.15749386594038e-05,
+      "loss": 1.2353,
+      "step": 2840
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1574542674158672e-05,
+      "loss": 1.2428,
+      "step": 2841
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.157414650855937e-05,
+      "loss": 1.1947,
+      "step": 2842
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1573750162613355e-05,
+      "loss": 1.3315,
+      "step": 2843
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1573353636328095e-05,
+      "loss": 1.2836,
+      "step": 2844
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.157295692971106e-05,
+      "loss": 1.3667,
+      "step": 2845
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.157256004276972e-05,
+      "loss": 1.2684,
+      "step": 2846
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.157216297551155e-05,
+      "loss": 1.2645,
+      "step": 2847
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1571765727944035e-05,
+      "loss": 1.3174,
+      "step": 2848
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1571368300074653e-05,
+      "loss": 1.2149,
+      "step": 2849
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1570970691910893e-05,
+      "loss": 1.3506,
+      "step": 2850
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1570572903460242e-05,
+      "loss": 1.3108,
+      "step": 2851
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1570174934730196e-05,
+      "loss": 1.3273,
+      "step": 2852
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.156977678572825e-05,
+      "loss": 1.3228,
+      "step": 2853
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1569378456461902e-05,
+      "loss": 1.3953,
+      "step": 2854
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.156897994693866e-05,
+      "loss": 1.3483,
+      "step": 2855
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1568581257166027e-05,
+      "loss": 1.3684,
+      "step": 2856
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1568182387151515e-05,
+      "loss": 1.4186,
+      "step": 2857
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1567783336902632e-05,
+      "loss": 1.2538,
+      "step": 2858
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1567384106426904e-05,
+      "loss": 1.3002,
+      "step": 2859
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1566984695731845e-05,
+      "loss": 1.42,
+      "step": 2860
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1566585104824983e-05,
+      "loss": 1.2549,
+      "step": 2861
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1566185333713837e-05,
+      "loss": 1.3502,
+      "step": 2862
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1565785382405945e-05,
+      "loss": 1.27,
+      "step": 2863
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.156538525090884e-05,
+      "loss": 1.2898,
+      "step": 2864
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1564984939230055e-05,
+      "loss": 1.2926,
+      "step": 2865
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1564584447377137e-05,
+      "loss": 1.3186,
+      "step": 2866
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1564183775357622e-05,
+      "loss": 1.2314,
+      "step": 2867
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1563782923179064e-05,
+      "loss": 1.4291,
+      "step": 2868
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1563381890849012e-05,
+      "loss": 1.2618,
+      "step": 2869
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1562980678375015e-05,
+      "loss": 1.238,
+      "step": 2870
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.156257928576464e-05,
+      "loss": 1.2687,
+      "step": 2871
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1562177713025446e-05,
+      "loss": 1.2761,
+      "step": 2872
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1561775960164988e-05,
+      "loss": 1.3308,
+      "step": 2873
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.156137402719084e-05,
+      "loss": 1.2953,
+      "step": 2874
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1560971914110577e-05,
+      "loss": 1.2618,
+      "step": 2875
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1560569620931773e-05,
+      "loss": 1.2218,
+      "step": 2876
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1560167147661997e-05,
+      "loss": 1.2217,
+      "step": 2877
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.155976449430884e-05,
+      "loss": 1.2464,
+      "step": 2878
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.155936166087988e-05,
+      "loss": 1.2322,
+      "step": 2879
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.155895864738271e-05,
+      "loss": 1.3078,
+      "step": 2880
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.155855545382492e-05,
+      "loss": 1.317,
+      "step": 2881
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1558152080214103e-05,
+      "loss": 1.2329,
+      "step": 2882
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.155774852655786e-05,
+      "loss": 1.2056,
+      "step": 2883
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.155734479286379e-05,
+      "loss": 1.2966,
+      "step": 2884
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.15569408791395e-05,
+      "loss": 1.1963,
+      "step": 2885
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.15565367853926e-05,
+      "loss": 1.2748,
+      "step": 2886
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1556132511630698e-05,
+      "loss": 1.1862,
+      "step": 2887
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.155572805786141e-05,
+      "loss": 1.2787,
+      "step": 2888
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1555323424092357e-05,
+      "loss": 1.1911,
+      "step": 2889
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.155491861033116e-05,
+      "loss": 1.2903,
+      "step": 2890
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1554513616585444e-05,
+      "loss": 1.2473,
+      "step": 2891
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1554108442862838e-05,
+      "loss": 1.2581,
+      "step": 2892
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1553703089170975e-05,
+      "loss": 1.264,
+      "step": 2893
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.155329755551749e-05,
+      "loss": 1.2049,
+      "step": 2894
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1552891841910018e-05,
+      "loss": 1.2563,
+      "step": 2895
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1552485948356207e-05,
+      "loss": 1.3083,
+      "step": 2896
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.15520798748637e-05,
+      "loss": 1.22,
+      "step": 2897
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.155167362144015e-05,
+      "loss": 1.1365,
+      "step": 2898
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1551267188093205e-05,
+      "loss": 1.1795,
+      "step": 2899
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1550860574830522e-05,
+      "loss": 1.3452,
+      "step": 2900
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1550453781659762e-05,
+      "loss": 1.1735,
+      "step": 2901
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1550046808588583e-05,
+      "loss": 1.226,
+      "step": 2902
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1549639655624658e-05,
+      "loss": 1.2178,
+      "step": 2903
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.154923232277565e-05,
+      "loss": 1.1939,
+      "step": 2904
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.154882481004924e-05,
+      "loss": 1.1857,
+      "step": 2905
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1548417117453094e-05,
+      "loss": 1.2353,
+      "step": 2906
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.15480092449949e-05,
+      "loss": 1.2913,
+      "step": 2907
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1547601192682336e-05,
+      "loss": 1.2821,
+      "step": 2908
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.154719296052309e-05,
+      "loss": 1.2184,
+      "step": 2909
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1546784548524853e-05,
+      "loss": 1.1648,
+      "step": 2910
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1546375956695317e-05,
+      "loss": 1.1899,
+      "step": 2911
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1545967185042177e-05,
+      "loss": 1.2159,
+      "step": 2912
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1545558233573138e-05,
+      "loss": 1.1786,
+      "step": 2913
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1545149102295897e-05,
+      "loss": 1.1926,
+      "step": 2914
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1544739791218166e-05,
+      "loss": 1.1552,
+      "step": 2915
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1544330300347656e-05,
+      "loss": 1.1758,
+      "step": 2916
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1543920629692072e-05,
+      "loss": 1.2293,
+      "step": 2917
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1543510779259136e-05,
+      "loss": 1.212,
+      "step": 2918
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1543100749056572e-05,
+      "loss": 1.0969,
+      "step": 2919
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.15426905390921e-05,
+      "loss": 1.1564,
+      "step": 2920
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1542280149373444e-05,
+      "loss": 1.2845,
+      "step": 2921
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1541869579908344e-05,
+      "loss": 1.2273,
+      "step": 2922
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1541458830704522e-05,
+      "loss": 1.2224,
+      "step": 2923
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1541047901769725e-05,
+      "loss": 1.1435,
+      "step": 2924
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1540636793111687e-05,
+      "loss": 1.2565,
+      "step": 2925
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1540225504738157e-05,
+      "loss": 1.1443,
+      "step": 2926
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1539814036656877e-05,
+      "loss": 1.2359,
+      "step": 2927
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.15394023888756e-05,
+      "loss": 1.1576,
+      "step": 2928
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1538990561402085e-05,
+      "loss": 1.1715,
+      "step": 2929
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1538578554244083e-05,
+      "loss": 1.1774,
+      "step": 2930
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1538166367409357e-05,
+      "loss": 1.1875,
+      "step": 2931
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.153775400090567e-05,
+      "loss": 1.1947,
+      "step": 2932
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1537341454740795e-05,
+      "loss": 1.1854,
+      "step": 2933
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1536928728922497e-05,
+      "loss": 1.1894,
+      "step": 2934
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.153651582345855e-05,
+      "loss": 1.1014,
+      "step": 2935
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.153610273835674e-05,
+      "loss": 1.1799,
+      "step": 2936
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.153568947362484e-05,
+      "loss": 1.2885,
+      "step": 2937
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1535276029270638e-05,
+      "loss": 1.2711,
+      "step": 2938
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1534862405301923e-05,
+      "loss": 1.2066,
+      "step": 2939
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1534448601726483e-05,
+      "loss": 1.1117,
+      "step": 2940
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1534034618552116e-05,
+      "loss": 1.1988,
+      "step": 2941
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1533620455786614e-05,
+      "loss": 1.1743,
+      "step": 2942
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.153320611343779e-05,
+      "loss": 1.1802,
+      "step": 2943
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1532791591513438e-05,
+      "loss": 1.2112,
+      "step": 2944
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.153237689002137e-05,
+      "loss": 1.2009,
+      "step": 2945
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1531962008969397e-05,
+      "loss": 1.135,
+      "step": 2946
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1531546948365336e-05,
+      "loss": 1.1161,
+      "step": 2947
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1531131708217006e-05,
+      "loss": 1.2217,
+      "step": 2948
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1530716288532227e-05,
+      "loss": 1.1808,
+      "step": 2949
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1530300689318824e-05,
+      "loss": 1.1848,
+      "step": 2950
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1529884910584628e-05,
+      "loss": 1.2293,
+      "step": 2951
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.152946895233747e-05,
+      "loss": 1.2192,
+      "step": 2952
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.152905281458518e-05,
+      "loss": 1.1892,
+      "step": 2953
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1528636497335603e-05,
+      "loss": 1.1643,
+      "step": 2954
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.152822000059658e-05,
+      "loss": 1.1847,
+      "step": 2955
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1527803324375952e-05,
+      "loss": 1.1878,
+      "step": 2956
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1527386468681576e-05,
+      "loss": 1.2619,
+      "step": 2957
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.15269694335213e-05,
+      "loss": 1.2568,
+      "step": 2958
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1526552218902977e-05,
+      "loss": 1.1258,
+      "step": 2959
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1526134824834467e-05,
+      "loss": 1.2264,
+      "step": 2960
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1525717251323636e-05,
+      "loss": 1.1846,
+      "step": 2961
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.152529949837835e-05,
+      "loss": 1.185,
+      "step": 2962
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1524881566006473e-05,
+      "loss": 1.2032,
+      "step": 2963
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.152446345421588e-05,
+      "loss": 1.1566,
+      "step": 2964
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1524045163014448e-05,
+      "loss": 1.1462,
+      "step": 2965
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1523626692410054e-05,
+      "loss": 1.2268,
+      "step": 2966
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1523208042410587e-05,
+      "loss": 1.1696,
+      "step": 2967
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.152278921302392e-05,
+      "loss": 1.2012,
+      "step": 2968
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1522370204257957e-05,
+      "loss": 1.2188,
+      "step": 2969
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1521951016120583e-05,
+      "loss": 1.1976,
+      "step": 2970
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1521531648619696e-05,
+      "loss": 1.2638,
+      "step": 2971
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1521112101763198e-05,
+      "loss": 1.1562,
+      "step": 2972
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1520692375558983e-05,
+      "loss": 1.2473,
+      "step": 2973
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1520272470014965e-05,
+      "loss": 1.243,
+      "step": 2974
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1519852385139053e-05,
+      "loss": 1.1741,
+      "step": 2975
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.151943212093916e-05,
+      "loss": 1.1781,
+      "step": 2976
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.15190116774232e-05,
+      "loss": 1.2016,
+      "step": 2977
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1518591054599096e-05,
+      "loss": 1.1739,
+      "step": 2978
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.151817025247477e-05,
+      "loss": 1.175,
+      "step": 2979
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.151774927105815e-05,
+      "loss": 1.2262,
+      "step": 2980
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.151732811035716e-05,
+      "loss": 1.1546,
+      "step": 2981
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1516906770379742e-05,
+      "loss": 1.178,
+      "step": 2982
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1516485251133826e-05,
+      "loss": 1.1794,
+      "step": 2983
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.151606355262735e-05,
+      "loss": 1.195,
+      "step": 2984
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1515641674868266e-05,
+      "loss": 1.1986,
+      "step": 2985
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1515219617864516e-05,
+      "loss": 1.1791,
+      "step": 2986
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.151479738162405e-05,
+      "loss": 1.0801,
+      "step": 2987
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1514374966154827e-05,
+      "loss": 1.2266,
+      "step": 2988
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1513952371464798e-05,
+      "loss": 1.2352,
+      "step": 2989
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.151352959756192e-05,
+      "loss": 1.157,
+      "step": 2990
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1513106644454165e-05,
+      "loss": 1.2275,
+      "step": 2991
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1512683512149495e-05,
+      "loss": 1.1493,
+      "step": 2992
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1512260200655886e-05,
+      "loss": 1.2033,
+      "step": 2993
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1511836709981307e-05,
+      "loss": 1.1835,
+      "step": 2994
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1511413040133733e-05,
+      "loss": 1.2027,
+      "step": 2995
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1510989191121148e-05,
+      "loss": 1.1723,
+      "step": 2996
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.151056516295154e-05,
+      "loss": 1.2251,
+      "step": 2997
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.151014095563289e-05,
+      "loss": 1.235,
+      "step": 2998
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.150971656917319e-05,
+      "loss": 1.2837,
+      "step": 2999
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.150929200358043e-05,
+      "loss": 1.2012,
+      "step": 3000
+    },
+    {
+      "epoch": 0.13,
+      "eval_text_loss": 0.8825516700744629,
+      "eval_text_runtime": 1.3913,
+      "eval_text_samples_per_second": 53.187,
+      "eval_text_steps_per_second": 1.437,
+      "step": 3000
+    },
+    {
+      "epoch": 0.13,
+      "eval_svit_loss": 1.2908438444137573,
+      "eval_svit_runtime": 4.8956,
+      "eval_svit_samples_per_second": 139.106,
+      "eval_svit_steps_per_second": 2.247,
+      "step": 3000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1508867258862623e-05,
+      "loss": 1.2032,
+      "step": 3001
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1508442335027752e-05,
+      "loss": 1.2053,
+      "step": 3002
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.150801723208383e-05,
+      "loss": 1.1988,
+      "step": 3003
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1507591950038865e-05,
+      "loss": 1.2714,
+      "step": 3004
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1507166488900865e-05,
+      "loss": 1.2091,
+      "step": 3005
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1506740848677846e-05,
+      "loss": 1.123,
+      "step": 3006
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1506315029377822e-05,
+      "loss": 1.1832,
+      "step": 3007
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1505889031008823e-05,
+      "loss": 1.2181,
+      "step": 3008
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.150546285357886e-05,
+      "loss": 1.2384,
+      "step": 3009
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1505036497095976e-05,
+      "loss": 1.2608,
+      "step": 3010
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1504609961568192e-05,
+      "loss": 1.1729,
+      "step": 3011
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1504183247003545e-05,
+      "loss": 1.3005,
+      "step": 3012
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1503756353410073e-05,
+      "loss": 1.1855,
+      "step": 3013
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1503329280795817e-05,
+      "loss": 1.2538,
+      "step": 3014
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1502902029168826e-05,
+      "loss": 1.188,
+      "step": 3015
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.150247459853714e-05,
+      "loss": 1.228,
+      "step": 3016
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.150204698890882e-05,
+      "loss": 1.2316,
+      "step": 3017
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.150161920029191e-05,
+      "loss": 1.1825,
+      "step": 3018
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1501191232694478e-05,
+      "loss": 1.1992,
+      "step": 3019
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1500763086124577e-05,
+      "loss": 1.137,
+      "step": 3020
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1500334760590278e-05,
+      "loss": 1.2961,
+      "step": 3021
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.149990625609965e-05,
+      "loss": 1.2472,
+      "step": 3022
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.149947757266076e-05,
+      "loss": 1.2358,
+      "step": 3023
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1499048710281687e-05,
+      "loss": 1.1915,
+      "step": 3024
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1498619668970508e-05,
+      "loss": 1.204,
+      "step": 3025
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1498190448735304e-05,
+      "loss": 1.2102,
+      "step": 3026
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.149776104958416e-05,
+      "loss": 1.2671,
+      "step": 3027
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1497331471525166e-05,
+      "loss": 1.2321,
+      "step": 3028
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1496901714566414e-05,
+      "loss": 1.2062,
+      "step": 3029
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1496471778715997e-05,
+      "loss": 1.277,
+      "step": 3030
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1496041663982013e-05,
+      "loss": 1.1999,
+      "step": 3031
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.149561137037257e-05,
+      "loss": 1.1461,
+      "step": 3032
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1495180897895767e-05,
+      "loss": 1.1804,
+      "step": 3033
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1494750246559717e-05,
+      "loss": 1.2519,
+      "step": 3034
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1494319416372527e-05,
+      "loss": 1.2394,
+      "step": 3035
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.149388840734232e-05,
+      "loss": 1.2443,
+      "step": 3036
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1493457219477207e-05,
+      "loss": 1.1457,
+      "step": 3037
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1493025852785315e-05,
+      "loss": 1.2788,
+      "step": 3038
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1492594307274773e-05,
+      "loss": 1.2136,
+      "step": 3039
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.14921625829537e-05,
+      "loss": 1.194,
+      "step": 3040
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1491730679830234e-05,
+      "loss": 1.1259,
+      "step": 3041
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.149129859791251e-05,
+      "loss": 1.1711,
+      "step": 3042
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.149086633720867e-05,
+      "loss": 1.1814,
+      "step": 3043
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1490433897726853e-05,
+      "loss": 1.205,
+      "step": 3044
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1490001279475203e-05,
+      "loss": 1.2089,
+      "step": 3045
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1489568482461873e-05,
+      "loss": 1.2049,
+      "step": 3046
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1489135506695017e-05,
+      "loss": 1.1799,
+      "step": 3047
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.148870235218279e-05,
+      "loss": 1.2315,
+      "step": 3048
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1488269018933345e-05,
+      "loss": 1.1774,
+      "step": 3049
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1487835506954847e-05,
+      "loss": 1.2572,
+      "step": 3050
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.148740181625547e-05,
+      "loss": 1.2315,
+      "step": 3051
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1486967946843372e-05,
+      "loss": 1.2161,
+      "step": 3052
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1486533898726736e-05,
+      "loss": 1.2415,
+      "step": 3053
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1486099671913732e-05,
+      "loss": 1.173,
+      "step": 3054
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.148566526641254e-05,
+      "loss": 1.2347,
+      "step": 3055
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.148523068223134e-05,
+      "loss": 1.1835,
+      "step": 3056
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.148479591937833e-05,
+      "loss": 1.1996,
+      "step": 3057
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.148436097786169e-05,
+      "loss": 1.1751,
+      "step": 3058
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.148392585768961e-05,
+      "loss": 1.2105,
+      "step": 3059
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.148349055887029e-05,
+      "loss": 1.1838,
+      "step": 3060
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1483055081411933e-05,
+      "loss": 1.2304,
+      "step": 3061
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.148261942532274e-05,
+      "loss": 1.1774,
+      "step": 3062
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1482183590610916e-05,
+      "loss": 1.0799,
+      "step": 3063
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.148174757728467e-05,
+      "loss": 1.2307,
+      "step": 3064
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.148131138535222e-05,
+      "loss": 1.1994,
+      "step": 3065
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1480875014821777e-05,
+      "loss": 1.2176,
+      "step": 3066
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.148043846570156e-05,
+      "loss": 1.1333,
+      "step": 3067
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1480001737999797e-05,
+      "loss": 1.0776,
+      "step": 3068
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1479564831724712e-05,
+      "loss": 1.1653,
+      "step": 3069
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1479127746884538e-05,
+      "loss": 1.1744,
+      "step": 3070
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1478690483487505e-05,
+      "loss": 1.2727,
+      "step": 3071
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.147825304154185e-05,
+      "loss": 1.1403,
+      "step": 3072
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.147781542105581e-05,
+      "loss": 1.2243,
+      "step": 3073
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.147737762203764e-05,
+      "loss": 1.1466,
+      "step": 3074
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1476939644495573e-05,
+      "loss": 1.1937,
+      "step": 3075
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1476501488437866e-05,
+      "loss": 1.1642,
+      "step": 3076
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.147606315387277e-05,
+      "loss": 1.2539,
+      "step": 3077
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1475624640808543e-05,
+      "loss": 1.2581,
+      "step": 3078
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.147518594925345e-05,
+      "loss": 1.1762,
+      "step": 3079
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1474747079215746e-05,
+      "loss": 1.1832,
+      "step": 3080
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.14743080307037e-05,
+      "loss": 1.1181,
+      "step": 3081
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.147386880372559e-05,
+      "loss": 1.2217,
+      "step": 3082
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.147342939828968e-05,
+      "loss": 1.1957,
+      "step": 3083
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.147298981440425e-05,
+      "loss": 1.1281,
+      "step": 3084
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1472550052077583e-05,
+      "loss": 1.1655,
+      "step": 3085
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.147211011131796e-05,
+      "loss": 1.1209,
+      "step": 3086
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.147166999213367e-05,
+      "loss": 1.1584,
+      "step": 3087
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1471229694533005e-05,
+      "loss": 1.1311,
+      "step": 3088
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.147078921852425e-05,
+      "loss": 1.2156,
+      "step": 3089
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.147034856411571e-05,
+      "loss": 1.0757,
+      "step": 3090
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.146990773131569e-05,
+      "loss": 1.1996,
+      "step": 3091
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.146946672013248e-05,
+      "loss": 1.2014,
+      "step": 3092
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1469025530574397e-05,
+      "loss": 1.1145,
+      "step": 3093
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1468584162649753e-05,
+      "loss": 1.1453,
+      "step": 3094
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.146814261636686e-05,
+      "loss": 1.2528,
+      "step": 3095
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1467700891734028e-05,
+      "loss": 1.1557,
+      "step": 3096
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1467258988759586e-05,
+      "loss": 1.1896,
+      "step": 3097
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1466816907451858e-05,
+      "loss": 1.1145,
+      "step": 3098
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1466374647819167e-05,
+      "loss": 1.2121,
+      "step": 3099
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.146593220986985e-05,
+      "loss": 1.1764,
+      "step": 3100
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1465489593612232e-05,
+      "loss": 1.1487,
+      "step": 3101
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1465046799054658e-05,
+      "loss": 1.2011,
+      "step": 3102
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1464603826205467e-05,
+      "loss": 1.1209,
+      "step": 3103
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1464160675073e-05,
+      "loss": 1.1766,
+      "step": 3104
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.146371734566561e-05,
+      "loss": 1.2891,
+      "step": 3105
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1463273837991644e-05,
+      "loss": 1.1169,
+      "step": 3106
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.146283015205946e-05,
+      "loss": 1.1059,
+      "step": 3107
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.146238628787741e-05,
+      "loss": 1.1967,
+      "step": 3108
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1461942245453858e-05,
+      "loss": 1.1953,
+      "step": 3109
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.146149802479717e-05,
+      "loss": 1.1184,
+      "step": 3110
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1461053625915713e-05,
+      "loss": 1.2587,
+      "step": 3111
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1460609048817853e-05,
+      "loss": 1.1641,
+      "step": 3112
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1460164293511976e-05,
+      "loss": 1.1786,
+      "step": 3113
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.145971936000645e-05,
+      "loss": 1.1655,
+      "step": 3114
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1459274248309655e-05,
+      "loss": 1.1641,
+      "step": 3115
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1458828958429982e-05,
+      "loss": 1.1502,
+      "step": 3116
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1458383490375814e-05,
+      "loss": 1.17,
+      "step": 3117
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1457937844155547e-05,
+      "loss": 1.1834,
+      "step": 3118
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1457492019777573e-05,
+      "loss": 1.1191,
+      "step": 3119
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1457046017250284e-05,
+      "loss": 1.2992,
+      "step": 3120
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1456599836582093e-05,
+      "loss": 1.1963,
+      "step": 3121
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1456153477781397e-05,
+      "loss": 1.1406,
+      "step": 3122
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1455706940856603e-05,
+      "loss": 1.154,
+      "step": 3123
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.145526022581613e-05,
+      "loss": 1.2025,
+      "step": 3124
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1454813332668384e-05,
+      "loss": 1.2077,
+      "step": 3125
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1454366261421787e-05,
+      "loss": 1.1646,
+      "step": 3126
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.145391901208476e-05,
+      "loss": 1.2165,
+      "step": 3127
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.145347158466573e-05,
+      "loss": 1.136,
+      "step": 3128
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1453023979173122e-05,
+      "loss": 1.2022,
+      "step": 3129
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1452576195615367e-05,
+      "loss": 1.1765,
+      "step": 3130
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1452128234000903e-05,
+      "loss": 1.1252,
+      "step": 3131
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1451680094338164e-05,
+      "loss": 1.1406,
+      "step": 3132
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.14512317766356e-05,
+      "loss": 1.2193,
+      "step": 3133
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1450783280901643e-05,
+      "loss": 1.1624,
+      "step": 3134
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1450334607144753e-05,
+      "loss": 1.2408,
+      "step": 3135
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1449885755373376e-05,
+      "loss": 1.1481,
+      "step": 3136
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1449436725595965e-05,
+      "loss": 1.2096,
+      "step": 3137
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1448987517820984e-05,
+      "loss": 1.1816,
+      "step": 3138
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.144853813205689e-05,
+      "loss": 1.1327,
+      "step": 3139
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1448088568312147e-05,
+      "loss": 1.1654,
+      "step": 3140
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.144763882659523e-05,
+      "loss": 1.1895,
+      "step": 3141
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.144718890691461e-05,
+      "loss": 1.2158,
+      "step": 3142
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1446738809278754e-05,
+      "loss": 1.0723,
+      "step": 3143
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1446288533696147e-05,
+      "loss": 1.2278,
+      "step": 3144
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1445838080175268e-05,
+      "loss": 1.2156,
+      "step": 3145
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1445387448724602e-05,
+      "loss": 1.0915,
+      "step": 3146
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.144493663935264e-05,
+      "loss": 1.0453,
+      "step": 3147
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.144448565206787e-05,
+      "loss": 1.1933,
+      "step": 3148
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1444034486878793e-05,
+      "loss": 1.1696,
+      "step": 3149
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1443583143793905e-05,
+      "loss": 1.1301,
+      "step": 3150
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1443131622821703e-05,
+      "loss": 1.1817,
+      "step": 3151
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.14426799239707e-05,
+      "loss": 1.1603,
+      "step": 3152
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.14422280472494e-05,
+      "loss": 1.1124,
+      "step": 3153
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1441775992666314e-05,
+      "loss": 1.1868,
+      "step": 3154
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.144132376022996e-05,
+      "loss": 1.19,
+      "step": 3155
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1440871349948857e-05,
+      "loss": 1.1755,
+      "step": 3156
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1440418761831524e-05,
+      "loss": 1.1206,
+      "step": 3157
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.143996599588649e-05,
+      "loss": 1.2197,
+      "step": 3158
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.143951305212228e-05,
+      "loss": 1.0845,
+      "step": 3159
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.143905993054743e-05,
+      "loss": 1.2127,
+      "step": 3160
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1438606631170468e-05,
+      "loss": 1.1486,
+      "step": 3161
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1438153153999943e-05,
+      "loss": 1.1611,
+      "step": 3162
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.143769949904439e-05,
+      "loss": 1.1075,
+      "step": 3163
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1437245666312354e-05,
+      "loss": 1.0885,
+      "step": 3164
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1436791655812393e-05,
+      "loss": 0.9837,
+      "step": 3165
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1436337467553044e-05,
+      "loss": 1.2348,
+      "step": 3166
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1435883101542875e-05,
+      "loss": 1.1125,
+      "step": 3167
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1435428557790442e-05,
+      "loss": 1.2605,
+      "step": 3168
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1434973836304303e-05,
+      "loss": 1.0993,
+      "step": 3169
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1434518937093026e-05,
+      "loss": 1.089,
+      "step": 3170
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.143406386016518e-05,
+      "loss": 1.1589,
+      "step": 3171
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1433608605529342e-05,
+      "loss": 1.2203,
+      "step": 3172
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1433153173194075e-05,
+      "loss": 1.1616,
+      "step": 3173
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1432697563167975e-05,
+      "loss": 1.157,
+      "step": 3174
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.143224177545961e-05,
+      "loss": 1.2318,
+      "step": 3175
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.143178581007757e-05,
+      "loss": 1.1778,
+      "step": 3176
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1431329667030445e-05,
+      "loss": 1.2172,
+      "step": 3177
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1430873346326827e-05,
+      "loss": 1.1825,
+      "step": 3178
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.143041684797531e-05,
+      "loss": 1.2112,
+      "step": 3179
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1429960171984497e-05,
+      "loss": 1.2283,
+      "step": 3180
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1429503318362988e-05,
+      "loss": 1.1611,
+      "step": 3181
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.142904628711939e-05,
+      "loss": 1.0934,
+      "step": 3182
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1428589078262307e-05,
+      "loss": 1.2049,
+      "step": 3183
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1428131691800354e-05,
+      "loss": 1.097,
+      "step": 3184
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.142767412774215e-05,
+      "loss": 1.1825,
+      "step": 3185
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1427216386096314e-05,
+      "loss": 1.1705,
+      "step": 3186
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1426758466871463e-05,
+      "loss": 1.1427,
+      "step": 3187
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1426300370076225e-05,
+      "loss": 1.158,
+      "step": 3188
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.142584209571923e-05,
+      "loss": 1.2012,
+      "step": 3189
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1425383643809113e-05,
+      "loss": 1.0914,
+      "step": 3190
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.142492501435451e-05,
+      "loss": 1.1833,
+      "step": 3191
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1424466207364052e-05,
+      "loss": 1.1808,
+      "step": 3192
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.142400722284639e-05,
+      "loss": 1.1373,
+      "step": 3193
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1423548060810163e-05,
+      "loss": 1.0951,
+      "step": 3194
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1423088721264027e-05,
+      "loss": 1.1756,
+      "step": 3195
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1422629204216628e-05,
+      "loss": 1.1403,
+      "step": 3196
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.142216950967663e-05,
+      "loss": 1.0398,
+      "step": 3197
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1421709637652684e-05,
+      "loss": 1.166,
+      "step": 3198
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1421249588153453e-05,
+      "loss": 1.1201,
+      "step": 3199
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1420789361187608e-05,
+      "loss": 1.1203,
+      "step": 3200
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.142032895676382e-05,
+      "loss": 1.1504,
+      "step": 3201
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1419868374890755e-05,
+      "loss": 1.2525,
+      "step": 3202
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.141940761557709e-05,
+      "loss": 1.1874,
+      "step": 3203
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1418946678831508e-05,
+      "loss": 1.1447,
+      "step": 3204
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.141848556466269e-05,
+      "loss": 1.2317,
+      "step": 3205
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.141802427307932e-05,
+      "loss": 1.131,
+      "step": 3206
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1417562804090087e-05,
+      "loss": 1.2202,
+      "step": 3207
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.141710115770369e-05,
+      "loss": 1.1302,
+      "step": 3208
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1416639333928812e-05,
+      "loss": 1.1778,
+      "step": 3209
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1416177332774163e-05,
+      "loss": 1.0809,
+      "step": 3210
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1415715154248447e-05,
+      "loss": 1.1519,
+      "step": 3211
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1415252798360364e-05,
+      "loss": 1.2008,
+      "step": 3212
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1414790265118625e-05,
+      "loss": 1.2159,
+      "step": 3213
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.141432755453194e-05,
+      "loss": 1.053,
+      "step": 3214
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1413864666609037e-05,
+      "loss": 1.1111,
+      "step": 3215
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1413401601358617e-05,
+      "loss": 1.1707,
+      "step": 3216
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1412938358789418e-05,
+      "loss": 1.1594,
+      "step": 3217
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1412474938910158e-05,
+      "loss": 1.1129,
+      "step": 3218
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.141201134172957e-05,
+      "loss": 1.169,
+      "step": 3219
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.141154756725638e-05,
+      "loss": 1.1385,
+      "step": 3220
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1411083615499333e-05,
+      "loss": 1.2406,
+      "step": 3221
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1410619486467166e-05,
+      "loss": 1.0966,
+      "step": 3222
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1410155180168618e-05,
+      "loss": 1.1881,
+      "step": 3223
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1409690696612437e-05,
+      "loss": 1.0249,
+      "step": 3224
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1409226035807375e-05,
+      "loss": 1.1369,
+      "step": 3225
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.140876119776218e-05,
+      "loss": 1.1914,
+      "step": 3226
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1408296182485614e-05,
+      "loss": 1.181,
+      "step": 3227
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.140783098998643e-05,
+      "loss": 1.0935,
+      "step": 3228
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1407365620273396e-05,
+      "loss": 1.1963,
+      "step": 3229
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1406900073355276e-05,
+      "loss": 1.202,
+      "step": 3230
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1406434349240837e-05,
+      "loss": 1.1894,
+      "step": 3231
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1405968447938857e-05,
+      "loss": 1.2505,
+      "step": 3232
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.140550236945811e-05,
+      "loss": 1.1638,
+      "step": 3233
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.140503611380737e-05,
+      "loss": 1.1092,
+      "step": 3234
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1404569680995428e-05,
+      "loss": 1.2216,
+      "step": 3235
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1404103071031064e-05,
+      "loss": 1.1523,
+      "step": 3236
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1403636283923074e-05,
+      "loss": 1.1937,
+      "step": 3237
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.140316931968024e-05,
+      "loss": 1.091,
+      "step": 3238
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1402702178311372e-05,
+      "loss": 1.0673,
+      "step": 3239
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.140223485982526e-05,
+      "loss": 1.0843,
+      "step": 3240
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.140176736423071e-05,
+      "loss": 1.1443,
+      "step": 3241
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1401299691536524e-05,
+      "loss": 1.1884,
+      "step": 3242
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.140083184175152e-05,
+      "loss": 1.2333,
+      "step": 3243
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.14003638148845e-05,
+      "loss": 1.1948,
+      "step": 3244
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.139989561094429e-05,
+      "loss": 1.167,
+      "step": 3245
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1399427229939705e-05,
+      "loss": 1.2377,
+      "step": 3246
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1398958671879566e-05,
+      "loss": 1.2642,
+      "step": 3247
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.13984899367727e-05,
+      "loss": 1.2155,
+      "step": 3248
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1398021024627942e-05,
+      "loss": 1.1741,
+      "step": 3249
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.139755193545412e-05,
+      "loss": 1.1197,
+      "step": 3250
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1397082669260067e-05,
+      "loss": 1.1745,
+      "step": 3251
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.139661322605463e-05,
+      "loss": 1.1442,
+      "step": 3252
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.139614360584665e-05,
+      "loss": 1.186,
+      "step": 3253
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.139567380864497e-05,
+      "loss": 1.1699,
+      "step": 3254
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1395203834458438e-05,
+      "loss": 1.1601,
+      "step": 3255
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1394733683295912e-05,
+      "loss": 1.1776,
+      "step": 3256
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1394263355166243e-05,
+      "loss": 1.2332,
+      "step": 3257
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1393792850078295e-05,
+      "loss": 1.0384,
+      "step": 3258
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.139332216804093e-05,
+      "loss": 1.2457,
+      "step": 3259
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1392851309063013e-05,
+      "loss": 1.1284,
+      "step": 3260
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1392380273153408e-05,
+      "loss": 1.2635,
+      "step": 3261
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1391909060321003e-05,
+      "loss": 1.2225,
+      "step": 3262
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1391437670574657e-05,
+      "loss": 1.1179,
+      "step": 3263
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.139096610392326e-05,
+      "loss": 1.0929,
+      "step": 3264
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1390494360375695e-05,
+      "loss": 1.1598,
+      "step": 3265
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1390022439940843e-05,
+      "loss": 1.1668,
+      "step": 3266
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1389550342627594e-05,
+      "loss": 1.1096,
+      "step": 3267
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1389078068444844e-05,
+      "loss": 1.1668,
+      "step": 3268
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.138860561740149e-05,
+      "loss": 1.1049,
+      "step": 3269
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1388132989506424e-05,
+      "loss": 1.1837,
+      "step": 3270
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1387660184768558e-05,
+      "loss": 1.0853,
+      "step": 3271
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1387187203196796e-05,
+      "loss": 1.109,
+      "step": 3272
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.138671404480004e-05,
+      "loss": 1.0779,
+      "step": 3273
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.138624070958721e-05,
+      "loss": 1.1795,
+      "step": 3274
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1385767197567224e-05,
+      "loss": 1.1155,
+      "step": 3275
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1385293508748995e-05,
+      "loss": 1.201,
+      "step": 3276
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.138481964314145e-05,
+      "loss": 1.143,
+      "step": 3277
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.138434560075351e-05,
+      "loss": 1.1486,
+      "step": 3278
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1383871381594114e-05,
+      "loss": 1.2097,
+      "step": 3279
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1383396985672188e-05,
+      "loss": 1.1954,
+      "step": 3280
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.138292241299667e-05,
+      "loss": 1.1883,
+      "step": 3281
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1382447663576496e-05,
+      "loss": 1.2281,
+      "step": 3282
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1381972737420615e-05,
+      "loss": 1.2213,
+      "step": 3283
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1381497634537965e-05,
+      "loss": 1.0849,
+      "step": 3284
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1381022354937502e-05,
+      "loss": 1.1495,
+      "step": 3285
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.138054689862818e-05,
+      "loss": 1.1395,
+      "step": 3286
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.138007126561895e-05,
+      "loss": 1.1209,
+      "step": 3287
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1379595455918774e-05,
+      "loss": 1.2304,
+      "step": 3288
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1379119469536613e-05,
+      "loss": 1.1913,
+      "step": 3289
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1378643306481437e-05,
+      "loss": 1.1966,
+      "step": 3290
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1378166966762208e-05,
+      "loss": 1.1761,
+      "step": 3291
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1377690450387906e-05,
+      "loss": 1.0982,
+      "step": 3292
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1377213757367505e-05,
+      "loss": 1.1806,
+      "step": 3293
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1376736887709983e-05,
+      "loss": 1.0475,
+      "step": 3294
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1376259841424324e-05,
+      "loss": 1.2485,
+      "step": 3295
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1375782618519515e-05,
+      "loss": 1.1059,
+      "step": 3296
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.137530521900454e-05,
+      "loss": 1.0624,
+      "step": 3297
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.13748276428884e-05,
+      "loss": 1.2183,
+      "step": 3298
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1374349890180084e-05,
+      "loss": 1.1236,
+      "step": 3299
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1373871960888595e-05,
+      "loss": 1.0675,
+      "step": 3300
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1373393855022936e-05,
+      "loss": 1.1361,
+      "step": 3301
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1372915572592112e-05,
+      "loss": 1.1153,
+      "step": 3302
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.137243711360513e-05,
+      "loss": 1.1047,
+      "step": 3303
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1371958478071003e-05,
+      "loss": 1.1362,
+      "step": 3304
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1371479665998753e-05,
+      "loss": 1.1311,
+      "step": 3305
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1371000677397394e-05,
+      "loss": 1.0564,
+      "step": 3306
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1370521512275947e-05,
+      "loss": 1.2498,
+      "step": 3307
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1370042170643446e-05,
+      "loss": 1.1013,
+      "step": 3308
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.136956265250891e-05,
+      "loss": 1.1568,
+      "step": 3309
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1369082957881377e-05,
+      "loss": 1.1581,
+      "step": 3310
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1368603086769885e-05,
+      "loss": 1.1642,
+      "step": 3311
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.136812303918347e-05,
+      "loss": 1.1476,
+      "step": 3312
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1367642815131176e-05,
+      "loss": 1.1062,
+      "step": 3313
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1367162414622045e-05,
+      "loss": 1.2232,
+      "step": 3314
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1366681837665133e-05,
+      "loss": 1.1203,
+      "step": 3315
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1366201084269487e-05,
+      "loss": 1.1577,
+      "step": 3316
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1365720154444165e-05,
+      "loss": 1.1282,
+      "step": 3317
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1365239048198228e-05,
+      "loss": 1.1578,
+      "step": 3318
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1364757765540736e-05,
+      "loss": 1.1671,
+      "step": 3319
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.136427630648075e-05,
+      "loss": 1.1928,
+      "step": 3320
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1363794671027352e-05,
+      "loss": 1.1809,
+      "step": 3321
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1363312859189603e-05,
+      "loss": 1.1935,
+      "step": 3322
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1362830870976587e-05,
+      "loss": 1.184,
+      "step": 3323
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1362348706397375e-05,
+      "loss": 1.1551,
+      "step": 3324
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1361866365461056e-05,
+      "loss": 1.0905,
+      "step": 3325
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.136138384817671e-05,
+      "loss": 1.1207,
+      "step": 3326
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1360901154553433e-05,
+      "loss": 1.2054,
+      "step": 3327
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1360418284600315e-05,
+      "loss": 0.9735,
+      "step": 3328
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1359935238326446e-05,
+      "loss": 1.1401,
+      "step": 3329
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.135945201574093e-05,
+      "loss": 1.176,
+      "step": 3330
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.135896861685287e-05,
+      "loss": 1.1542,
+      "step": 3331
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1358485041671373e-05,
+      "loss": 1.2003,
+      "step": 3332
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1358001290205543e-05,
+      "loss": 1.0745,
+      "step": 3333
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1357517362464497e-05,
+      "loss": 1.1156,
+      "step": 3334
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.135703325845735e-05,
+      "loss": 1.2594,
+      "step": 3335
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1356548978193217e-05,
+      "loss": 1.1712,
+      "step": 3336
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1356064521681228e-05,
+      "loss": 1.108,
+      "step": 3337
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1355579888930495e-05,
+      "loss": 1.1162,
+      "step": 3338
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1355095079950164e-05,
+      "loss": 1.1773,
+      "step": 3339
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1354610094749354e-05,
+      "loss": 1.1628,
+      "step": 3340
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1354124933337207e-05,
+      "loss": 1.1946,
+      "step": 3341
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1353639595722864e-05,
+      "loss": 1.1763,
+      "step": 3342
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1353154081915457e-05,
+      "loss": 1.1577,
+      "step": 3343
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1352668391924146e-05,
+      "loss": 1.2416,
+      "step": 3344
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1352182525758066e-05,
+      "loss": 1.1113,
+      "step": 3345
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.135169648342638e-05,
+      "loss": 1.204,
+      "step": 3346
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1351210264938232e-05,
+      "loss": 1.1096,
+      "step": 3347
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.135072387030279e-05,
+      "loss": 1.1682,
+      "step": 3348
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1350237299529216e-05,
+      "loss": 1.075,
+      "step": 3349
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1349750552626673e-05,
+      "loss": 1.1811,
+      "step": 3350
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.134926362960433e-05,
+      "loss": 1.2231,
+      "step": 3351
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.134877653047136e-05,
+      "loss": 1.1441,
+      "step": 3352
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1348289255236936e-05,
+      "loss": 1.1635,
+      "step": 3353
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1347801803910238e-05,
+      "loss": 1.177,
+      "step": 3354
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.134731417650045e-05,
+      "loss": 1.2154,
+      "step": 3355
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1346826373016754e-05,
+      "loss": 1.0922,
+      "step": 3356
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.134633839346834e-05,
+      "loss": 1.2021,
+      "step": 3357
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.13458502378644e-05,
+      "loss": 1.0851,
+      "step": 3358
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.134536190621413e-05,
+      "loss": 1.1755,
+      "step": 3359
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1344873398526734e-05,
+      "loss": 1.1849,
+      "step": 3360
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1344384714811397e-05,
+      "loss": 1.1292,
+      "step": 3361
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1343895855077345e-05,
+      "loss": 1.0865,
+      "step": 3362
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1343406819333773e-05,
+      "loss": 1.0659,
+      "step": 3363
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1342917607589898e-05,
+      "loss": 1.1062,
+      "step": 3364
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1342428219854934e-05,
+      "loss": 1.0933,
+      "step": 3365
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1341938656138098e-05,
+      "loss": 1.1932,
+      "step": 3366
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1341448916448615e-05,
+      "loss": 1.1708,
+      "step": 3367
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.134095900079571e-05,
+      "loss": 1.0885,
+      "step": 3368
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1340468909188607e-05,
+      "loss": 1.1901,
+      "step": 3369
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1339978641636543e-05,
+      "loss": 1.1322,
+      "step": 3370
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.133948819814875e-05,
+      "loss": 1.1248,
+      "step": 3371
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1338997578734467e-05,
+      "loss": 1.1385,
+      "step": 3372
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.133850678340294e-05,
+      "loss": 1.1081,
+      "step": 3373
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1338015812163405e-05,
+      "loss": 1.169,
+      "step": 3374
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1337524665025116e-05,
+      "loss": 1.1742,
+      "step": 3375
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1337033341997324e-05,
+      "loss": 1.2647,
+      "step": 3376
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.133654184308929e-05,
+      "loss": 1.1676,
+      "step": 3377
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.133605016831026e-05,
+      "loss": 1.1866,
+      "step": 3378
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1335558317669503e-05,
+      "loss": 1.1567,
+      "step": 3379
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1335066291176285e-05,
+      "loss": 1.169,
+      "step": 3380
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.133457408883987e-05,
+      "loss": 1.0894,
+      "step": 3381
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1334081710669534e-05,
+      "loss": 1.0653,
+      "step": 3382
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.133358915667455e-05,
+      "loss": 1.0654,
+      "step": 3383
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1333096426864195e-05,
+      "loss": 1.1456,
+      "step": 3384
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1332603521247753e-05,
+      "loss": 1.2264,
+      "step": 3385
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.13321104398345e-05,
+      "loss": 1.1541,
+      "step": 3386
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1331617182633735e-05,
+      "loss": 1.2512,
+      "step": 3387
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.133112374965475e-05,
+      "loss": 1.172,
+      "step": 3388
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.133063014090683e-05,
+      "loss": 1.1025,
+      "step": 3389
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.133013635639928e-05,
+      "loss": 1.2072,
+      "step": 3390
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1329642396141397e-05,
+      "loss": 1.2317,
+      "step": 3391
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.132914826014249e-05,
+      "loss": 1.1539,
+      "step": 3392
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1328653948411867e-05,
+      "loss": 1.147,
+      "step": 3393
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1328159460958835e-05,
+      "loss": 1.1215,
+      "step": 3394
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1327664797792714e-05,
+      "loss": 1.1974,
+      "step": 3395
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1327169958922815e-05,
+      "loss": 1.184,
+      "step": 3396
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1326674944358465e-05,
+      "loss": 1.1,
+      "step": 3397
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1326179754108986e-05,
+      "loss": 1.132,
+      "step": 3398
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1325684388183708e-05,
+      "loss": 1.1302,
+      "step": 3399
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.132518884659196e-05,
+      "loss": 1.1175,
+      "step": 3400
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.132469312934308e-05,
+      "loss": 1.0691,
+      "step": 3401
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1324197236446398e-05,
+      "loss": 1.2218,
+      "step": 3402
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.132370116791126e-05,
+      "loss": 1.1193,
+      "step": 3403
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.132320492374701e-05,
+      "loss": 1.1903,
+      "step": 3404
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1322708503963e-05,
+      "loss": 1.1596,
+      "step": 3405
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1322211908568574e-05,
+      "loss": 1.1432,
+      "step": 3406
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.132171513757309e-05,
+      "loss": 1.1587,
+      "step": 3407
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1321218190985907e-05,
+      "loss": 1.1402,
+      "step": 3408
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1320721068816377e-05,
+      "loss": 1.1577,
+      "step": 3409
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1320223771073878e-05,
+      "loss": 1.1558,
+      "step": 3410
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1319726297767764e-05,
+      "loss": 1.2552,
+      "step": 3411
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1319228648907415e-05,
+      "loss": 1.1616,
+      "step": 3412
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1318730824502208e-05,
+      "loss": 1.2178,
+      "step": 3413
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1318232824561508e-05,
+      "loss": 1.1671,
+      "step": 3414
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1317734649094704e-05,
+      "loss": 1.1785,
+      "step": 3415
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1317236298111177e-05,
+      "loss": 1.1034,
+      "step": 3416
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.131673777162032e-05,
+      "loss": 1.1873,
+      "step": 3417
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1316239069631517e-05,
+      "loss": 1.0922,
+      "step": 3418
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1315740192154167e-05,
+      "loss": 1.162,
+      "step": 3419
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1315241139197662e-05,
+      "loss": 1.0872,
+      "step": 3420
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1314741910771407e-05,
+      "loss": 1.172,
+      "step": 3421
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1314242506884802e-05,
+      "loss": 1.1889,
+      "step": 3422
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.131374292754726e-05,
+      "loss": 1.1794,
+      "step": 3423
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1313243172768184e-05,
+      "loss": 1.1903,
+      "step": 3424
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1312743242556996e-05,
+      "loss": 1.218,
+      "step": 3425
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.131224313692311e-05,
+      "loss": 1.1134,
+      "step": 3426
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.131174285587594e-05,
+      "loss": 1.3343,
+      "step": 3427
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.131124239942492e-05,
+      "loss": 1.1457,
+      "step": 3428
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.131074176757947e-05,
+      "loss": 1.2231,
+      "step": 3429
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.131024096034902e-05,
+      "loss": 1.2056,
+      "step": 3430
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.130973997774301e-05,
+      "loss": 1.1426,
+      "step": 3431
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.130923881977087e-05,
+      "loss": 1.1225,
+      "step": 3432
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1308737486442046e-05,
+      "loss": 1.2176,
+      "step": 3433
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1308235977765977e-05,
+      "loss": 1.0972,
+      "step": 3434
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.130773429375211e-05,
+      "loss": 1.1523,
+      "step": 3435
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.13072324344099e-05,
+      "loss": 1.1942,
+      "step": 3436
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1306730399748796e-05,
+      "loss": 1.1584,
+      "step": 3437
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1306228189778256e-05,
+      "loss": 1.1047,
+      "step": 3438
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1305725804507743e-05,
+      "loss": 1.099,
+      "step": 3439
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1305223243946712e-05,
+      "loss": 1.1893,
+      "step": 3440
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1304720508104637e-05,
+      "loss": 1.2043,
+      "step": 3441
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1304217596990984e-05,
+      "loss": 1.1899,
+      "step": 3442
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.130371451061523e-05,
+      "loss": 1.2874,
+      "step": 3443
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1303211248986852e-05,
+      "loss": 1.2121,
+      "step": 3444
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1302707812115326e-05,
+      "loss": 1.135,
+      "step": 3445
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1302204200010136e-05,
+      "loss": 1.2777,
+      "step": 3446
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.130170041268077e-05,
+      "loss": 1.1624,
+      "step": 3447
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.130119645013672e-05,
+      "loss": 1.1814,
+      "step": 3448
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.130069231238747e-05,
+      "loss": 1.1699,
+      "step": 3449
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.130018799944253e-05,
+      "loss": 1.1284,
+      "step": 3450
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1299683511311383e-05,
+      "loss": 1.2148,
+      "step": 3451
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1299178848003546e-05,
+      "loss": 1.1619,
+      "step": 3452
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1298674009528523e-05,
+      "loss": 1.1007,
+      "step": 3453
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1298168995895818e-05,
+      "loss": 1.1204,
+      "step": 3454
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1297663807114944e-05,
+      "loss": 1.0825,
+      "step": 3455
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1297158443195422e-05,
+      "loss": 1.149,
+      "step": 3456
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1296652904146774e-05,
+      "loss": 1.1769,
+      "step": 3457
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1296147189978512e-05,
+      "loss": 1.1777,
+      "step": 3458
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1295641300700175e-05,
+      "loss": 1.1951,
+      "step": 3459
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.129513523632128e-05,
+      "loss": 1.1254,
+      "step": 3460
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1294628996851365e-05,
+      "loss": 1.1761,
+      "step": 3461
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1294122582299966e-05,
+      "loss": 1.0398,
+      "step": 3462
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1293615992676627e-05,
+      "loss": 1.1706,
+      "step": 3463
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.129310922799088e-05,
+      "loss": 1.1687,
+      "step": 3464
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.129260228825228e-05,
+      "loss": 1.2601,
+      "step": 3465
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1292095173470374e-05,
+      "loss": 1.2029,
+      "step": 3466
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1291587883654714e-05,
+      "loss": 1.1109,
+      "step": 3467
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1291080418814853e-05,
+      "loss": 1.1757,
+      "step": 3468
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.129057277896035e-05,
+      "loss": 1.1058,
+      "step": 3469
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1290064964100774e-05,
+      "loss": 1.1464,
+      "step": 3470
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1289556974245685e-05,
+      "loss": 1.3237,
+      "step": 3471
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1289048809404654e-05,
+      "loss": 1.143,
+      "step": 3472
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.128854046958725e-05,
+      "loss": 1.1278,
+      "step": 3473
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.128803195480305e-05,
+      "loss": 1.1424,
+      "step": 3474
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1287523265061636e-05,
+      "loss": 1.1705,
+      "step": 3475
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1287014400372585e-05,
+      "loss": 1.1431,
+      "step": 3476
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.128650536074549e-05,
+      "loss": 1.2024,
+      "step": 3477
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1285996146189933e-05,
+      "loss": 1.0527,
+      "step": 3478
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1285486756715508e-05,
+      "loss": 1.1591,
+      "step": 3479
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1284977192331808e-05,
+      "loss": 1.1281,
+      "step": 3480
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1284467453048436e-05,
+      "loss": 1.0877,
+      "step": 3481
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.128395753887499e-05,
+      "loss": 1.1561,
+      "step": 3482
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1283447449821077e-05,
+      "loss": 1.188,
+      "step": 3483
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1282937185896302e-05,
+      "loss": 1.1547,
+      "step": 3484
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1282426747110286e-05,
+      "loss": 1.1617,
+      "step": 3485
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1281916133472637e-05,
+      "loss": 1.1352,
+      "step": 3486
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.128140534499297e-05,
+      "loss": 1.2025,
+      "step": 3487
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.128089438168092e-05,
+      "loss": 1.1322,
+      "step": 3488
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1280383243546093e-05,
+      "loss": 1.1804,
+      "step": 3489
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.127987193059813e-05,
+      "loss": 1.1668,
+      "step": 3490
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1279360442846662e-05,
+      "loss": 1.1247,
+      "step": 3491
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.127884878030132e-05,
+      "loss": 1.0896,
+      "step": 3492
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.127833694297175e-05,
+      "loss": 1.0833,
+      "step": 3493
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.127782493086758e-05,
+      "loss": 1.2218,
+      "step": 3494
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1277312743998464e-05,
+      "loss": 1.1476,
+      "step": 3495
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1276800382374043e-05,
+      "loss": 1.1658,
+      "step": 3496
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1276287846003978e-05,
+      "loss": 1.1108,
+      "step": 3497
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.127577513489792e-05,
+      "loss": 1.1697,
+      "step": 3498
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.127526224906552e-05,
+      "loss": 1.2079,
+      "step": 3499
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.127474918851645e-05,
+      "loss": 1.1175,
+      "step": 3500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.127423595326037e-05,
+      "loss": 1.1518,
+      "step": 3501
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1273722543306942e-05,
+      "loss": 1.1168,
+      "step": 3502
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1273208958665842e-05,
+      "loss": 1.2124,
+      "step": 3503
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1272695199346744e-05,
+      "loss": 1.0591,
+      "step": 3504
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.127218126535933e-05,
+      "loss": 1.1215,
+      "step": 3505
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1271667156713274e-05,
+      "loss": 1.1933,
+      "step": 3506
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1271152873418263e-05,
+      "loss": 1.1335,
+      "step": 3507
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1270638415483982e-05,
+      "loss": 1.0803,
+      "step": 3508
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1270123782920127e-05,
+      "loss": 1.1362,
+      "step": 3509
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.126960897573639e-05,
+      "loss": 1.1225,
+      "step": 3510
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1269093993942467e-05,
+      "loss": 1.1197,
+      "step": 3511
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1268578837548057e-05,
+      "loss": 1.1529,
+      "step": 3512
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1268063506562866e-05,
+      "loss": 1.1402,
+      "step": 3513
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1267548000996603e-05,
+      "loss": 1.1654,
+      "step": 3514
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1267032320858976e-05,
+      "loss": 1.1106,
+      "step": 3515
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1266516466159698e-05,
+      "loss": 1.1286,
+      "step": 3516
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1266000436908487e-05,
+      "loss": 1.0853,
+      "step": 3517
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1265484233115066e-05,
+      "loss": 1.2128,
+      "step": 3518
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1264967854789157e-05,
+      "loss": 1.1372,
+      "step": 3519
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.126445130194049e-05,
+      "loss": 1.2187,
+      "step": 3520
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1263934574578787e-05,
+      "loss": 1.1513,
+      "step": 3521
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1263417672713787e-05,
+      "loss": 1.1476,
+      "step": 3522
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1262900596355223e-05,
+      "loss": 1.1046,
+      "step": 3523
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1262383345512842e-05,
+      "loss": 1.15,
+      "step": 3524
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1261865920196382e-05,
+      "loss": 1.141,
+      "step": 3525
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1261348320415593e-05,
+      "loss": 1.1379,
+      "step": 3526
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.126083054618022e-05,
+      "loss": 1.1368,
+      "step": 3527
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1260312597500022e-05,
+      "loss": 1.1347,
+      "step": 3528
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1259794474384752e-05,
+      "loss": 1.1635,
+      "step": 3529
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.125927617684417e-05,
+      "loss": 1.1353,
+      "step": 3530
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.125875770488804e-05,
+      "loss": 1.0711,
+      "step": 3531
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.125823905852613e-05,
+      "loss": 1.2054,
+      "step": 3532
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.125772023776821e-05,
+      "loss": 1.0621,
+      "step": 3533
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1257201242624046e-05,
+      "loss": 1.1487,
+      "step": 3534
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.125668207310342e-05,
+      "loss": 1.1447,
+      "step": 3535
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.125616272921611e-05,
+      "loss": 1.1392,
+      "step": 3536
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.12556432109719e-05,
+      "loss": 1.1236,
+      "step": 3537
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1255123518380574e-05,
+      "loss": 1.1337,
+      "step": 3538
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1254603651451928e-05,
+      "loss": 1.185,
+      "step": 3539
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1254083610195746e-05,
+      "loss": 1.2367,
+      "step": 3540
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1253563394621826e-05,
+      "loss": 1.1294,
+      "step": 3541
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1253043004739968e-05,
+      "loss": 1.1756,
+      "step": 3542
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.125252244055998e-05,
+      "loss": 1.2186,
+      "step": 3543
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.125200170209166e-05,
+      "loss": 1.1023,
+      "step": 3544
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.125148078934482e-05,
+      "loss": 1.1715,
+      "step": 3545
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.125095970232927e-05,
+      "loss": 1.1439,
+      "step": 3546
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.125043844105483e-05,
+      "loss": 1.0922,
+      "step": 3547
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1249917005531316e-05,
+      "loss": 1.1524,
+      "step": 3548
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.124939539576855e-05,
+      "loss": 1.231,
+      "step": 3549
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.124887361177636e-05,
+      "loss": 1.1618,
+      "step": 3550
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1248351653564575e-05,
+      "loss": 1.1872,
+      "step": 3551
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1247829521143024e-05,
+      "loss": 1.1322,
+      "step": 3552
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1247307214521544e-05,
+      "loss": 1.1592,
+      "step": 3553
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1246784733709975e-05,
+      "loss": 1.1187,
+      "step": 3554
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1246262078718153e-05,
+      "loss": 1.2157,
+      "step": 3555
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.124573924955593e-05,
+      "loss": 1.1375,
+      "step": 3556
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.124521624623315e-05,
+      "loss": 1.1155,
+      "step": 3557
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.124469306875967e-05,
+      "loss": 1.233,
+      "step": 3558
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1244169717145338e-05,
+      "loss": 1.2275,
+      "step": 3559
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1243646191400014e-05,
+      "loss": 1.2164,
+      "step": 3560
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1243122491533565e-05,
+      "loss": 1.1803,
+      "step": 3561
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1242598617555852e-05,
+      "loss": 1.0696,
+      "step": 3562
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.124207456947674e-05,
+      "loss": 1.0717,
+      "step": 3563
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.124155034730611e-05,
+      "loss": 1.1383,
+      "step": 3564
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.124102595105383e-05,
+      "loss": 1.14,
+      "step": 3565
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.124050138072977e-05,
+      "loss": 1.0702,
+      "step": 3566
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1239976636343826e-05,
+      "loss": 1.1562,
+      "step": 3567
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.123945171790588e-05,
+      "loss": 1.2056,
+      "step": 3568
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.123892662542581e-05,
+      "loss": 1.1311,
+      "step": 3569
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1238401358913514e-05,
+      "loss": 1.1492,
+      "step": 3570
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1237875918378885e-05,
+      "loss": 1.1877,
+      "step": 3571
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1237350303831824e-05,
+      "loss": 1.2121,
+      "step": 3572
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1236824515282226e-05,
+      "loss": 1.0434,
+      "step": 3573
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.123629855274e-05,
+      "loss": 1.1616,
+      "step": 3574
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.123577241621505e-05,
+      "loss": 1.1036,
+      "step": 3575
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.123524610571729e-05,
+      "loss": 1.1129,
+      "step": 3576
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1234719621256634e-05,
+      "loss": 1.2299,
+      "step": 3577
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1234192962842997e-05,
+      "loss": 1.1104,
+      "step": 3578
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.12336661304863e-05,
+      "loss": 1.1859,
+      "step": 3579
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1233139124196465e-05,
+      "loss": 1.147,
+      "step": 3580
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.123261194398342e-05,
+      "loss": 1.1219,
+      "step": 3581
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1232084589857104e-05,
+      "loss": 1.1172,
+      "step": 3582
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.123155706182744e-05,
+      "loss": 1.1834,
+      "step": 3583
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1231029359904367e-05,
+      "loss": 1.1075,
+      "step": 3584
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1230501484097827e-05,
+      "loss": 1.1363,
+      "step": 3585
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1229973434417762e-05,
+      "loss": 1.1496,
+      "step": 3586
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1229445210874124e-05,
+      "loss": 1.1716,
+      "step": 3587
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1228916813476856e-05,
+      "loss": 1.1704,
+      "step": 3588
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1228388242235914e-05,
+      "loss": 1.1359,
+      "step": 3589
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1227859497161257e-05,
+      "loss": 1.1214,
+      "step": 3590
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1227330578262842e-05,
+      "loss": 1.2043,
+      "step": 3591
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1226801485550633e-05,
+      "loss": 1.0578,
+      "step": 3592
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1226272219034596e-05,
+      "loss": 1.155,
+      "step": 3593
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.12257427787247e-05,
+      "loss": 1.0885,
+      "step": 3594
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1225213164630926e-05,
+      "loss": 1.2053,
+      "step": 3595
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1224683376763237e-05,
+      "loss": 1.0999,
+      "step": 3596
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.122415341513162e-05,
+      "loss": 1.058,
+      "step": 3597
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1223623279746055e-05,
+      "loss": 1.1745,
+      "step": 3598
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1223092970616536e-05,
+      "loss": 1.0534,
+      "step": 3599
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1222562487753042e-05,
+      "loss": 1.0671,
+      "step": 3600
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.122203183116557e-05,
+      "loss": 1.1194,
+      "step": 3601
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1221501000864115e-05,
+      "loss": 1.169,
+      "step": 3602
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.122096999685868e-05,
+      "loss": 1.1773,
+      "step": 3603
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.122043881915926e-05,
+      "loss": 1.1043,
+      "step": 3604
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.121990746777587e-05,
+      "loss": 1.1375,
+      "step": 3605
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.121937594271851e-05,
+      "loss": 1.2348,
+      "step": 3606
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1218844243997196e-05,
+      "loss": 1.1507,
+      "step": 3607
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1218312371621946e-05,
+      "loss": 1.0864,
+      "step": 3608
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1217780325602776e-05,
+      "loss": 1.1503,
+      "step": 3609
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.121724810594971e-05,
+      "loss": 1.1278,
+      "step": 3610
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.121671571267277e-05,
+      "loss": 1.1094,
+      "step": 3611
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1216183145781985e-05,
+      "loss": 1.1431,
+      "step": 3612
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1215650405287394e-05,
+      "loss": 1.1421,
+      "step": 3613
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.121511749119902e-05,
+      "loss": 1.2171,
+      "step": 3614
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1214584403526914e-05,
+      "loss": 1.0658,
+      "step": 3615
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1214051142281108e-05,
+      "loss": 1.1939,
+      "step": 3616
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1213517707471654e-05,
+      "loss": 1.1627,
+      "step": 3617
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1212984099108596e-05,
+      "loss": 1.157,
+      "step": 3618
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1212450317201988e-05,
+      "loss": 1.1764,
+      "step": 3619
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.121191636176188e-05,
+      "loss": 1.1109,
+      "step": 3620
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.121138223279833e-05,
+      "loss": 1.1448,
+      "step": 3621
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1210847930321408e-05,
+      "loss": 1.2008,
+      "step": 3622
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1210313454341173e-05,
+      "loss": 1.1325,
+      "step": 3623
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1209778804867692e-05,
+      "loss": 1.0422,
+      "step": 3624
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1209243981911036e-05,
+      "loss": 1.0651,
+      "step": 3625
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.120870898548128e-05,
+      "loss": 1.1995,
+      "step": 3626
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1208173815588507e-05,
+      "loss": 1.0942,
+      "step": 3627
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1207638472242787e-05,
+      "loss": 1.154,
+      "step": 3628
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1207102955454216e-05,
+      "loss": 1.1333,
+      "step": 3629
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1206567265232868e-05,
+      "loss": 1.1506,
+      "step": 3630
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1206031401588845e-05,
+      "loss": 1.1694,
+      "step": 3631
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1205495364532242e-05,
+      "loss": 1.0945,
+      "step": 3632
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1204959154073144e-05,
+      "loss": 1.1278,
+      "step": 3633
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1204422770221665e-05,
+      "loss": 1.1617,
+      "step": 3634
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.12038862129879e-05,
+      "loss": 1.1061,
+      "step": 3635
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.120334948238196e-05,
+      "loss": 1.0796,
+      "step": 3636
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1202812578413954e-05,
+      "loss": 1.0744,
+      "step": 3637
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1202275501093995e-05,
+      "loss": 1.2152,
+      "step": 3638
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1201738250432204e-05,
+      "loss": 1.122,
+      "step": 3639
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1201200826438693e-05,
+      "loss": 1.1636,
+      "step": 3640
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1200663229123592e-05,
+      "loss": 1.108,
+      "step": 3641
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1200125458497026e-05,
+      "loss": 1.1052,
+      "step": 3642
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1199587514569124e-05,
+      "loss": 1.1041,
+      "step": 3643
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1199049397350022e-05,
+      "loss": 1.1037,
+      "step": 3644
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1198511106849856e-05,
+      "loss": 1.112,
+      "step": 3645
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.119797264307876e-05,
+      "loss": 1.2145,
+      "step": 3646
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1197434006046878e-05,
+      "loss": 1.1341,
+      "step": 3647
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1196895195764364e-05,
+      "loss": 1.1357,
+      "step": 3648
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1196356212241363e-05,
+      "loss": 1.114,
+      "step": 3649
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.119581705548802e-05,
+      "loss": 1.1471,
+      "step": 3650
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.119527772551451e-05,
+      "loss": 1.1134,
+      "step": 3651
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1194738222330973e-05,
+      "loss": 1.1192,
+      "step": 3652
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1194198545947584e-05,
+      "loss": 1.156,
+      "step": 3653
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.11936586963745e-05,
+      "loss": 1.0957,
+      "step": 3654
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1193118673621896e-05,
+      "loss": 1.1438,
+      "step": 3655
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.119257847769994e-05,
+      "loss": 1.2143,
+      "step": 3656
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1192038108618813e-05,
+      "loss": 1.0495,
+      "step": 3657
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1191497566388686e-05,
+      "loss": 1.1229,
+      "step": 3658
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1190956851019752e-05,
+      "loss": 1.1674,
+      "step": 3659
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1190415962522187e-05,
+      "loss": 1.0266,
+      "step": 3660
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1189874900906185e-05,
+      "loss": 1.154,
+      "step": 3661
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1189333666181934e-05,
+      "loss": 1.1673,
+      "step": 3662
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1188792258359632e-05,
+      "loss": 1.1474,
+      "step": 3663
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.118825067744948e-05,
+      "loss": 1.0807,
+      "step": 3664
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1187708923461668e-05,
+      "loss": 1.137,
+      "step": 3665
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1187166996406414e-05,
+      "loss": 1.0618,
+      "step": 3666
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1186624896293917e-05,
+      "loss": 1.0875,
+      "step": 3667
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.11860826231344e-05,
+      "loss": 1.1581,
+      "step": 3668
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1185540176938066e-05,
+      "loss": 1.0529,
+      "step": 3669
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.118499755771514e-05,
+      "loss": 1.1542,
+      "step": 3670
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1184454765475835e-05,
+      "loss": 1.1844,
+      "step": 3671
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1183911800230385e-05,
+      "loss": 1.0922,
+      "step": 3672
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1183368661989015e-05,
+      "loss": 1.0524,
+      "step": 3673
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1182825350761952e-05,
+      "loss": 1.0805,
+      "step": 3674
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1182281866559435e-05,
+      "loss": 1.0577,
+      "step": 3675
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1181738209391696e-05,
+      "loss": 1.0884,
+      "step": 3676
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1181194379268984e-05,
+      "loss": 1.1374,
+      "step": 3677
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1180650376201537e-05,
+      "loss": 1.129,
+      "step": 3678
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1180106200199602e-05,
+      "loss": 1.089,
+      "step": 3679
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1179561851273434e-05,
+      "loss": 1.1845,
+      "step": 3680
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1179017329433283e-05,
+      "loss": 1.2004,
+      "step": 3681
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1178472634689408e-05,
+      "loss": 1.1263,
+      "step": 3682
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1177927767052068e-05,
+      "loss": 1.1022,
+      "step": 3683
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.117738272653153e-05,
+      "loss": 1.0863,
+      "step": 3684
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.117683751313805e-05,
+      "loss": 1.1496,
+      "step": 3685
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.117629212688191e-05,
+      "loss": 1.034,
+      "step": 3686
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.117574656777338e-05,
+      "loss": 1.1002,
+      "step": 3687
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.117520083582274e-05,
+      "loss": 1.1204,
+      "step": 3688
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1174654931040262e-05,
+      "loss": 1.0939,
+      "step": 3689
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1174108853436235e-05,
+      "loss": 1.1273,
+      "step": 3690
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1173562603020942e-05,
+      "loss": 1.1025,
+      "step": 3691
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1173016179804674e-05,
+      "loss": 1.186,
+      "step": 3692
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1172469583797723e-05,
+      "loss": 1.0512,
+      "step": 3693
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.117192281501039e-05,
+      "loss": 1.1538,
+      "step": 3694
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1171375873452966e-05,
+      "loss": 1.1034,
+      "step": 3695
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.117082875913576e-05,
+      "loss": 1.0965,
+      "step": 3696
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1170281472069074e-05,
+      "loss": 1.1014,
+      "step": 3697
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1169734012263222e-05,
+      "loss": 1.15,
+      "step": 3698
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1169186379728515e-05,
+      "loss": 1.0565,
+      "step": 3699
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1168638574475265e-05,
+      "loss": 1.1587,
+      "step": 3700
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1168090596513792e-05,
+      "loss": 1.0702,
+      "step": 3701
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.116754244585442e-05,
+      "loss": 1.1589,
+      "step": 3702
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1166994122507476e-05,
+      "loss": 1.1486,
+      "step": 3703
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1166445626483285e-05,
+      "loss": 1.1074,
+      "step": 3704
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.116589695779218e-05,
+      "loss": 1.1043,
+      "step": 3705
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1165348116444496e-05,
+      "loss": 1.0751,
+      "step": 3706
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1164799102450575e-05,
+      "loss": 1.1309,
+      "step": 3707
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1164249915820754e-05,
+      "loss": 1.0647,
+      "step": 3708
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.116370055656538e-05,
+      "loss": 1.1371,
+      "step": 3709
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.11631510246948e-05,
+      "loss": 1.1664,
+      "step": 3710
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1162601320219365e-05,
+      "loss": 1.1325,
+      "step": 3711
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1162051443149435e-05,
+      "loss": 1.1109,
+      "step": 3712
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1161501393495363e-05,
+      "loss": 1.0852,
+      "step": 3713
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.116095117126751e-05,
+      "loss": 1.154,
+      "step": 3714
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1160400776476243e-05,
+      "loss": 1.1654,
+      "step": 3715
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1159850209131923e-05,
+      "loss": 1.2155,
+      "step": 3716
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1159299469244933e-05,
+      "loss": 1.0215,
+      "step": 3717
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.115874855682564e-05,
+      "loss": 1.1173,
+      "step": 3718
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.115819747188442e-05,
+      "loss": 1.1873,
+      "step": 3719
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1157646214431654e-05,
+      "loss": 1.1289,
+      "step": 3720
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.115709478447773e-05,
+      "loss": 1.1906,
+      "step": 3721
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.115654318203303e-05,
+      "loss": 1.0568,
+      "step": 3722
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.115599140710795e-05,
+      "loss": 1.1397,
+      "step": 3723
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.115543945971288e-05,
+      "loss": 1.0537,
+      "step": 3724
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.115488733985822e-05,
+      "loss": 1.0631,
+      "step": 3725
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1154335047554365e-05,
+      "loss": 1.1187,
+      "step": 3726
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1153782582811724e-05,
+      "loss": 1.1452,
+      "step": 3727
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1153229945640696e-05,
+      "loss": 1.1054,
+      "step": 3728
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1152677136051705e-05,
+      "loss": 1.2144,
+      "step": 3729
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.115212415405515e-05,
+      "loss": 1.1065,
+      "step": 3730
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.115157099966145e-05,
+      "loss": 1.1937,
+      "step": 3731
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1151017672881033e-05,
+      "loss": 1.2207,
+      "step": 3732
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1150464173724315e-05,
+      "loss": 1.1083,
+      "step": 3733
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.114991050220172e-05,
+      "loss": 1.0661,
+      "step": 3734
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1149356658323684e-05,
+      "loss": 1.1607,
+      "step": 3735
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.114880264210064e-05,
+      "loss": 1.1138,
+      "step": 3736
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1148248453543016e-05,
+      "loss": 1.1516,
+      "step": 3737
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1147694092661255e-05,
+      "loss": 1.149,
+      "step": 3738
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1147139559465803e-05,
+      "loss": 1.0733,
+      "step": 3739
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1146584853967103e-05,
+      "loss": 1.121,
+      "step": 3740
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1146029976175602e-05,
+      "loss": 1.1518,
+      "step": 3741
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1145474926101753e-05,
+      "loss": 1.1006,
+      "step": 3742
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1144919703756013e-05,
+      "loss": 1.2528,
+      "step": 3743
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1144364309148844e-05,
+      "loss": 1.1325,
+      "step": 3744
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.11438087422907e-05,
+      "loss": 1.0494,
+      "step": 3745
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1143253003192053e-05,
+      "loss": 1.1035,
+      "step": 3746
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1142697091863364e-05,
+      "loss": 1.1625,
+      "step": 3747
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.114214100831511e-05,
+      "loss": 1.0836,
+      "step": 3748
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1141584752557766e-05,
+      "loss": 1.1539,
+      "step": 3749
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.114102832460181e-05,
+      "loss": 1.1204,
+      "step": 3750
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1140471724457718e-05,
+      "loss": 1.1747,
+      "step": 3751
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1139914952135983e-05,
+      "loss": 1.1291,
+      "step": 3752
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1139358007647086e-05,
+      "loss": 1.193,
+      "step": 3753
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.113880089100152e-05,
+      "loss": 1.1547,
+      "step": 3754
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1138243602209785e-05,
+      "loss": 1.0957,
+      "step": 3755
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.113768614128237e-05,
+      "loss": 1.1702,
+      "step": 3756
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1137128508229778e-05,
+      "loss": 1.1995,
+      "step": 3757
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1136570703062518e-05,
+      "loss": 1.0422,
+      "step": 3758
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1136012725791092e-05,
+      "loss": 1.0937,
+      "step": 3759
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.113545457642601e-05,
+      "loss": 1.0872,
+      "step": 3760
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.113489625497779e-05,
+      "loss": 1.123,
+      "step": 3761
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.113433776145695e-05,
+      "loss": 1.0682,
+      "step": 3762
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1133779095874006e-05,
+      "loss": 1.1592,
+      "step": 3763
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1133220258239482e-05,
+      "loss": 1.09,
+      "step": 3764
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1132661248563904e-05,
+      "loss": 1.0987,
+      "step": 3765
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1132102066857804e-05,
+      "loss": 1.1577,
+      "step": 3766
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1131542713131716e-05,
+      "loss": 1.166,
+      "step": 3767
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1130983187396175e-05,
+      "loss": 1.1116,
+      "step": 3768
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.113042348966172e-05,
+      "loss": 1.1119,
+      "step": 3769
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1129863619938896e-05,
+      "loss": 1.1434,
+      "step": 3770
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1129303578238245e-05,
+      "loss": 1.0831,
+      "step": 3771
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1128743364570323e-05,
+      "loss": 1.1798,
+      "step": 3772
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1128182978945677e-05,
+      "loss": 1.1128,
+      "step": 3773
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1127622421374867e-05,
+      "loss": 1.1155,
+      "step": 3774
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1127061691868447e-05,
+      "loss": 1.1023,
+      "step": 3775
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1126500790436986e-05,
+      "loss": 1.093,
+      "step": 3776
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1125939717091046e-05,
+      "loss": 1.1261,
+      "step": 3777
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1125378471841197e-05,
+      "loss": 1.2081,
+      "step": 3778
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1124817054698004e-05,
+      "loss": 1.1777,
+      "step": 3779
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1124255465672054e-05,
+      "loss": 1.1351,
+      "step": 3780
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1123693704773916e-05,
+      "loss": 1.1843,
+      "step": 3781
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.112313177201418e-05,
+      "loss": 1.0453,
+      "step": 3782
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1122569667403424e-05,
+      "loss": 1.1606,
+      "step": 3783
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.112200739095224e-05,
+      "loss": 1.12,
+      "step": 3784
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1121444942671218e-05,
+      "loss": 1.085,
+      "step": 3785
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1120882322570953e-05,
+      "loss": 1.1654,
+      "step": 3786
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1120319530662046e-05,
+      "loss": 1.1711,
+      "step": 3787
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1119756566955093e-05,
+      "loss": 1.2216,
+      "step": 3788
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.11191934314607e-05,
+      "loss": 1.1412,
+      "step": 3789
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1118630124189477e-05,
+      "loss": 1.1367,
+      "step": 3790
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1118066645152033e-05,
+      "loss": 1.1166,
+      "step": 3791
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1117502994358985e-05,
+      "loss": 1.1502,
+      "step": 3792
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1116939171820946e-05,
+      "loss": 1.1571,
+      "step": 3793
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1116375177548543e-05,
+      "loss": 1.0642,
+      "step": 3794
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.111581101155239e-05,
+      "loss": 1.1335,
+      "step": 3795
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1115246673843122e-05,
+      "loss": 1.1292,
+      "step": 3796
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.111468216443137e-05,
+      "loss": 1.1288,
+      "step": 3797
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1114117483327762e-05,
+      "loss": 1.1329,
+      "step": 3798
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1113552630542936e-05,
+      "loss": 1.0714,
+      "step": 3799
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1112987606087534e-05,
+      "loss": 1.1873,
+      "step": 3800
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1112422409972205e-05,
+      "loss": 1.1986,
+      "step": 3801
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1111857042207583e-05,
+      "loss": 1.0118,
+      "step": 3802
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1111291502804325e-05,
+      "loss": 1.1424,
+      "step": 3803
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1110725791773086e-05,
+      "loss": 1.1524,
+      "step": 3804
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1110159909124516e-05,
+      "loss": 1.0925,
+      "step": 3805
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1109593854869282e-05,
+      "loss": 1.1109,
+      "step": 3806
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1109027629018035e-05,
+      "loss": 1.121,
+      "step": 3807
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1108461231581457e-05,
+      "loss": 1.0502,
+      "step": 3808
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1107894662570203e-05,
+      "loss": 1.1248,
+      "step": 3809
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.110732792199495e-05,
+      "loss": 1.1211,
+      "step": 3810
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1106761009866378e-05,
+      "loss": 1.1104,
+      "step": 3811
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.110619392619516e-05,
+      "loss": 1.2022,
+      "step": 3812
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1105626670991984e-05,
+      "loss": 1.1545,
+      "step": 3813
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1105059244267527e-05,
+      "loss": 1.1951,
+      "step": 3814
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1104491646032486e-05,
+      "loss": 1.1066,
+      "step": 3815
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1103923876297545e-05,
+      "loss": 1.1684,
+      "step": 3816
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1103355935073408e-05,
+      "loss": 1.1261,
+      "step": 3817
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1102787822370765e-05,
+      "loss": 1.1191,
+      "step": 3818
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.110221953820032e-05,
+      "loss": 1.084,
+      "step": 3819
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.110165108257278e-05,
+      "loss": 1.129,
+      "step": 3820
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.110108245549885e-05,
+      "loss": 1.1421,
+      "step": 3821
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1100513656989245e-05,
+      "loss": 1.1683,
+      "step": 3822
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1099944687054673e-05,
+      "loss": 1.0925,
+      "step": 3823
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1099375545705857e-05,
+      "loss": 1.0996,
+      "step": 3824
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1098806232953517e-05,
+      "loss": 1.1613,
+      "step": 3825
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1098236748808377e-05,
+      "loss": 1.2018,
+      "step": 3826
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.109766709328116e-05,
+      "loss": 1.1328,
+      "step": 3827
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.10970972663826e-05,
+      "loss": 1.0204,
+      "step": 3828
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1096527268123435e-05,
+      "loss": 1.2048,
+      "step": 3829
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1095957098514392e-05,
+      "loss": 1.1532,
+      "step": 3830
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1095386757566224e-05,
+      "loss": 1.1631,
+      "step": 3831
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1094816245289663e-05,
+      "loss": 1.1074,
+      "step": 3832
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1094245561695457e-05,
+      "loss": 1.1699,
+      "step": 3833
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1093674706794364e-05,
+      "loss": 1.0056,
+      "step": 3834
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.109310368059713e-05,
+      "loss": 1.1018,
+      "step": 3835
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1092532483114514e-05,
+      "loss": 1.157,
+      "step": 3836
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1091961114357275e-05,
+      "loss": 1.1609,
+      "step": 3837
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1091389574336175e-05,
+      "loss": 1.1115,
+      "step": 3838
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1090817863061983e-05,
+      "loss": 1.0594,
+      "step": 3839
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1090245980545462e-05,
+      "loss": 1.1419,
+      "step": 3840
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1089673926797388e-05,
+      "loss": 1.0836,
+      "step": 3841
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.108910170182854e-05,
+      "loss": 1.0741,
+      "step": 3842
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1088529305649693e-05,
+      "loss": 1.1708,
+      "step": 3843
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.108795673827163e-05,
+      "loss": 1.0807,
+      "step": 3844
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1087383999705138e-05,
+      "loss": 1.0918,
+      "step": 3845
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1086811089961e-05,
+      "loss": 1.0225,
+      "step": 3846
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1086238009050017e-05,
+      "loss": 1.1471,
+      "step": 3847
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1085664756982976e-05,
+      "loss": 1.13,
+      "step": 3848
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1085091333770678e-05,
+      "loss": 1.0998,
+      "step": 3849
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1084517739423924e-05,
+      "loss": 1.1035,
+      "step": 3850
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.108394397395352e-05,
+      "loss": 1.0844,
+      "step": 3851
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1083370037370277e-05,
+      "loss": 1.1148,
+      "step": 3852
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1082795929684996e-05,
+      "loss": 1.0761,
+      "step": 3853
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1082221650908506e-05,
+      "loss": 1.0592,
+      "step": 3854
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.108164720105161e-05,
+      "loss": 1.0756,
+      "step": 3855
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1081072580125138e-05,
+      "loss": 1.0887,
+      "step": 3856
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1080497788139906e-05,
+      "loss": 1.2013,
+      "step": 3857
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1079922825106752e-05,
+      "loss": 1.0338,
+      "step": 3858
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.10793476910365e-05,
+      "loss": 1.0901,
+      "step": 3859
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1078772385939984e-05,
+      "loss": 1.155,
+      "step": 3860
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.107819690982804e-05,
+      "loss": 1.1201,
+      "step": 3861
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1077621262711514e-05,
+      "loss": 1.0667,
+      "step": 3862
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1077045444601243e-05,
+      "loss": 1.1405,
+      "step": 3863
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1076469455508074e-05,
+      "loss": 1.1476,
+      "step": 3864
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1075893295442863e-05,
+      "loss": 1.1473,
+      "step": 3865
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1075316964416455e-05,
+      "loss": 1.1757,
+      "step": 3866
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1074740462439707e-05,
+      "loss": 1.126,
+      "step": 3867
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.107416378952349e-05,
+      "loss": 1.112,
+      "step": 3868
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.107358694567865e-05,
+      "loss": 1.1949,
+      "step": 3869
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1073009930916064e-05,
+      "loss": 1.0975,
+      "step": 3870
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.10724327452466e-05,
+      "loss": 1.1162,
+      "step": 3871
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1071855388681126e-05,
+      "loss": 1.1205,
+      "step": 3872
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1071277861230524e-05,
+      "loss": 1.0865,
+      "step": 3873
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1070700162905664e-05,
+      "loss": 1.1086,
+      "step": 3874
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1070122293717436e-05,
+      "loss": 1.0521,
+      "step": 3875
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.106954425367672e-05,
+      "loss": 1.0971,
+      "step": 3876
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1068966042794406e-05,
+      "loss": 1.1161,
+      "step": 3877
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.106838766108139e-05,
+      "loss": 1.1082,
+      "step": 3878
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1067809108548563e-05,
+      "loss": 1.1495,
+      "step": 3879
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.106723038520682e-05,
+      "loss": 1.1188,
+      "step": 3880
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1066651491067065e-05,
+      "loss": 1.1159,
+      "step": 3881
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1066072426140204e-05,
+      "loss": 1.0872,
+      "step": 3882
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1065493190437145e-05,
+      "loss": 1.1144,
+      "step": 3883
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1064913783968798e-05,
+      "loss": 1.1039,
+      "step": 3884
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1064334206746076e-05,
+      "loss": 1.091,
+      "step": 3885
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.10637544587799e-05,
+      "loss": 1.1275,
+      "step": 3886
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1063174540081184e-05,
+      "loss": 1.0876,
+      "step": 3887
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1062594450660858e-05,
+      "loss": 1.0751,
+      "step": 3888
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1062014190529847e-05,
+      "loss": 1.1069,
+      "step": 3889
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.106143375969908e-05,
+      "loss": 1.1024,
+      "step": 3890
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1060853158179495e-05,
+      "loss": 1.1138,
+      "step": 3891
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1060272385982024e-05,
+      "loss": 1.1072,
+      "step": 3892
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1059691443117605e-05,
+      "loss": 1.1033,
+      "step": 3893
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1059110329597186e-05,
+      "loss": 1.1783,
+      "step": 3894
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.105852904543171e-05,
+      "loss": 1.2196,
+      "step": 3895
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1057947590632133e-05,
+      "loss": 1.1451,
+      "step": 3896
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1057365965209402e-05,
+      "loss": 1.074,
+      "step": 3897
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1056784169174473e-05,
+      "loss": 1.1912,
+      "step": 3898
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1056202202538306e-05,
+      "loss": 1.1736,
+      "step": 3899
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.105562006531186e-05,
+      "loss": 1.0877,
+      "step": 3900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.105503775750611e-05,
+      "loss": 1.0897,
+      "step": 3901
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.105445527913202e-05,
+      "loss": 1.1681,
+      "step": 3902
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1053872630200558e-05,
+      "loss": 1.1085,
+      "step": 3903
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1053289810722702e-05,
+      "loss": 1.1715,
+      "step": 3904
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1052706820709432e-05,
+      "loss": 1.1427,
+      "step": 3905
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.105212366017173e-05,
+      "loss": 1.0618,
+      "step": 3906
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.105154032912058e-05,
+      "loss": 1.1449,
+      "step": 3907
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.105095682756697e-05,
+      "loss": 1.0779,
+      "step": 3908
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1050373155521887e-05,
+      "loss": 1.1247,
+      "step": 3909
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1049789312996332e-05,
+      "loss": 1.1639,
+      "step": 3910
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.10492053000013e-05,
+      "loss": 1.1319,
+      "step": 3911
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1048621116547794e-05,
+      "loss": 1.1691,
+      "step": 3912
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1048036762646813e-05,
+      "loss": 1.1505,
+      "step": 3913
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1047452238309372e-05,
+      "loss": 1.0977,
+      "step": 3914
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1046867543546473e-05,
+      "loss": 1.1027,
+      "step": 3915
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1046282678369137e-05,
+      "loss": 1.1,
+      "step": 3916
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1045697642788376e-05,
+      "loss": 1.1345,
+      "step": 3917
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.104511243681522e-05,
+      "loss": 1.1098,
+      "step": 3918
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.104452706046068e-05,
+      "loss": 1.1384,
+      "step": 3919
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.104394151373578e-05,
+      "loss": 1.0981,
+      "step": 3920
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1043355796651566e-05,
+      "loss": 1.1398,
+      "step": 3921
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.104276990921906e-05,
+      "loss": 1.1414,
+      "step": 3922
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1042183851449303e-05,
+      "loss": 1.1694,
+      "step": 3923
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.104159762335333e-05,
+      "loss": 1.0884,
+      "step": 3924
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1041011224942187e-05,
+      "loss": 1.1007,
+      "step": 3925
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.104042465622692e-05,
+      "loss": 1.182,
+      "step": 3926
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.103983791721857e-05,
+      "loss": 1.0868,
+      "step": 3927
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.10392510079282e-05,
+      "loss": 1.1582,
+      "step": 3928
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1038663928366865e-05,
+      "loss": 1.1181,
+      "step": 3929
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.103807667854562e-05,
+      "loss": 1.0383,
+      "step": 3930
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1037489258475526e-05,
+      "loss": 1.0721,
+      "step": 3931
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.103690166816765e-05,
+      "loss": 1.1776,
+      "step": 3932
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1036313907633058e-05,
+      "loss": 1.1719,
+      "step": 3933
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1035725976882828e-05,
+      "loss": 1.1309,
+      "step": 3934
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1035137875928025e-05,
+      "loss": 1.0393,
+      "step": 3935
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1034549604779733e-05,
+      "loss": 1.1387,
+      "step": 3936
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1033961163449033e-05,
+      "loss": 1.091,
+      "step": 3937
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1033372551947012e-05,
+      "loss": 1.1219,
+      "step": 3938
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.103278377028475e-05,
+      "loss": 0.9947,
+      "step": 3939
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1032194818473343e-05,
+      "loss": 1.1292,
+      "step": 3940
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1031605696523884e-05,
+      "loss": 1.1433,
+      "step": 3941
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.103101640444747e-05,
+      "loss": 1.1782,
+      "step": 3942
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1030426942255204e-05,
+      "loss": 1.0772,
+      "step": 3943
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1029837309958186e-05,
+      "loss": 1.0833,
+      "step": 3944
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1029247507567526e-05,
+      "loss": 1.1076,
+      "step": 3945
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1028657535094328e-05,
+      "loss": 1.1992,
+      "step": 3946
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1028067392549714e-05,
+      "loss": 1.1358,
+      "step": 3947
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.102747707994479e-05,
+      "loss": 1.176,
+      "step": 3948
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1026886597290685e-05,
+      "loss": 1.1615,
+      "step": 3949
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.102629594459852e-05,
+      "loss": 1.1048,
+      "step": 3950
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1025705121879416e-05,
+      "loss": 1.0453,
+      "step": 3951
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1025114129144506e-05,
+      "loss": 1.0569,
+      "step": 3952
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1024522966404923e-05,
+      "loss": 1.1398,
+      "step": 3953
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.10239316336718e-05,
+      "loss": 1.1316,
+      "step": 3954
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1023340130956283e-05,
+      "loss": 1.0743,
+      "step": 3955
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1022748458269503e-05,
+      "loss": 1.0955,
+      "step": 3956
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.102215661562261e-05,
+      "loss": 1.0626,
+      "step": 3957
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1021564603026757e-05,
+      "loss": 1.1896,
+      "step": 3958
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.102097242049309e-05,
+      "loss": 1.0514,
+      "step": 3959
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.102038006803276e-05,
+      "loss": 1.1528,
+      "step": 3960
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.101978754565694e-05,
+      "loss": 1.184,
+      "step": 3961
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1019194853376777e-05,
+      "loss": 1.1086,
+      "step": 3962
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1018601991203442e-05,
+      "loss": 1.1846,
+      "step": 3963
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1018008959148098e-05,
+      "loss": 1.1866,
+      "step": 3964
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1017415757221923e-05,
+      "loss": 1.1327,
+      "step": 3965
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1016822385436087e-05,
+      "loss": 1.1234,
+      "step": 3966
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1016228843801764e-05,
+      "loss": 1.0622,
+      "step": 3967
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.101563513233014e-05,
+      "loss": 1.1034,
+      "step": 3968
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1015041251032393e-05,
+      "loss": 1.0719,
+      "step": 3969
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1014447199919717e-05,
+      "loss": 1.1421,
+      "step": 3970
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1013852979003297e-05,
+      "loss": 1.1033,
+      "step": 3971
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1013258588294325e-05,
+      "loss": 1.075,
+      "step": 3972
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1012664027804007e-05,
+      "loss": 1.1351,
+      "step": 3973
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1012069297543528e-05,
+      "loss": 1.0504,
+      "step": 3974
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.10114743975241e-05,
+      "loss": 1.0867,
+      "step": 3975
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.101087932775693e-05,
+      "loss": 1.0759,
+      "step": 3976
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1010284088253224e-05,
+      "loss": 1.1026,
+      "step": 3977
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1009688679024193e-05,
+      "loss": 1.1156,
+      "step": 3978
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1009093100081055e-05,
+      "loss": 1.0643,
+      "step": 3979
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1008497351435033e-05,
+      "loss": 1.1309,
+      "step": 3980
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.100790143309734e-05,
+      "loss": 1.206,
+      "step": 3981
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.100730534507921e-05,
+      "loss": 1.0665,
+      "step": 3982
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1006709087391863e-05,
+      "loss": 1.063,
+      "step": 3983
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1006112660046536e-05,
+      "loss": 1.1302,
+      "step": 3984
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1005516063054462e-05,
+      "loss": 0.9796,
+      "step": 3985
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1004919296426883e-05,
+      "loss": 1.0341,
+      "step": 3986
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1004322360175033e-05,
+      "loss": 1.0999,
+      "step": 3987
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1003725254310162e-05,
+      "loss": 1.0755,
+      "step": 3988
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1003127978843515e-05,
+      "loss": 1.1015,
+      "step": 3989
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1002530533786343e-05,
+      "loss": 1.1345,
+      "step": 3990
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.10019329191499e-05,
+      "loss": 1.0602,
+      "step": 3991
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1001335134945448e-05,
+      "loss": 1.0714,
+      "step": 3992
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.100073718118424e-05,
+      "loss": 1.0995,
+      "step": 3993
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1000139057877544e-05,
+      "loss": 1.1313,
+      "step": 3994
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0999540765036625e-05,
+      "loss": 1.1224,
+      "step": 3995
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0998942302672754e-05,
+      "loss": 1.0976,
+      "step": 3996
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.09983436707972e-05,
+      "loss": 1.1301,
+      "step": 3997
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.099774486942125e-05,
+      "loss": 1.0883,
+      "step": 3998
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0997145898556168e-05,
+      "loss": 1.0511,
+      "step": 3999
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.099654675821325e-05,
+      "loss": 1.1515,
+      "step": 4000
+    },
+    {
+      "epoch": 0.17,
+      "eval_text_loss": 0.814426064491272,
+      "eval_text_runtime": 1.4139,
+      "eval_text_samples_per_second": 52.336,
+      "eval_text_steps_per_second": 1.414,
+      "step": 4000
+    },
+    {
+      "epoch": 0.17,
+      "eval_svit_loss": 1.1891210079193115,
+      "eval_svit_runtime": 4.8683,
+      "eval_svit_samples_per_second": 139.885,
+      "eval_svit_steps_per_second": 2.26,
+      "step": 4000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0995947448403777e-05,
+      "loss": 1.1269,
+      "step": 4001
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0995347969139036e-05,
+      "loss": 1.1072,
+      "step": 4002
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0994748320430323e-05,
+      "loss": 1.0669,
+      "step": 4003
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0994148502288933e-05,
+      "loss": 1.0929,
+      "step": 4004
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0993548514726163e-05,
+      "loss": 1.1085,
+      "step": 4005
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0992948357753315e-05,
+      "loss": 0.9914,
+      "step": 4006
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0992348031381693e-05,
+      "loss": 1.1149,
+      "step": 4007
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.099174753562261e-05,
+      "loss": 1.1195,
+      "step": 4008
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0991146870487373e-05,
+      "loss": 1.1254,
+      "step": 4009
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0990546035987297e-05,
+      "loss": 1.0751,
+      "step": 4010
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0989945032133703e-05,
+      "loss": 1.087,
+      "step": 4011
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0989343858937907e-05,
+      "loss": 1.0232,
+      "step": 4012
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0988742516411237e-05,
+      "loss": 1.0715,
+      "step": 4013
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.098814100456502e-05,
+      "loss": 1.1301,
+      "step": 4014
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0987539323410585e-05,
+      "loss": 1.0927,
+      "step": 4015
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0986937472959266e-05,
+      "loss": 1.0526,
+      "step": 4016
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0986335453222402e-05,
+      "loss": 1.093,
+      "step": 4017
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0985733264211333e-05,
+      "loss": 1.0598,
+      "step": 4018
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.09851309059374e-05,
+      "loss": 1.0825,
+      "step": 4019
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.098452837841195e-05,
+      "loss": 1.1265,
+      "step": 4020
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.098392568164633e-05,
+      "loss": 1.1445,
+      "step": 4021
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0983322815651904e-05,
+      "loss": 1.1264,
+      "step": 4022
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0982719780440017e-05,
+      "loss": 1.1232,
+      "step": 4023
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.098211657602203e-05,
+      "loss": 1.0688,
+      "step": 4024
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.098151320240931e-05,
+      "loss": 1.1283,
+      "step": 4025
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.098090965961322e-05,
+      "loss": 1.0271,
+      "step": 4026
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0980305947645123e-05,
+      "loss": 1.138,
+      "step": 4027
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0979702066516404e-05,
+      "loss": 1.1404,
+      "step": 4028
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0979098016238426e-05,
+      "loss": 1.0676,
+      "step": 4029
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0978493796822572e-05,
+      "loss": 1.0683,
+      "step": 4030
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0977889408280226e-05,
+      "loss": 1.1538,
+      "step": 4031
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.097728485062277e-05,
+      "loss": 1.117,
+      "step": 4032
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0976680123861595e-05,
+      "loss": 1.2303,
+      "step": 4033
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.097607522800809e-05,
+      "loss": 1.1384,
+      "step": 4034
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0975470163073647e-05,
+      "loss": 1.2178,
+      "step": 4035
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0974864929069663e-05,
+      "loss": 1.0967,
+      "step": 4036
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0974259526007543e-05,
+      "loss": 1.1295,
+      "step": 4037
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.097365395389869e-05,
+      "loss": 1.0822,
+      "step": 4038
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0973048212754512e-05,
+      "loss": 1.079,
+      "step": 4039
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0972442302586414e-05,
+      "loss": 1.037,
+      "step": 4040
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0971836223405817e-05,
+      "loss": 1.1552,
+      "step": 4041
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.097122997522413e-05,
+      "loss": 1.1391,
+      "step": 4042
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.097062355805278e-05,
+      "loss": 1.0224,
+      "step": 4043
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0970016971903183e-05,
+      "loss": 1.0137,
+      "step": 4044
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.096941021678677e-05,
+      "loss": 1.0892,
+      "step": 4045
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0968803292714966e-05,
+      "loss": 1.126,
+      "step": 4046
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.096819619969921e-05,
+      "loss": 1.1001,
+      "step": 4047
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0967588937750934e-05,
+      "loss": 1.0978,
+      "step": 4048
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0966981506881574e-05,
+      "loss": 1.0876,
+      "step": 4049
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0966373907102578e-05,
+      "loss": 1.1761,
+      "step": 4050
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.096576613842539e-05,
+      "loss": 1.0824,
+      "step": 4051
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.096515820086145e-05,
+      "loss": 1.1463,
+      "step": 4052
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0964550094422224e-05,
+      "loss": 1.0642,
+      "step": 4053
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0963941819119156e-05,
+      "loss": 1.1031,
+      "step": 4054
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0963333374963705e-05,
+      "loss": 1.1308,
+      "step": 4055
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.096272476196734e-05,
+      "loss": 1.2138,
+      "step": 4056
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0962115980141513e-05,
+      "loss": 1.0533,
+      "step": 4057
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0961507029497704e-05,
+      "loss": 1.148,
+      "step": 4058
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0960897910047373e-05,
+      "loss": 1.0684,
+      "step": 4059
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0960288621802005e-05,
+      "loss": 1.0572,
+      "step": 4060
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0959679164773066e-05,
+      "loss": 1.099,
+      "step": 4061
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0959069538972044e-05,
+      "loss": 1.1315,
+      "step": 4062
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.095845974441042e-05,
+      "loss": 1.0791,
+      "step": 4063
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0957849781099678e-05,
+      "loss": 1.0627,
+      "step": 4064
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.095723964905131e-05,
+      "loss": 1.0989,
+      "step": 4065
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0956629348276808e-05,
+      "loss": 1.0832,
+      "step": 4066
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.095601887878767e-05,
+      "loss": 1.1664,
+      "step": 4067
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0955408240595397e-05,
+      "loss": 1.1375,
+      "step": 4068
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0954797433711487e-05,
+      "loss": 1.0634,
+      "step": 4069
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0954186458147448e-05,
+      "loss": 1.1816,
+      "step": 4070
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0953575313914788e-05,
+      "loss": 1.165,
+      "step": 4071
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.095296400102502e-05,
+      "loss": 1.1216,
+      "step": 4072
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.095235251948966e-05,
+      "loss": 1.0191,
+      "step": 4073
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.095174086932022e-05,
+      "loss": 1.169,
+      "step": 4074
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0951129050528232e-05,
+      "loss": 1.1523,
+      "step": 4075
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0950517063125214e-05,
+      "loss": 1.0839,
+      "step": 4076
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0949904907122693e-05,
+      "loss": 1.0957,
+      "step": 4077
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0949292582532203e-05,
+      "loss": 1.1251,
+      "step": 4078
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.094868008936528e-05,
+      "loss": 1.0494,
+      "step": 4079
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0948067427633457e-05,
+      "loss": 1.1106,
+      "step": 4080
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0947454597348278e-05,
+      "loss": 1.1823,
+      "step": 4081
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0946841598521286e-05,
+      "loss": 1.1562,
+      "step": 4082
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0946228431164025e-05,
+      "loss": 1.0678,
+      "step": 4083
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0945615095288048e-05,
+      "loss": 1.0491,
+      "step": 4084
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0945001590904912e-05,
+      "loss": 0.9925,
+      "step": 4085
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0944387918026163e-05,
+      "loss": 1.064,
+      "step": 4086
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0943774076663373e-05,
+      "loss": 1.0564,
+      "step": 4087
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0943160066828094e-05,
+      "loss": 1.1757,
+      "step": 4088
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0942545888531898e-05,
+      "loss": 1.132,
+      "step": 4089
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0941931541786354e-05,
+      "loss": 1.0718,
+      "step": 4090
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0941317026603034e-05,
+      "loss": 1.1824,
+      "step": 4091
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0940702342993513e-05,
+      "loss": 1.0355,
+      "step": 4092
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.094008749096937e-05,
+      "loss": 1.1344,
+      "step": 4093
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0939472470542186e-05,
+      "loss": 1.2154,
+      "step": 4094
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0938857281723544e-05,
+      "loss": 1.1448,
+      "step": 4095
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0938241924525038e-05,
+      "loss": 0.985,
+      "step": 4096
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0937626398958256e-05,
+      "loss": 1.049,
+      "step": 4097
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.093701070503479e-05,
+      "loss": 1.0505,
+      "step": 4098
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.093639484276624e-05,
+      "loss": 1.112,
+      "step": 4099
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0935778812164213e-05,
+      "loss": 1.1751,
+      "step": 4100
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0935162613240302e-05,
+      "loss": 1.0408,
+      "step": 4101
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0934546246006124e-05,
+      "loss": 1.1318,
+      "step": 4102
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.093392971047328e-05,
+      "loss": 1.0609,
+      "step": 4103
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0933313006653394e-05,
+      "loss": 1.1072,
+      "step": 4104
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0932696134558075e-05,
+      "loss": 1.0689,
+      "step": 4105
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0932079094198946e-05,
+      "loss": 1.1333,
+      "step": 4106
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0931461885587628e-05,
+      "loss": 1.0764,
+      "step": 4107
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0930844508735747e-05,
+      "loss": 1.1214,
+      "step": 4108
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0930226963654936e-05,
+      "loss": 1.1885,
+      "step": 4109
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.092960925035683e-05,
+      "loss": 1.0863,
+      "step": 4110
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0928991368853054e-05,
+      "loss": 1.0836,
+      "step": 4111
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0928373319155256e-05,
+      "loss": 1.0585,
+      "step": 4112
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0927755101275078e-05,
+      "loss": 1.2879,
+      "step": 4113
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.092713671522416e-05,
+      "loss": 1.0472,
+      "step": 4114
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0926518161014156e-05,
+      "loss": 1.1,
+      "step": 4115
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.092589943865671e-05,
+      "loss": 1.123,
+      "step": 4116
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0925280548163487e-05,
+      "loss": 1.0833,
+      "step": 4117
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0924661489546137e-05,
+      "loss": 1.1435,
+      "step": 4118
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0924042262816322e-05,
+      "loss": 1.0722,
+      "step": 4119
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.092342286798571e-05,
+      "loss": 1.0953,
+      "step": 4120
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0922803305065964e-05,
+      "loss": 1.1177,
+      "step": 4121
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.092218357406876e-05,
+      "loss": 1.1463,
+      "step": 4122
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.092156367500577e-05,
+      "loss": 1.0852,
+      "step": 4123
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0920943607888667e-05,
+      "loss": 1.0489,
+      "step": 4124
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0920323372729134e-05,
+      "loss": 1.1525,
+      "step": 4125
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0919702969538856e-05,
+      "loss": 1.1907,
+      "step": 4126
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0919082398329514e-05,
+      "loss": 1.1564,
+      "step": 4127
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0918461659112806e-05,
+      "loss": 1.1302,
+      "step": 4128
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0917840751900415e-05,
+      "loss": 1.1609,
+      "step": 4129
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0917219676704046e-05,
+      "loss": 1.1224,
+      "step": 4130
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.091659843353539e-05,
+      "loss": 1.1386,
+      "step": 4131
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0915977022406157e-05,
+      "loss": 1.1153,
+      "step": 4132
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0915355443328046e-05,
+      "loss": 1.1298,
+      "step": 4133
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.091473369631277e-05,
+      "loss": 1.0536,
+      "step": 4134
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.091411178137204e-05,
+      "loss": 1.1336,
+      "step": 4135
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0913489698517567e-05,
+      "loss": 1.0784,
+      "step": 4136
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0912867447761075e-05,
+      "loss": 1.1454,
+      "step": 4137
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.091224502911428e-05,
+      "loss": 1.1928,
+      "step": 4138
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.091162244258891e-05,
+      "loss": 1.0879,
+      "step": 4139
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.091099968819669e-05,
+      "loss": 1.0833,
+      "step": 4140
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0910376765949353e-05,
+      "loss": 1.0978,
+      "step": 4141
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.090975367585863e-05,
+      "loss": 1.1126,
+      "step": 4142
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0909130417936263e-05,
+      "loss": 1.1738,
+      "step": 4143
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0908506992193986e-05,
+      "loss": 1.1172,
+      "step": 4144
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0907883398643548e-05,
+      "loss": 1.1344,
+      "step": 4145
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.090725963729669e-05,
+      "loss": 1.1164,
+      "step": 4146
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0906635708165168e-05,
+      "loss": 1.0668,
+      "step": 4147
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.090601161126073e-05,
+      "loss": 1.0965,
+      "step": 4148
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0905387346595134e-05,
+      "loss": 1.1865,
+      "step": 4149
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.090476291418014e-05,
+      "loss": 1.1236,
+      "step": 4150
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0904138314027507e-05,
+      "loss": 1.1236,
+      "step": 4151
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0903513546149002e-05,
+      "loss": 1.0541,
+      "step": 4152
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0902888610556395e-05,
+      "loss": 1.1413,
+      "step": 4153
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0902263507261456e-05,
+      "loss": 1.1296,
+      "step": 4154
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0901638236275965e-05,
+      "loss": 1.159,
+      "step": 4155
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0901012797611693e-05,
+      "loss": 1.1014,
+      "step": 4156
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0900387191280424e-05,
+      "loss": 1.0728,
+      "step": 4157
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0899761417293945e-05,
+      "loss": 1.063,
+      "step": 4158
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.089913547566404e-05,
+      "loss": 1.1332,
+      "step": 4159
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.08985093664025e-05,
+      "loss": 1.0833,
+      "step": 4160
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0897883089521122e-05,
+      "loss": 1.1465,
+      "step": 4161
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.08972566450317e-05,
+      "loss": 1.2024,
+      "step": 4162
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0896630032946035e-05,
+      "loss": 1.0656,
+      "step": 4163
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0896003253275936e-05,
+      "loss": 1.1344,
+      "step": 4164
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.08953763060332e-05,
+      "loss": 1.0519,
+      "step": 4165
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.089474919122964e-05,
+      "loss": 1.0424,
+      "step": 4166
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0894121908877074e-05,
+      "loss": 1.1514,
+      "step": 4167
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0893494458987313e-05,
+      "loss": 1.1272,
+      "step": 4168
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0892866841572175e-05,
+      "loss": 1.0301,
+      "step": 4169
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0892239056643488e-05,
+      "loss": 1.163,
+      "step": 4170
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.089161110421307e-05,
+      "loss": 1.139,
+      "step": 4171
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0890982984292757e-05,
+      "loss": 1.0231,
+      "step": 4172
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0890354696894375e-05,
+      "loss": 1.0896,
+      "step": 4173
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0889726242029763e-05,
+      "loss": 1.1127,
+      "step": 4174
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.088909761971076e-05,
+      "loss": 1.1151,
+      "step": 4175
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.08884688299492e-05,
+      "loss": 1.1687,
+      "step": 4176
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0887839872756935e-05,
+      "loss": 1.0898,
+      "step": 4177
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0887210748145808e-05,
+      "loss": 1.0556,
+      "step": 4178
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0886581456127675e-05,
+      "loss": 1.0833,
+      "step": 4179
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0885951996714385e-05,
+      "loss": 1.0573,
+      "step": 4180
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0885322369917792e-05,
+      "loss": 1.1069,
+      "step": 4181
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0884692575749763e-05,
+      "loss": 1.097,
+      "step": 4182
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.088406261422216e-05,
+      "loss": 1.0746,
+      "step": 4183
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.088343248534685e-05,
+      "loss": 1.0679,
+      "step": 4184
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0882802189135696e-05,
+      "loss": 1.1952,
+      "step": 4185
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.088217172560058e-05,
+      "loss": 1.0946,
+      "step": 4186
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0881541094753377e-05,
+      "loss": 1.1109,
+      "step": 4187
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0880910296605957e-05,
+      "loss": 1.058,
+      "step": 4188
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.088027933117021e-05,
+      "loss": 1.0217,
+      "step": 4189
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.087964819845802e-05,
+      "loss": 1.1667,
+      "step": 4190
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0879016898481273e-05,
+      "loss": 1.077,
+      "step": 4191
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0878385431251865e-05,
+      "loss": 1.1554,
+      "step": 4192
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0877753796781688e-05,
+      "loss": 1.0792,
+      "step": 4193
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.087712199508264e-05,
+      "loss": 1.0037,
+      "step": 4194
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0876490026166625e-05,
+      "loss": 1.0727,
+      "step": 4195
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0875857890045545e-05,
+      "loss": 1.0971,
+      "step": 4196
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0875225586731304e-05,
+      "loss": 1.1711,
+      "step": 4197
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.087459311623582e-05,
+      "loss": 1.0879,
+      "step": 4198
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0873960478571e-05,
+      "loss": 1.0732,
+      "step": 4199
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0873327673748766e-05,
+      "loss": 1.1568,
+      "step": 4200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.087269470178103e-05,
+      "loss": 1.1091,
+      "step": 4201
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.087206156267973e-05,
+      "loss": 1.0904,
+      "step": 4202
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0871428256456777e-05,
+      "loss": 1.0986,
+      "step": 4203
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0870794783124112e-05,
+      "loss": 1.1099,
+      "step": 4204
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.087016114269366e-05,
+      "loss": 1.1576,
+      "step": 4205
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0869527335177356e-05,
+      "loss": 1.1887,
+      "step": 4206
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0868893360587144e-05,
+      "loss": 1.1623,
+      "step": 4207
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.086825921893497e-05,
+      "loss": 1.0655,
+      "step": 4208
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0867624910232764e-05,
+      "loss": 1.1368,
+      "step": 4209
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.086699043449249e-05,
+      "loss": 1.1694,
+      "step": 4210
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0866355791726094e-05,
+      "loss": 1.1357,
+      "step": 4211
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.086572098194553e-05,
+      "loss": 1.0454,
+      "step": 4212
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0865086005162753e-05,
+      "loss": 1.1411,
+      "step": 4213
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.086445086138973e-05,
+      "loss": 1.1742,
+      "step": 4214
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0863815550638415e-05,
+      "loss": 1.0644,
+      "step": 4215
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0863180072920788e-05,
+      "loss": 1.0693,
+      "step": 4216
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0862544428248812e-05,
+      "loss": 1.036,
+      "step": 4217
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0861908616634466e-05,
+      "loss": 1.1352,
+      "step": 4218
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.086127263808972e-05,
+      "loss": 1.1166,
+      "step": 4219
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0860636492626556e-05,
+      "loss": 1.1902,
+      "step": 4220
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0860000180256956e-05,
+      "loss": 1.084,
+      "step": 4221
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.085936370099291e-05,
+      "loss": 1.0455,
+      "step": 4222
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0858727054846408e-05,
+      "loss": 1.0582,
+      "step": 4223
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0858090241829437e-05,
+      "loss": 1.1256,
+      "step": 4224
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.085745326195399e-05,
+      "loss": 1.1001,
+      "step": 4225
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0856816115232077e-05,
+      "loss": 1.2335,
+      "step": 4226
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0856178801675693e-05,
+      "loss": 1.2011,
+      "step": 4227
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0855541321296842e-05,
+      "loss": 1.1226,
+      "step": 4228
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.085490367410753e-05,
+      "loss": 1.1358,
+      "step": 4229
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0854265860119778e-05,
+      "loss": 1.0,
+      "step": 4230
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.085362787934559e-05,
+      "loss": 1.1,
+      "step": 4231
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.085298973179699e-05,
+      "loss": 1.119,
+      "step": 4232
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0852351417485995e-05,
+      "loss": 1.081,
+      "step": 4233
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0851712936424632e-05,
+      "loss": 1.0672,
+      "step": 4234
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0851074288624925e-05,
+      "loss": 1.1181,
+      "step": 4235
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0850435474098905e-05,
+      "loss": 1.1013,
+      "step": 4236
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.084979649285861e-05,
+      "loss": 1.0675,
+      "step": 4237
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0849157344916067e-05,
+      "loss": 1.1481,
+      "step": 4238
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0848518030283323e-05,
+      "loss": 1.0879,
+      "step": 4239
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0847878548972418e-05,
+      "loss": 1.1306,
+      "step": 4240
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0847238900995397e-05,
+      "loss": 1.1344,
+      "step": 4241
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0846599086364308e-05,
+      "loss": 1.1246,
+      "step": 4242
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.084595910509121e-05,
+      "loss": 1.1032,
+      "step": 4243
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.084531895718815e-05,
+      "loss": 1.0657,
+      "step": 4244
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.084467864266719e-05,
+      "loss": 1.0858,
+      "step": 4245
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.084403816154039e-05,
+      "loss": 1.0838,
+      "step": 4246
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.084339751381982e-05,
+      "loss": 1.1827,
+      "step": 4247
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.084275669951754e-05,
+      "loss": 1.1245,
+      "step": 4248
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.084211571864563e-05,
+      "loss": 1.1208,
+      "step": 4249
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0841474571216152e-05,
+      "loss": 1.1621,
+      "step": 4250
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.084083325724119e-05,
+      "loss": 1.121,
+      "step": 4251
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0840191776732824e-05,
+      "loss": 1.1075,
+      "step": 4252
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0839550129703143e-05,
+      "loss": 1.0563,
+      "step": 4253
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0838908316164222e-05,
+      "loss": 1.1792,
+      "step": 4254
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0838266336128163e-05,
+      "loss": 1.2173,
+      "step": 4255
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.083762418960705e-05,
+      "loss": 1.0655,
+      "step": 4256
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.083698187661298e-05,
+      "loss": 1.1304,
+      "step": 4257
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0836339397158054e-05,
+      "loss": 1.127,
+      "step": 4258
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0835696751254375e-05,
+      "loss": 1.0222,
+      "step": 4259
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.083505393891405e-05,
+      "loss": 1.1726,
+      "step": 4260
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0834410960149185e-05,
+      "loss": 1.0407,
+      "step": 4261
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.083376781497189e-05,
+      "loss": 1.1055,
+      "step": 4262
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.083312450339428e-05,
+      "loss": 1.0962,
+      "step": 4263
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.083248102542848e-05,
+      "loss": 1.0491,
+      "step": 4264
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.08318373810866e-05,
+      "loss": 1.0842,
+      "step": 4265
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0831193570380774e-05,
+      "loss": 1.0613,
+      "step": 4266
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0830549593323124e-05,
+      "loss": 1.1132,
+      "step": 4267
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.082990544992578e-05,
+      "loss": 1.1259,
+      "step": 4268
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.082926114020088e-05,
+      "loss": 1.2266,
+      "step": 4269
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.082861666416056e-05,
+      "loss": 1.1086,
+      "step": 4270
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0827972021816954e-05,
+      "loss": 1.0469,
+      "step": 4271
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.082732721318221e-05,
+      "loss": 1.1059,
+      "step": 4272
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0826682238268477e-05,
+      "loss": 1.1717,
+      "step": 4273
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0826037097087897e-05,
+      "loss": 1.0821,
+      "step": 4274
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.082539178965263e-05,
+      "loss": 1.1955,
+      "step": 4275
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0824746315974823e-05,
+      "loss": 1.1857,
+      "step": 4276
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.082410067606664e-05,
+      "loss": 1.0714,
+      "step": 4277
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0823454869940244e-05,
+      "loss": 1.2067,
+      "step": 4278
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.08228088976078e-05,
+      "loss": 1.1216,
+      "step": 4279
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.082216275908147e-05,
+      "loss": 1.087,
+      "step": 4280
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0821516454373434e-05,
+      "loss": 1.1619,
+      "step": 4281
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0820869983495855e-05,
+      "loss": 1.0953,
+      "step": 4282
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.082022334646092e-05,
+      "loss": 1.1427,
+      "step": 4283
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.081957654328081e-05,
+      "loss": 1.0916,
+      "step": 4284
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0818929573967703e-05,
+      "loss": 1.1565,
+      "step": 4285
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.081828243853379e-05,
+      "loss": 1.1228,
+      "step": 4286
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.081763513699126e-05,
+      "loss": 1.1506,
+      "step": 4287
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.08169876693523e-05,
+      "loss": 1.0462,
+      "step": 4288
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0816340035629115e-05,
+      "loss": 1.1138,
+      "step": 4289
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0815692235833904e-05,
+      "loss": 1.1443,
+      "step": 4290
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0815044269978864e-05,
+      "loss": 1.0697,
+      "step": 4291
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0814396138076204e-05,
+      "loss": 1.088,
+      "step": 4292
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.081374784013813e-05,
+      "loss": 1.1154,
+      "step": 4293
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.081309937617686e-05,
+      "loss": 1.1118,
+      "step": 4294
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0812450746204603e-05,
+      "loss": 1.1044,
+      "step": 4295
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0811801950233577e-05,
+      "loss": 1.079,
+      "step": 4296
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0811152988276007e-05,
+      "loss": 1.1467,
+      "step": 4297
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0810503860344116e-05,
+      "loss": 1.0568,
+      "step": 4298
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.080985456645013e-05,
+      "loss": 1.1137,
+      "step": 4299
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.080920510660628e-05,
+      "loss": 1.1809,
+      "step": 4300
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.08085554808248e-05,
+      "loss": 1.0959,
+      "step": 4301
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0807905689117933e-05,
+      "loss": 1.05,
+      "step": 4302
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.080725573149791e-05,
+      "loss": 1.1782,
+      "step": 4303
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0806605607976975e-05,
+      "loss": 1.0964,
+      "step": 4304
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.080595531856738e-05,
+      "loss": 1.1313,
+      "step": 4305
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0805304863281373e-05,
+      "loss": 1.0422,
+      "step": 4306
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0804654242131204e-05,
+      "loss": 1.2052,
+      "step": 4307
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.080400345512913e-05,
+      "loss": 1.1521,
+      "step": 4308
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0803352502287407e-05,
+      "loss": 1.0985,
+      "step": 4309
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.08027013836183e-05,
+      "loss": 1.1615,
+      "step": 4310
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0802050099134075e-05,
+      "loss": 1.0852,
+      "step": 4311
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0801398648847e-05,
+      "loss": 1.1138,
+      "step": 4312
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0800747032769344e-05,
+      "loss": 1.0876,
+      "step": 4313
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.080009525091338e-05,
+      "loss": 1.084,
+      "step": 4314
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0799443303291386e-05,
+      "loss": 1.1648,
+      "step": 4315
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0798791189915646e-05,
+      "loss": 1.132,
+      "step": 4316
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0798138910798442e-05,
+      "loss": 1.1793,
+      "step": 4317
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0797486465952062e-05,
+      "loss": 1.1813,
+      "step": 4318
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0796833855388793e-05,
+      "loss": 1.149,
+      "step": 4319
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.079618107912093e-05,
+      "loss": 1.0407,
+      "step": 4320
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.079552813716077e-05,
+      "loss": 1.0877,
+      "step": 4321
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0794875029520614e-05,
+      "loss": 1.0641,
+      "step": 4322
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.079422175621276e-05,
+      "loss": 1.1043,
+      "step": 4323
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0793568317249515e-05,
+      "loss": 1.1221,
+      "step": 4324
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.079291471264319e-05,
+      "loss": 1.1934,
+      "step": 4325
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0792260942406094e-05,
+      "loss": 1.174,
+      "step": 4326
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0791607006550545e-05,
+      "loss": 1.0855,
+      "step": 4327
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0790952905088857e-05,
+      "loss": 1.0708,
+      "step": 4328
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0790298638033355e-05,
+      "loss": 1.1967,
+      "step": 4329
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.078964420539636e-05,
+      "loss": 1.0937,
+      "step": 4330
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0788989607190205e-05,
+      "loss": 1.176,
+      "step": 4331
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0788334843427215e-05,
+      "loss": 1.1514,
+      "step": 4332
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0787679914119725e-05,
+      "loss": 1.1441,
+      "step": 4333
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0787024819280074e-05,
+      "loss": 1.157,
+      "step": 4334
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.07863695589206e-05,
+      "loss": 1.0904,
+      "step": 4335
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0785714133053645e-05,
+      "loss": 1.0321,
+      "step": 4336
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0785058541691553e-05,
+      "loss": 1.1447,
+      "step": 4337
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0784402784846684e-05,
+      "loss": 1.0495,
+      "step": 4338
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0783746862531378e-05,
+      "loss": 1.0917,
+      "step": 4339
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0783090774757994e-05,
+      "loss": 1.1234,
+      "step": 4340
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.07824345215389e-05,
+      "loss": 1.1192,
+      "step": 4341
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0781778102886444e-05,
+      "loss": 1.1052,
+      "step": 4342
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0781121518813e-05,
+      "loss": 1.1293,
+      "step": 4343
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0780464769330932e-05,
+      "loss": 1.0563,
+      "step": 4344
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.077980785445261e-05,
+      "loss": 1.1554,
+      "step": 4345
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0779150774190413e-05,
+      "loss": 1.1379,
+      "step": 4346
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0778493528556716e-05,
+      "loss": 1.1633,
+      "step": 4347
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0777836117563895e-05,
+      "loss": 1.0615,
+      "step": 4348
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.077717854122434e-05,
+      "loss": 1.0923,
+      "step": 4349
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0776520799550433e-05,
+      "loss": 1.1097,
+      "step": 4350
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0775862892554566e-05,
+      "loss": 1.0974,
+      "step": 4351
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0775204820249137e-05,
+      "loss": 1.0616,
+      "step": 4352
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0774546582646532e-05,
+      "loss": 1.1213,
+      "step": 4353
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0773888179759157e-05,
+      "loss": 1.0163,
+      "step": 4354
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0773229611599413e-05,
+      "loss": 1.0173,
+      "step": 4355
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.07725708781797e-05,
+      "loss": 1.1663,
+      "step": 4356
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0771911979512434e-05,
+      "loss": 1.156,
+      "step": 4357
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0771252915610023e-05,
+      "loss": 1.0814,
+      "step": 4358
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0770593686484884e-05,
+      "loss": 1.088,
+      "step": 4359
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.076993429214943e-05,
+      "loss": 1.1485,
+      "step": 4360
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0769274732616086e-05,
+      "loss": 1.1248,
+      "step": 4361
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0768615007897275e-05,
+      "loss": 1.0897,
+      "step": 4362
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0767955118005424e-05,
+      "loss": 1.0375,
+      "step": 4363
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0767295062952962e-05,
+      "loss": 1.1682,
+      "step": 4364
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.076663484275233e-05,
+      "loss": 1.1341,
+      "step": 4365
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0765974457415948e-05,
+      "loss": 1.1318,
+      "step": 4366
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.076531390695627e-05,
+      "loss": 1.1269,
+      "step": 4367
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.076465319138574e-05,
+      "loss": 1.0367,
+      "step": 4368
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0763992310716792e-05,
+      "loss": 1.1515,
+      "step": 4369
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0763331264961884e-05,
+      "loss": 1.0448,
+      "step": 4370
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0762670054133464e-05,
+      "loss": 1.0681,
+      "step": 4371
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0762008678243988e-05,
+      "loss": 1.067,
+      "step": 4372
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0761347137305914e-05,
+      "loss": 1.1173,
+      "step": 4373
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0760685431331703e-05,
+      "loss": 1.0155,
+      "step": 4374
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0760023560333824e-05,
+      "loss": 1.1619,
+      "step": 4375
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0759361524324738e-05,
+      "loss": 1.0856,
+      "step": 4376
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.075869932331692e-05,
+      "loss": 1.0728,
+      "step": 4377
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.075803695732284e-05,
+      "loss": 1.1982,
+      "step": 4378
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0757374426354977e-05,
+      "loss": 1.1546,
+      "step": 4379
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0756711730425812e-05,
+      "loss": 1.1108,
+      "step": 4380
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.075604886954783e-05,
+      "loss": 1.2021,
+      "step": 4381
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.075538584373351e-05,
+      "loss": 1.1173,
+      "step": 4382
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0754722652995348e-05,
+      "loss": 1.1113,
+      "step": 4383
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0754059297345837e-05,
+      "loss": 1.1104,
+      "step": 4384
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0753395776797463e-05,
+      "loss": 1.0976,
+      "step": 4385
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.075273209136274e-05,
+      "loss": 1.1339,
+      "step": 4386
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0752068241054152e-05,
+      "loss": 1.2121,
+      "step": 4387
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.075140422588422e-05,
+      "loss": 1.1281,
+      "step": 4388
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.075074004586544e-05,
+      "loss": 1.0277,
+      "step": 4389
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0750075701010333e-05,
+      "loss": 1.1112,
+      "step": 4390
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0749411191331402e-05,
+      "loss": 1.0378,
+      "step": 4391
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0748746516841174e-05,
+      "loss": 1.1269,
+      "step": 4392
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0748081677552168e-05,
+      "loss": 1.1046,
+      "step": 4393
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.07474166734769e-05,
+      "loss": 1.19,
+      "step": 4394
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0746751504627907e-05,
+      "loss": 1.0255,
+      "step": 4395
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.074608617101771e-05,
+      "loss": 1.1656,
+      "step": 4396
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0745420672658846e-05,
+      "loss": 1.1305,
+      "step": 4397
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.074475500956385e-05,
+      "loss": 1.1237,
+      "step": 4398
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0744089181745264e-05,
+      "loss": 0.9809,
+      "step": 4399
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0743423189215622e-05,
+      "loss": 1.137,
+      "step": 4400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0742757031987477e-05,
+      "loss": 1.1158,
+      "step": 4401
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0742090710073376e-05,
+      "loss": 1.1207,
+      "step": 4402
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.074142422348587e-05,
+      "loss": 1.096,
+      "step": 4403
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.074075757223751e-05,
+      "loss": 1.0491,
+      "step": 4404
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0740090756340856e-05,
+      "loss": 1.0569,
+      "step": 4405
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0739423775808467e-05,
+      "loss": 1.1108,
+      "step": 4406
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0738756630652914e-05,
+      "loss": 1.0871,
+      "step": 4407
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0738089320886757e-05,
+      "loss": 1.0653,
+      "step": 4408
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0737421846522566e-05,
+      "loss": 1.0917,
+      "step": 4409
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.073675420757292e-05,
+      "loss": 1.0717,
+      "step": 4410
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0736086404050384e-05,
+      "loss": 1.1798,
+      "step": 4411
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0735418435967547e-05,
+      "loss": 1.0573,
+      "step": 4412
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0734750303336993e-05,
+      "loss": 1.0344,
+      "step": 4413
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0734082006171302e-05,
+      "loss": 1.1178,
+      "step": 4414
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.073341354448306e-05,
+      "loss": 1.035,
+      "step": 4415
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0732744918284867e-05,
+      "loss": 1.109,
+      "step": 4416
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0732076127589307e-05,
+      "loss": 1.1071,
+      "step": 4417
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.073140717240899e-05,
+      "loss": 1.0425,
+      "step": 4418
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.073073805275651e-05,
+      "loss": 1.1564,
+      "step": 4419
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0730068768644472e-05,
+      "loss": 1.1479,
+      "step": 4420
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0729399320085486e-05,
+      "loss": 1.0758,
+      "step": 4421
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0728729707092157e-05,
+      "loss": 1.0419,
+      "step": 4422
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0728059929677098e-05,
+      "loss": 1.0929,
+      "step": 4423
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0727389987852934e-05,
+      "loss": 1.0705,
+      "step": 4424
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.072671988163228e-05,
+      "loss": 1.1389,
+      "step": 4425
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0726049611027754e-05,
+      "loss": 1.1621,
+      "step": 4426
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0725379176051985e-05,
+      "loss": 1.1246,
+      "step": 4427
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.072470857671761e-05,
+      "loss": 1.0795,
+      "step": 4428
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.072403781303725e-05,
+      "loss": 1.0713,
+      "step": 4429
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.072336688502354e-05,
+      "loss": 1.1578,
+      "step": 4430
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0722695792689123e-05,
+      "loss": 1.1281,
+      "step": 4431
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0722024536046645e-05,
+      "loss": 1.0517,
+      "step": 4432
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0721353115108738e-05,
+      "loss": 1.1609,
+      "step": 4433
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0720681529888058e-05,
+      "loss": 1.1508,
+      "step": 4434
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0720009780397254e-05,
+      "loss": 1.0765,
+      "step": 4435
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.071933786664898e-05,
+      "loss": 1.1245,
+      "step": 4436
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0718665788655894e-05,
+      "loss": 1.0867,
+      "step": 4437
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0717993546430652e-05,
+      "loss": 1.0796,
+      "step": 4438
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0717321139985916e-05,
+      "loss": 1.1102,
+      "step": 4439
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0716648569334356e-05,
+      "loss": 1.0615,
+      "step": 4440
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.071597583448864e-05,
+      "loss": 1.12,
+      "step": 4441
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.071530293546144e-05,
+      "loss": 1.114,
+      "step": 4442
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.071462987226543e-05,
+      "loss": 1.1321,
+      "step": 4443
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0713956644913296e-05,
+      "loss": 1.1223,
+      "step": 4444
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0713283253417706e-05,
+      "loss": 1.0501,
+      "step": 4445
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0712609697791353e-05,
+      "loss": 1.1638,
+      "step": 4446
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0711935978046924e-05,
+      "loss": 1.0359,
+      "step": 4447
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.071126209419711e-05,
+      "loss": 1.1359,
+      "step": 4448
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0710588046254604e-05,
+      "loss": 1.0652,
+      "step": 4449
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0709913834232103e-05,
+      "loss": 1.1186,
+      "step": 4450
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0709239458142308e-05,
+      "loss": 1.2065,
+      "step": 4451
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0708564917997918e-05,
+      "loss": 1.1672,
+      "step": 4452
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0707890213811646e-05,
+      "loss": 1.1526,
+      "step": 4453
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0707215345596193e-05,
+      "loss": 1.1487,
+      "step": 4454
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.070654031336428e-05,
+      "loss": 1.0428,
+      "step": 4455
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0705865117128616e-05,
+      "loss": 1.0925,
+      "step": 4456
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0705189756901922e-05,
+      "loss": 1.0693,
+      "step": 4457
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.070451423269692e-05,
+      "loss": 1.1242,
+      "step": 4458
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.070383854452634e-05,
+      "loss": 1.1289,
+      "step": 4459
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0703162692402896e-05,
+      "loss": 1.0576,
+      "step": 4460
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.070248667633933e-05,
+      "loss": 1.0856,
+      "step": 4461
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0701810496348374e-05,
+      "loss": 1.1519,
+      "step": 4462
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0701134152442766e-05,
+      "loss": 1.052,
+      "step": 4463
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.070045764463524e-05,
+      "loss": 1.118,
+      "step": 4464
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0699780972938547e-05,
+      "loss": 1.0796,
+      "step": 4465
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.069910413736543e-05,
+      "loss": 1.0734,
+      "step": 4466
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0698427137928638e-05,
+      "loss": 1.1231,
+      "step": 4467
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0697749974640925e-05,
+      "loss": 1.1093,
+      "step": 4468
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0697072647515043e-05,
+      "loss": 1.1033,
+      "step": 4469
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0696395156563754e-05,
+      "loss": 1.1329,
+      "step": 4470
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.069571750179982e-05,
+      "loss": 1.0617,
+      "step": 4471
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0695039683236007e-05,
+      "loss": 1.1412,
+      "step": 4472
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.069436170088508e-05,
+      "loss": 1.0656,
+      "step": 4473
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0693683554759812e-05,
+      "loss": 1.1162,
+      "step": 4474
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0693005244872973e-05,
+      "loss": 1.0877,
+      "step": 4475
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0692326771237345e-05,
+      "loss": 1.1145,
+      "step": 4476
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0691648133865705e-05,
+      "loss": 1.0848,
+      "step": 4477
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.069096933277084e-05,
+      "loss": 1.1226,
+      "step": 4478
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0690290367965536e-05,
+      "loss": 1.0907,
+      "step": 4479
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.068961123946258e-05,
+      "loss": 1.1347,
+      "step": 4480
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0688931947274766e-05,
+      "loss": 1.2154,
+      "step": 4481
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0688252491414887e-05,
+      "loss": 1.1177,
+      "step": 4482
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0687572871895748e-05,
+      "loss": 1.1042,
+      "step": 4483
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0686893088730145e-05,
+      "loss": 1.1276,
+      "step": 4484
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0686213141930885e-05,
+      "loss": 1.094,
+      "step": 4485
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0685533031510778e-05,
+      "loss": 1.0492,
+      "step": 4486
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.068485275748263e-05,
+      "loss": 1.1641,
+      "step": 4487
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.068417231985926e-05,
+      "loss": 1.1202,
+      "step": 4488
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0683491718653488e-05,
+      "loss": 1.1181,
+      "step": 4489
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0682810953878126e-05,
+      "loss": 1.0884,
+      "step": 4490
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0682130025546004e-05,
+      "loss": 1.0685,
+      "step": 4491
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0681448933669943e-05,
+      "loss": 1.0628,
+      "step": 4492
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0680767678262778e-05,
+      "loss": 1.1076,
+      "step": 4493
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.068008625933734e-05,
+      "loss": 1.0983,
+      "step": 4494
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0679404676906463e-05,
+      "loss": 1.1728,
+      "step": 4495
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0678722930982985e-05,
+      "loss": 1.1397,
+      "step": 4496
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0678041021579756e-05,
+      "loss": 1.087,
+      "step": 4497
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.067735894870961e-05,
+      "loss": 1.1378,
+      "step": 4498
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0676676712385404e-05,
+      "loss": 1.1365,
+      "step": 4499
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0675994312619983e-05,
+      "loss": 1.0644,
+      "step": 4500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.067531174942621e-05,
+      "loss": 1.0062,
+      "step": 4501
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0674629022816926e-05,
+      "loss": 1.1567,
+      "step": 4502
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0673946132805006e-05,
+      "loss": 1.0417,
+      "step": 4503
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.067326307940331e-05,
+      "loss": 1.0924,
+      "step": 4504
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.06725798626247e-05,
+      "loss": 1.0228,
+      "step": 4505
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0671896482482052e-05,
+      "loss": 1.0607,
+      "step": 4506
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.067121293898824e-05,
+      "loss": 1.1119,
+      "step": 4507
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.067052923215613e-05,
+      "loss": 1.0673,
+      "step": 4508
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0669845361998608e-05,
+      "loss": 1.1924,
+      "step": 4509
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0669161328528554e-05,
+      "loss": 1.1235,
+      "step": 4510
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0668477131758856e-05,
+      "loss": 1.0739,
+      "step": 4511
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0667792771702398e-05,
+      "loss": 1.1306,
+      "step": 4512
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0667108248372072e-05,
+      "loss": 1.1547,
+      "step": 4513
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0666423561780774e-05,
+      "loss": 1.0858,
+      "step": 4514
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.06657387119414e-05,
+      "loss": 1.0711,
+      "step": 4515
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0665053698866852e-05,
+      "loss": 1.1711,
+      "step": 4516
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0664368522570032e-05,
+      "loss": 1.1144,
+      "step": 4517
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0663683183063847e-05,
+      "loss": 1.1128,
+      "step": 4518
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0662997680361204e-05,
+      "loss": 1.096,
+      "step": 4519
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0662312014475024e-05,
+      "loss": 1.1331,
+      "step": 4520
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0661626185418212e-05,
+      "loss": 1.1084,
+      "step": 4521
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0660940193203696e-05,
+      "loss": 1.0763,
+      "step": 4522
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0660254037844392e-05,
+      "loss": 1.1162,
+      "step": 4523
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0659567719353224e-05,
+      "loss": 1.1305,
+      "step": 4524
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0658881237743125e-05,
+      "loss": 1.1408,
+      "step": 4525
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0658194593027026e-05,
+      "loss": 1.1071,
+      "step": 4526
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0657507785217854e-05,
+      "loss": 1.1853,
+      "step": 4527
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0656820814328558e-05,
+      "loss": 1.1159,
+      "step": 4528
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0656133680372067e-05,
+      "loss": 1.0834,
+      "step": 4529
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0655446383361333e-05,
+      "loss": 1.0403,
+      "step": 4530
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0654758923309293e-05,
+      "loss": 1.0915,
+      "step": 4531
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.065407130022891e-05,
+      "loss": 1.0597,
+      "step": 4532
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0653383514133124e-05,
+      "loss": 1.106,
+      "step": 4533
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.06526955650349e-05,
+      "loss": 1.0984,
+      "step": 4534
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.065200745294719e-05,
+      "loss": 1.0621,
+      "step": 4535
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.065131917788296e-05,
+      "loss": 1.1079,
+      "step": 4536
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.065063073985517e-05,
+      "loss": 1.1251,
+      "step": 4537
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0649942138876793e-05,
+      "loss": 1.188,
+      "step": 4538
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0649253374960798e-05,
+      "loss": 1.1029,
+      "step": 4539
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.064856444812016e-05,
+      "loss": 1.1143,
+      "step": 4540
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0647875358367857e-05,
+      "loss": 1.0954,
+      "step": 4541
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.064718610571687e-05,
+      "loss": 1.0641,
+      "step": 4542
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0646496690180178e-05,
+      "loss": 1.0829,
+      "step": 4543
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.064580711177077e-05,
+      "loss": 1.0698,
+      "step": 4544
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0645117370501633e-05,
+      "loss": 1.148,
+      "step": 4545
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0644427466385765e-05,
+      "loss": 1.1188,
+      "step": 4546
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.064373739943616e-05,
+      "loss": 1.1479,
+      "step": 4547
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0643047169665813e-05,
+      "loss": 1.1104,
+      "step": 4548
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0642356777087725e-05,
+      "loss": 1.0894,
+      "step": 4549
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0641666221714905e-05,
+      "loss": 1.0361,
+      "step": 4550
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.064097550356036e-05,
+      "loss": 1.1108,
+      "step": 4551
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0640284622637103e-05,
+      "loss": 1.1165,
+      "step": 4552
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0639593578958145e-05,
+      "loss": 1.0788,
+      "step": 4553
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0638902372536503e-05,
+      "loss": 1.1131,
+      "step": 4554
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.06382110033852e-05,
+      "loss": 1.1641,
+      "step": 4555
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0637519471517254e-05,
+      "loss": 1.1454,
+      "step": 4556
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0636827776945697e-05,
+      "loss": 1.1553,
+      "step": 4557
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0636135919683552e-05,
+      "loss": 1.0973,
+      "step": 4558
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0635443899743855e-05,
+      "loss": 1.065,
+      "step": 4559
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0634751717139646e-05,
+      "loss": 1.1609,
+      "step": 4560
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0634059371883958e-05,
+      "loss": 1.0697,
+      "step": 4561
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.063336686398983e-05,
+      "loss": 1.0545,
+      "step": 4562
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0632674193470313e-05,
+      "loss": 1.1693,
+      "step": 4563
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0631981360338454e-05,
+      "loss": 1.0901,
+      "step": 4564
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0631288364607298e-05,
+      "loss": 1.0752,
+      "step": 4565
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0630595206289906e-05,
+      "loss": 1.1248,
+      "step": 4566
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0629901885399328e-05,
+      "loss": 1.1412,
+      "step": 4567
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.062920840194863e-05,
+      "loss": 1.0864,
+      "step": 4568
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.062851475595087e-05,
+      "loss": 1.0996,
+      "step": 4569
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0627820947419122e-05,
+      "loss": 1.084,
+      "step": 4570
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0627126976366447e-05,
+      "loss": 1.1489,
+      "step": 4571
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.062643284280592e-05,
+      "loss": 1.1447,
+      "step": 4572
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0625738546750618e-05,
+      "loss": 1.1358,
+      "step": 4573
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0625044088213616e-05,
+      "loss": 1.1385,
+      "step": 4574
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0624349467207997e-05,
+      "loss": 1.1509,
+      "step": 4575
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0623654683746846e-05,
+      "loss": 1.127,
+      "step": 4576
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.062295973784325e-05,
+      "loss": 1.0605,
+      "step": 4577
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.06222646295103e-05,
+      "loss": 1.0716,
+      "step": 4578
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.062156935876109e-05,
+      "loss": 1.0967,
+      "step": 4579
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0620873925608718e-05,
+      "loss": 1.0167,
+      "step": 4580
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.062017833006628e-05,
+      "loss": 1.1364,
+      "step": 4581
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.061948257214688e-05,
+      "loss": 1.1358,
+      "step": 4582
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0618786651863625e-05,
+      "loss": 1.1239,
+      "step": 4583
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.061809056922962e-05,
+      "loss": 1.1305,
+      "step": 4584
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.061739432425799e-05,
+      "loss": 1.1538,
+      "step": 4585
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.061669791696183e-05,
+      "loss": 1.1257,
+      "step": 4586
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0616001347354276e-05,
+      "loss": 0.9744,
+      "step": 4587
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.061530461544844e-05,
+      "loss": 1.2618,
+      "step": 4588
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0614607721257445e-05,
+      "loss": 1.0421,
+      "step": 4589
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0613910664794425e-05,
+      "loss": 1.0942,
+      "step": 4590
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0613213446072507e-05,
+      "loss": 1.0638,
+      "step": 4591
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0612516065104823e-05,
+      "loss": 1.0725,
+      "step": 4592
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.061181852190451e-05,
+      "loss": 1.14,
+      "step": 4593
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.061112081648471e-05,
+      "loss": 1.1075,
+      "step": 4594
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0610422948858564e-05,
+      "loss": 1.0684,
+      "step": 4595
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0609724919039214e-05,
+      "loss": 0.9616,
+      "step": 4596
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0609026727039814e-05,
+      "loss": 1.1008,
+      "step": 4597
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0608328372873513e-05,
+      "loss": 1.1131,
+      "step": 4598
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0607629856553467e-05,
+      "loss": 1.0247,
+      "step": 4599
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0606931178092833e-05,
+      "loss": 1.0781,
+      "step": 4600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.060623233750477e-05,
+      "loss": 1.0833,
+      "step": 4601
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.060553333480245e-05,
+      "loss": 1.1239,
+      "step": 4602
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.060483416999903e-05,
+      "loss": 1.1272,
+      "step": 4603
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0604134843107686e-05,
+      "loss": 1.1239,
+      "step": 4604
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0603435354141588e-05,
+      "loss": 1.0647,
+      "step": 4605
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0602735703113915e-05,
+      "loss": 1.1074,
+      "step": 4606
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0602035890037846e-05,
+      "loss": 1.1382,
+      "step": 4607
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.060133591492656e-05,
+      "loss": 1.1266,
+      "step": 4608
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.060063577779324e-05,
+      "loss": 1.0815,
+      "step": 4609
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0599935478651083e-05,
+      "loss": 1.1187,
+      "step": 4610
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.059923501751328e-05,
+      "loss": 1.1401,
+      "step": 4611
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.059853439439302e-05,
+      "loss": 1.0079,
+      "step": 4612
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.05978336093035e-05,
+      "loss": 1.1105,
+      "step": 4613
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.059713266225792e-05,
+      "loss": 1.0935,
+      "step": 4614
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.059643155326949e-05,
+      "loss": 1.0352,
+      "step": 4615
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.059573028235141e-05,
+      "loss": 1.0761,
+      "step": 4616
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.05950288495169e-05,
+      "loss": 1.0611,
+      "step": 4617
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.059432725477916e-05,
+      "loss": 1.0826,
+      "step": 4618
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0593625498151414e-05,
+      "loss": 1.1713,
+      "step": 4619
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0592923579646876e-05,
+      "loss": 1.0394,
+      "step": 4620
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.059222149927877e-05,
+      "loss": 1.1115,
+      "step": 4621
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.059151925706032e-05,
+      "loss": 1.0481,
+      "step": 4622
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.059081685300476e-05,
+      "loss": 1.0928,
+      "step": 4623
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0590114287125314e-05,
+      "loss": 1.0877,
+      "step": 4624
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0589411559435216e-05,
+      "loss": 1.001,
+      "step": 4625
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.058870866994771e-05,
+      "loss": 1.0594,
+      "step": 4626
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.058800561867603e-05,
+      "loss": 1.049,
+      "step": 4627
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.058730240563342e-05,
+      "loss": 1.0329,
+      "step": 4628
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.058659903083313e-05,
+      "loss": 1.2537,
+      "step": 4629
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0585895494288403e-05,
+      "loss": 1.1307,
+      "step": 4630
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0585191796012496e-05,
+      "loss": 1.1298,
+      "step": 4631
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0584487936018665e-05,
+      "loss": 1.1424,
+      "step": 4632
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0583783914320165e-05,
+      "loss": 1.0793,
+      "step": 4633
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0583079730930265e-05,
+      "loss": 1.1383,
+      "step": 4634
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0582375385862216e-05,
+      "loss": 1.0479,
+      "step": 4635
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.05816708791293e-05,
+      "loss": 1.152,
+      "step": 4636
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.058096621074478e-05,
+      "loss": 1.0481,
+      "step": 4637
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0580261380721933e-05,
+      "loss": 1.1078,
+      "step": 4638
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0579556389074032e-05,
+      "loss": 1.0503,
+      "step": 4639
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0578851235814362e-05,
+      "loss": 1.0631,
+      "step": 4640
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0578145920956198e-05,
+      "loss": 1.1231,
+      "step": 4641
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.057744044451283e-05,
+      "loss": 1.0387,
+      "step": 4642
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0576734806497554e-05,
+      "loss": 1.0859,
+      "step": 4643
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0576029006923654e-05,
+      "loss": 1.1118,
+      "step": 4644
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0575323045804424e-05,
+      "loss": 1.1322,
+      "step": 4645
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0574616923153164e-05,
+      "loss": 1.1193,
+      "step": 4646
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0573910638983177e-05,
+      "loss": 0.9795,
+      "step": 4647
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0573204193307768e-05,
+      "loss": 1.1068,
+      "step": 4648
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0572497586140242e-05,
+      "loss": 1.0992,
+      "step": 4649
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.057179081749391e-05,
+      "loss": 1.1117,
+      "step": 4650
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0571083887382084e-05,
+      "loss": 1.0125,
+      "step": 4651
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.057037679581808e-05,
+      "loss": 0.9845,
+      "step": 4652
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0569669542815215e-05,
+      "loss": 1.0659,
+      "step": 4653
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.056896212838682e-05,
+      "loss": 1.0818,
+      "step": 4654
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0568254552546217e-05,
+      "loss": 1.1004,
+      "step": 4655
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0567546815306727e-05,
+      "loss": 1.1245,
+      "step": 4656
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0566838916681693e-05,
+      "loss": 1.0315,
+      "step": 4657
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.056613085668444e-05,
+      "loss": 0.9991,
+      "step": 4658
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0565422635328312e-05,
+      "loss": 1.1277,
+      "step": 4659
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0564714252626643e-05,
+      "loss": 1.0782,
+      "step": 4660
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0564005708592787e-05,
+      "loss": 1.0385,
+      "step": 4661
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.056329700324008e-05,
+      "loss": 1.0338,
+      "step": 4662
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.056258813658188e-05,
+      "loss": 1.0777,
+      "step": 4663
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0561879108631535e-05,
+      "loss": 1.0697,
+      "step": 4664
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.05611699194024e-05,
+      "loss": 1.1157,
+      "step": 4665
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0560460568907837e-05,
+      "loss": 1.0368,
+      "step": 4666
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.055975105716121e-05,
+      "loss": 1.0999,
+      "step": 4667
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0559041384175876e-05,
+      "loss": 1.1292,
+      "step": 4668
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0558331549965212e-05,
+      "loss": 1.1006,
+      "step": 4669
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0557621554542583e-05,
+      "loss": 1.0588,
+      "step": 4670
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0556911397921367e-05,
+      "loss": 1.0406,
+      "step": 4671
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0556201080114935e-05,
+      "loss": 1.0604,
+      "step": 4672
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0555490601136676e-05,
+      "loss": 1.1345,
+      "step": 4673
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0554779960999964e-05,
+      "loss": 1.0636,
+      "step": 4674
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.055406915971819e-05,
+      "loss": 1.1356,
+      "step": 4675
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0553358197304745e-05,
+      "loss": 1.2066,
+      "step": 4676
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.055264707377302e-05,
+      "loss": 1.1173,
+      "step": 4677
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.055193578913641e-05,
+      "loss": 1.0521,
+      "step": 4678
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0551224343408312e-05,
+      "loss": 1.1218,
+      "step": 4679
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0550512736602127e-05,
+      "loss": 1.0434,
+      "step": 4680
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.054980096873126e-05,
+      "loss": 1.0775,
+      "step": 4681
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.054908903980912e-05,
+      "loss": 1.1007,
+      "step": 4682
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.054837694984912e-05,
+      "loss": 1.1096,
+      "step": 4683
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0547664698864667e-05,
+      "loss": 1.1011,
+      "step": 4684
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0546952286869182e-05,
+      "loss": 1.0758,
+      "step": 4685
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0546239713876078e-05,
+      "loss": 1.039,
+      "step": 4686
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.054552697989879e-05,
+      "loss": 1.1054,
+      "step": 4687
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0544814084950732e-05,
+      "loss": 1.0768,
+      "step": 4688
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.054410102904534e-05,
+      "loss": 1.0647,
+      "step": 4689
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0543387812196042e-05,
+      "loss": 1.1507,
+      "step": 4690
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0542674434416274e-05,
+      "loss": 1.0952,
+      "step": 4691
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0541960895719474e-05,
+      "loss": 1.1599,
+      "step": 4692
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0541247196119082e-05,
+      "loss": 1.0186,
+      "step": 4693
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0540533335628543e-05,
+      "loss": 0.9892,
+      "step": 4694
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.05398193142613e-05,
+      "loss": 1.0757,
+      "step": 4695
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.053910513203081e-05,
+      "loss": 1.1037,
+      "step": 4696
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0538390788950518e-05,
+      "loss": 1.0765,
+      "step": 4697
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0537676285033887e-05,
+      "loss": 1.0633,
+      "step": 4698
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0536961620294374e-05,
+      "loss": 1.0897,
+      "step": 4699
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0536246794745438e-05,
+      "loss": 1.0354,
+      "step": 4700
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0535531808400545e-05,
+      "loss": 1.1261,
+      "step": 4701
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0534816661273166e-05,
+      "loss": 1.1295,
+      "step": 4702
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0534101353376772e-05,
+      "loss": 1.0459,
+      "step": 4703
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.053338588472483e-05,
+      "loss": 1.0122,
+      "step": 4704
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0532670255330827e-05,
+      "loss": 1.1184,
+      "step": 4705
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0531954465208235e-05,
+      "loss": 1.1156,
+      "step": 4706
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0531238514370547e-05,
+      "loss": 1.1327,
+      "step": 4707
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0530522402831237e-05,
+      "loss": 1.0128,
+      "step": 4708
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0529806130603804e-05,
+      "loss": 1.1502,
+      "step": 4709
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0529089697701736e-05,
+      "loss": 1.136,
+      "step": 4710
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.052837310413853e-05,
+      "loss": 1.1135,
+      "step": 4711
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.052765634992768e-05,
+      "loss": 1.1878,
+      "step": 4712
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0526939435082693e-05,
+      "loss": 1.1155,
+      "step": 4713
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0526222359617074e-05,
+      "loss": 1.1095,
+      "step": 4714
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0525505123544325e-05,
+      "loss": 1.1253,
+      "step": 4715
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0524787726877957e-05,
+      "loss": 1.1042,
+      "step": 4716
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0524070169631494e-05,
+      "loss": 1.0314,
+      "step": 4717
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0523352451818436e-05,
+      "loss": 1.1591,
+      "step": 4718
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0522634573452315e-05,
+      "loss": 1.0765,
+      "step": 4719
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0521916534546645e-05,
+      "loss": 1.0302,
+      "step": 4720
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0521198335114962e-05,
+      "loss": 1.1376,
+      "step": 4721
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0520479975170784e-05,
+      "loss": 1.0939,
+      "step": 4722
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.051976145472765e-05,
+      "loss": 1.1141,
+      "step": 4723
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0519042773799088e-05,
+      "loss": 1.1031,
+      "step": 4724
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.051832393239864e-05,
+      "loss": 1.1408,
+      "step": 4725
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.051760493053985e-05,
+      "loss": 1.0598,
+      "step": 4726
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0516885768236256e-05,
+      "loss": 1.0809,
+      "step": 4727
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0516166445501406e-05,
+      "loss": 1.2223,
+      "step": 4728
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0515446962348852e-05,
+      "loss": 1.1006,
+      "step": 4729
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.051472731879214e-05,
+      "loss": 1.0792,
+      "step": 4730
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0514007514844836e-05,
+      "loss": 1.0548,
+      "step": 4731
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0513287550520496e-05,
+      "loss": 1.107,
+      "step": 4732
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0512567425832676e-05,
+      "loss": 1.0501,
+      "step": 4733
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0511847140794946e-05,
+      "loss": 1.0582,
+      "step": 4734
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0511126695420872e-05,
+      "loss": 1.1023,
+      "step": 4735
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0510406089724027e-05,
+      "loss": 1.1019,
+      "step": 4736
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.050968532371798e-05,
+      "loss": 1.0744,
+      "step": 4737
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0508964397416313e-05,
+      "loss": 0.973,
+      "step": 4738
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0508243310832603e-05,
+      "loss": 1.0814,
+      "step": 4739
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0507522063980437e-05,
+      "loss": 1.1443,
+      "step": 4740
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0506800656873398e-05,
+      "loss": 1.0091,
+      "step": 4741
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0506079089525073e-05,
+      "loss": 1.0959,
+      "step": 4742
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.050535736194906e-05,
+      "loss": 1.0645,
+      "step": 4743
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.050463547415895e-05,
+      "loss": 1.0486,
+      "step": 4744
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0503913426168338e-05,
+      "loss": 1.071,
+      "step": 4745
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.050319121799083e-05,
+      "loss": 1.1368,
+      "step": 4746
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.050246884964003e-05,
+      "loss": 1.1722,
+      "step": 4747
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0501746321129543e-05,
+      "loss": 1.1155,
+      "step": 4748
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0501023632472978e-05,
+      "loss": 1.0326,
+      "step": 4749
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0500300783683957e-05,
+      "loss": 1.0554,
+      "step": 4750
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0499577774776083e-05,
+      "loss": 1.1007,
+      "step": 4751
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0498854605762983e-05,
+      "loss": 1.0457,
+      "step": 4752
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.049813127665828e-05,
+      "loss": 1.0974,
+      "step": 4753
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0497407787475598e-05,
+      "loss": 1.1598,
+      "step": 4754
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.049668413822856e-05,
+      "loss": 1.1189,
+      "step": 4755
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0495960328930807e-05,
+      "loss": 1.1258,
+      "step": 4756
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0495236359595964e-05,
+      "loss": 1.0982,
+      "step": 4757
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.049451223023767e-05,
+      "loss": 1.1342,
+      "step": 4758
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0493787940869573e-05,
+      "loss": 1.0847,
+      "step": 4759
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.049306349150531e-05,
+      "loss": 1.0822,
+      "step": 4760
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0492338882158527e-05,
+      "loss": 1.0697,
+      "step": 4761
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.049161411284287e-05,
+      "loss": 1.0772,
+      "step": 4762
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0490889183572003e-05,
+      "loss": 1.0419,
+      "step": 4763
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0490164094359573e-05,
+      "loss": 1.0612,
+      "step": 4764
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0489438845219238e-05,
+      "loss": 1.0778,
+      "step": 4765
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0488713436164662e-05,
+      "loss": 1.2493,
+      "step": 4766
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0487987867209505e-05,
+      "loss": 1.0572,
+      "step": 4767
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0487262138367442e-05,
+      "loss": 1.0258,
+      "step": 4768
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0486536249652136e-05,
+      "loss": 1.0891,
+      "step": 4769
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0485810201077267e-05,
+      "loss": 1.1613,
+      "step": 4770
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.048508399265651e-05,
+      "loss": 1.0277,
+      "step": 4771
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.048435762440354e-05,
+      "loss": 1.1332,
+      "step": 4772
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.048363109633204e-05,
+      "loss": 1.0703,
+      "step": 4773
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.04829044084557e-05,
+      "loss": 1.0121,
+      "step": 4774
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0482177560788207e-05,
+      "loss": 1.1736,
+      "step": 4775
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.048145055334325e-05,
+      "loss": 1.0178,
+      "step": 4776
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.048072338613453e-05,
+      "loss": 1.1038,
+      "step": 4777
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0479996059175734e-05,
+      "loss": 1.137,
+      "step": 4778
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.047926857248057e-05,
+      "loss": 1.0785,
+      "step": 4779
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0478540926062744e-05,
+      "loss": 1.1176,
+      "step": 4780
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0477813119935957e-05,
+      "loss": 1.1553,
+      "step": 4781
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0477085154113914e-05,
+      "loss": 1.1097,
+      "step": 4782
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.047635702861034e-05,
+      "loss": 1.0078,
+      "step": 4783
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0475628743438945e-05,
+      "loss": 1.2,
+      "step": 4784
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0474900298613445e-05,
+      "loss": 1.0971,
+      "step": 4785
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0474171694147558e-05,
+      "loss": 1.071,
+      "step": 4786
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.047344293005502e-05,
+      "loss": 1.0974,
+      "step": 4787
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0472714006349555e-05,
+      "loss": 1.0587,
+      "step": 4788
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0471984923044887e-05,
+      "loss": 1.0232,
+      "step": 4789
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0471255680154753e-05,
+      "loss": 1.0591,
+      "step": 4790
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0470526277692895e-05,
+      "loss": 1.0758,
+      "step": 4791
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0469796715673046e-05,
+      "loss": 1.0301,
+      "step": 4792
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0469066994108947e-05,
+      "loss": 1.0147,
+      "step": 4793
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.046833711301435e-05,
+      "loss": 1.1529,
+      "step": 4794
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0467607072403004e-05,
+      "loss": 1.2285,
+      "step": 4795
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0466876872288654e-05,
+      "loss": 1.0691,
+      "step": 4796
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.046614651268506e-05,
+      "loss": 1.0813,
+      "step": 4797
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.046541599360598e-05,
+      "loss": 1.1614,
+      "step": 4798
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.046468531506517e-05,
+      "loss": 1.2327,
+      "step": 4799
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.04639544770764e-05,
+      "loss": 1.0926,
+      "step": 4800
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.046322347965343e-05,
+      "loss": 1.0617,
+      "step": 4801
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0462492322810034e-05,
+      "loss": 1.112,
+      "step": 4802
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0461761006559983e-05,
+      "loss": 1.0834,
+      "step": 4803
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0461029530917055e-05,
+      "loss": 1.114,
+      "step": 4804
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0460297895895025e-05,
+      "loss": 1.0826,
+      "step": 4805
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0459566101507676e-05,
+      "loss": 1.1377,
+      "step": 4806
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.04588341477688e-05,
+      "loss": 0.9832,
+      "step": 4807
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.045810203469217e-05,
+      "loss": 1.0874,
+      "step": 4808
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0457369762291587e-05,
+      "loss": 1.1383,
+      "step": 4809
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0456637330580843e-05,
+      "loss": 1.0122,
+      "step": 4810
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0455904739573736e-05,
+      "loss": 1.1048,
+      "step": 4811
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.045517198928406e-05,
+      "loss": 1.0131,
+      "step": 4812
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0454439079725623e-05,
+      "loss": 1.076,
+      "step": 4813
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.045370601091223e-05,
+      "loss": 1.1084,
+      "step": 4814
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0452972782857688e-05,
+      "loss": 1.1392,
+      "step": 4815
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.045223939557581e-05,
+      "loss": 1.1418,
+      "step": 4816
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.045150584908041e-05,
+      "loss": 1.0718,
+      "step": 4817
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.045077214338531e-05,
+      "loss": 1.0033,
+      "step": 4818
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.045003827850432e-05,
+      "loss": 1.0966,
+      "step": 4819
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0449304254451274e-05,
+      "loss": 1.1124,
+      "step": 4820
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0448570071239998e-05,
+      "loss": 1.1411,
+      "step": 4821
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0447835728884316e-05,
+      "loss": 1.0613,
+      "step": 4822
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0447101227398065e-05,
+      "loss": 1.0988,
+      "step": 4823
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0446366566795076e-05,
+      "loss": 1.1684,
+      "step": 4824
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0445631747089193e-05,
+      "loss": 1.0997,
+      "step": 4825
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0444896768294256e-05,
+      "loss": 1.1929,
+      "step": 4826
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0444161630424112e-05,
+      "loss": 1.1064,
+      "step": 4827
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.04434263334926e-05,
+      "loss": 1.1288,
+      "step": 4828
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.044269087751358e-05,
+      "loss": 1.0885,
+      "step": 4829
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.04419552625009e-05,
+      "loss": 1.1089,
+      "step": 4830
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0441219488468422e-05,
+      "loss": 1.1598,
+      "step": 4831
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0440483555430002e-05,
+      "loss": 1.0739,
+      "step": 4832
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0439747463399503e-05,
+      "loss": 1.073,
+      "step": 4833
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.043901121239079e-05,
+      "loss": 1.1006,
+      "step": 4834
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0438274802417732e-05,
+      "loss": 1.0776,
+      "step": 4835
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0437538233494206e-05,
+      "loss": 1.101,
+      "step": 4836
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0436801505634075e-05,
+      "loss": 1.029,
+      "step": 4837
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0436064618851227e-05,
+      "loss": 1.1441,
+      "step": 4838
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.043532757315954e-05,
+      "loss": 0.992,
+      "step": 4839
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0434590368572897e-05,
+      "loss": 1.1058,
+      "step": 4840
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0433853005105182e-05,
+      "loss": 1.1348,
+      "step": 4841
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0433115482770288e-05,
+      "loss": 1.0441,
+      "step": 4842
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0432377801582106e-05,
+      "loss": 1.0903,
+      "step": 4843
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0431639961554532e-05,
+      "loss": 1.0313,
+      "step": 4844
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0430901962701467e-05,
+      "loss": 1.0801,
+      "step": 4845
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0430163805036808e-05,
+      "loss": 1.034,
+      "step": 4846
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0429425488574462e-05,
+      "loss": 1.1246,
+      "step": 4847
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.042868701332834e-05,
+      "loss": 1.0774,
+      "step": 4848
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0427948379312344e-05,
+      "loss": 1.1607,
+      "step": 4849
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0427209586540393e-05,
+      "loss": 1.1116,
+      "step": 4850
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0426470635026407e-05,
+      "loss": 1.1328,
+      "step": 4851
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.04257315247843e-05,
+      "loss": 1.1124,
+      "step": 4852
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0424992255827992e-05,
+      "loss": 1.144,
+      "step": 4853
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0424252828171416e-05,
+      "loss": 1.044,
+      "step": 4854
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0423513241828497e-05,
+      "loss": 1.0607,
+      "step": 4855
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0422773496813165e-05,
+      "loss": 1.0228,
+      "step": 4856
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0422033593139357e-05,
+      "loss": 1.0937,
+      "step": 4857
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.042129353082101e-05,
+      "loss": 0.9951,
+      "step": 4858
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0420553309872062e-05,
+      "loss": 1.1407,
+      "step": 4859
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0419812930306457e-05,
+      "loss": 1.0589,
+      "step": 4860
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0419072392138147e-05,
+      "loss": 1.1269,
+      "step": 4861
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0418331695381075e-05,
+      "loss": 1.028,
+      "step": 4862
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0417590840049193e-05,
+      "loss": 1.0605,
+      "step": 4863
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0416849826156463e-05,
+      "loss": 1.0688,
+      "step": 4864
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0416108653716838e-05,
+      "loss": 1.0852,
+      "step": 4865
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.041536732274428e-05,
+      "loss": 1.04,
+      "step": 4866
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0414625833252752e-05,
+      "loss": 1.1398,
+      "step": 4867
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0413884185256227e-05,
+      "loss": 1.0807,
+      "step": 4868
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.041314237876867e-05,
+      "loss": 1.1516,
+      "step": 4869
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0412400413804056e-05,
+      "loss": 1.1114,
+      "step": 4870
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.041165829037636e-05,
+      "loss": 1.1418,
+      "step": 4871
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0410916008499564e-05,
+      "loss": 1.056,
+      "step": 4872
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0410173568187647e-05,
+      "loss": 1.1006,
+      "step": 4873
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0409430969454596e-05,
+      "loss": 0.998,
+      "step": 4874
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.04086882123144e-05,
+      "loss": 1.0865,
+      "step": 4875
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0407945296781047e-05,
+      "loss": 1.0559,
+      "step": 4876
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0407202222868536e-05,
+      "loss": 1.1041,
+      "step": 4877
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.040645899059086e-05,
+      "loss": 1.1018,
+      "step": 4878
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0405715599962023e-05,
+      "loss": 1.085,
+      "step": 4879
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0404972050996024e-05,
+      "loss": 1.1022,
+      "step": 4880
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.040422834370687e-05,
+      "loss": 1.1411,
+      "step": 4881
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0403484478108572e-05,
+      "loss": 1.0688,
+      "step": 4882
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0402740454215143e-05,
+      "loss": 1.1077,
+      "step": 4883
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0401996272040594e-05,
+      "loss": 1.108,
+      "step": 4884
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0401251931598945e-05,
+      "loss": 1.1506,
+      "step": 4885
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0400507432904223e-05,
+      "loss": 1.1465,
+      "step": 4886
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.039976277597044e-05,
+      "loss": 1.1442,
+      "step": 4887
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.039901796081163e-05,
+      "loss": 1.0576,
+      "step": 4888
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0398272987441822e-05,
+      "loss": 1.0828,
+      "step": 4889
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0397527855875053e-05,
+      "loss": 1.0955,
+      "step": 4890
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.039678256612535e-05,
+      "loss": 1.0729,
+      "step": 4891
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0396037118206762e-05,
+      "loss": 1.1646,
+      "step": 4892
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0395291512133325e-05,
+      "loss": 0.9412,
+      "step": 4893
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.039454574791908e-05,
+      "loss": 1.148,
+      "step": 4894
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0393799825578082e-05,
+      "loss": 1.1323,
+      "step": 4895
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.039305374512438e-05,
+      "loss": 0.987,
+      "step": 4896
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.039230750657203e-05,
+      "loss": 1.1043,
+      "step": 4897
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0391561109935085e-05,
+      "loss": 1.1023,
+      "step": 4898
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0390814555227603e-05,
+      "loss": 1.1204,
+      "step": 4899
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0390067842463655e-05,
+      "loss": 1.0846,
+      "step": 4900
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0389320971657294e-05,
+      "loss": 1.1183,
+      "step": 4901
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0388573942822598e-05,
+      "loss": 1.1423,
+      "step": 4902
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0387826755973642e-05,
+      "loss": 1.0731,
+      "step": 4903
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.038707941112449e-05,
+      "loss": 1.0999,
+      "step": 4904
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0386331908289228e-05,
+      "loss": 1.1278,
+      "step": 4905
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.038558424748193e-05,
+      "loss": 1.1052,
+      "step": 4906
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0384836428716687e-05,
+      "loss": 0.979,
+      "step": 4907
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.038408845200758e-05,
+      "loss": 1.1347,
+      "step": 4908
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0383340317368698e-05,
+      "loss": 1.0533,
+      "step": 4909
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0382592024814137e-05,
+      "loss": 1.1094,
+      "step": 4910
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.038184357435799e-05,
+      "loss": 1.0963,
+      "step": 4911
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.038109496601436e-05,
+      "loss": 1.0235,
+      "step": 4912
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0380346199797343e-05,
+      "loss": 1.1244,
+      "step": 4913
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0379597275721044e-05,
+      "loss": 1.1627,
+      "step": 4914
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0378848193799573e-05,
+      "loss": 1.1362,
+      "step": 4915
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.037809895404704e-05,
+      "loss": 1.1694,
+      "step": 4916
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0377349556477556e-05,
+      "loss": 1.059,
+      "step": 4917
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.037660000110524e-05,
+      "loss": 1.0992,
+      "step": 4918
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0375850287944208e-05,
+      "loss": 1.1441,
+      "step": 4919
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0375100417008585e-05,
+      "loss": 1.0974,
+      "step": 4920
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0374350388312493e-05,
+      "loss": 1.073,
+      "step": 4921
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0373600201870063e-05,
+      "loss": 1.1214,
+      "step": 4922
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0372849857695432e-05,
+      "loss": 1.1324,
+      "step": 4923
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.037209935580272e-05,
+      "loss": 1.1223,
+      "step": 4924
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0371348696206077e-05,
+      "loss": 1.0655,
+      "step": 4925
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0370597878919634e-05,
+      "loss": 1.1577,
+      "step": 4926
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0369846903957537e-05,
+      "loss": 1.1392,
+      "step": 4927
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.036909577133394e-05,
+      "loss": 1.1089,
+      "step": 4928
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.036834448106298e-05,
+      "loss": 1.125,
+      "step": 4929
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0367593033158815e-05,
+      "loss": 1.0596,
+      "step": 4930
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.03668414276356e-05,
+      "loss": 1.0923,
+      "step": 4931
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.036608966450749e-05,
+      "loss": 1.0791,
+      "step": 4932
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0365337743788648e-05,
+      "loss": 1.1788,
+      "step": 4933
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0364585665493235e-05,
+      "loss": 1.0114,
+      "step": 4934
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0363833429635423e-05,
+      "loss": 1.1299,
+      "step": 4935
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0363081036229374e-05,
+      "loss": 1.1277,
+      "step": 4936
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.036232848528927e-05,
+      "loss": 1.1166,
+      "step": 4937
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.036157577682928e-05,
+      "loss": 1.0236,
+      "step": 4938
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0360822910863592e-05,
+      "loss": 1.0592,
+      "step": 4939
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0360069887406373e-05,
+      "loss": 1.083,
+      "step": 4940
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0359316706471815e-05,
+      "loss": 1.1909,
+      "step": 4941
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.035856336807411e-05,
+      "loss": 1.0772,
+      "step": 4942
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.035780987222744e-05,
+      "loss": 1.1247,
+      "step": 4943
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0357056218946004e-05,
+      "loss": 1.0757,
+      "step": 4944
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0356302408243998e-05,
+      "loss": 1.0811,
+      "step": 4945
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0355548440135622e-05,
+      "loss": 1.0396,
+      "step": 4946
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.035479431463507e-05,
+      "loss": 1.1089,
+      "step": 4947
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.035404003175656e-05,
+      "loss": 0.9953,
+      "step": 4948
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0353285591514294e-05,
+      "loss": 1.1165,
+      "step": 4949
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0352530993922485e-05,
+      "loss": 1.0896,
+      "step": 4950
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0351776238995345e-05,
+      "loss": 1.0027,
+      "step": 4951
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0351021326747092e-05,
+      "loss": 1.0924,
+      "step": 4952
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0350266257191947e-05,
+      "loss": 1.1329,
+      "step": 4953
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.034951103034413e-05,
+      "loss": 1.0327,
+      "step": 4954
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0348755646217872e-05,
+      "loss": 1.1394,
+      "step": 4955
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0348000104827397e-05,
+      "loss": 1.1431,
+      "step": 4956
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0347244406186943e-05,
+      "loss": 1.1065,
+      "step": 4957
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0346488550310743e-05,
+      "loss": 1.1436,
+      "step": 4958
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.034573253721303e-05,
+      "loss": 1.1085,
+      "step": 4959
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.034497636690805e-05,
+      "loss": 1.074,
+      "step": 4960
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0344220039410043e-05,
+      "loss": 1.0568,
+      "step": 4961
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0343463554733263e-05,
+      "loss": 1.0793,
+      "step": 4962
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.034270691289195e-05,
+      "loss": 1.1285,
+      "step": 4963
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0341950113900364e-05,
+      "loss": 1.0489,
+      "step": 4964
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.034119315777276e-05,
+      "loss": 1.1058,
+      "step": 4965
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0340436044523394e-05,
+      "loss": 1.1326,
+      "step": 4966
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0339678774166526e-05,
+      "loss": 1.0596,
+      "step": 4967
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0338921346716427e-05,
+      "loss": 1.0961,
+      "step": 4968
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0338163762187358e-05,
+      "loss": 1.106,
+      "step": 4969
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0337406020593596e-05,
+      "loss": 1.0907,
+      "step": 4970
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0336648121949406e-05,
+      "loss": 1.059,
+      "step": 4971
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.033589006626907e-05,
+      "loss": 1.0991,
+      "step": 4972
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.033513185356687e-05,
+      "loss": 1.0767,
+      "step": 4973
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0334373483857084e-05,
+      "loss": 1.1425,
+      "step": 4974
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0333614957154003e-05,
+      "loss": 1.1005,
+      "step": 4975
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.03328562734719e-05,
+      "loss": 1.0971,
+      "step": 4976
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0332097432825084e-05,
+      "loss": 1.0452,
+      "step": 4977
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.033133843522784e-05,
+      "loss": 1.1608,
+      "step": 4978
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0330579280694465e-05,
+      "loss": 1.0995,
+      "step": 4979
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0329819969239264e-05,
+      "loss": 1.1034,
+      "step": 4980
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0329060500876537e-05,
+      "loss": 1.1133,
+      "step": 4981
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0328300875620587e-05,
+      "loss": 1.0792,
+      "step": 4982
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0327541093485732e-05,
+      "loss": 1.0881,
+      "step": 4983
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0326781154486275e-05,
+      "loss": 1.1105,
+      "step": 4984
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.032602105863653e-05,
+      "loss": 1.0854,
+      "step": 4985
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.032526080595082e-05,
+      "loss": 1.0869,
+      "step": 4986
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.032450039644347e-05,
+      "loss": 1.1029,
+      "step": 4987
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.032373983012879e-05,
+      "loss": 1.0752,
+      "step": 4988
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.032297910702112e-05,
+      "loss": 1.0083,
+      "step": 4989
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.032221822713478e-05,
+      "loss": 1.1796,
+      "step": 4990
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.032145719048411e-05,
+      "loss": 1.0203,
+      "step": 4991
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0320695997083444e-05,
+      "loss": 1.0872,
+      "step": 4992
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0319934646947117e-05,
+      "loss": 1.1562,
+      "step": 4993
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0319173140089474e-05,
+      "loss": 1.0873,
+      "step": 4994
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0318411476524858e-05,
+      "loss": 1.083,
+      "step": 4995
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0317649656267613e-05,
+      "loss": 1.0893,
+      "step": 4996
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0316887679332095e-05,
+      "loss": 1.0783,
+      "step": 4997
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0316125545732654e-05,
+      "loss": 1.0482,
+      "step": 4998
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0315363255483647e-05,
+      "loss": 1.0785,
+      "step": 4999
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.031460080859943e-05,
+      "loss": 1.0364,
+      "step": 5000
+    },
+    {
+      "epoch": 0.21,
+      "eval_text_loss": 0.8074743151664734,
+      "eval_text_runtime": 1.5697,
+      "eval_text_samples_per_second": 47.142,
+      "eval_text_steps_per_second": 1.274,
+      "step": 5000
+    },
+    {
+      "epoch": 0.21,
+      "eval_svit_loss": 1.1665762662887573,
+      "eval_svit_runtime": 4.9288,
+      "eval_svit_samples_per_second": 138.168,
+      "eval_svit_steps_per_second": 2.232,
+      "step": 5000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0313838205094374e-05,
+      "loss": 1.1003,
+      "step": 5001
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0313075444982835e-05,
+      "loss": 1.0854,
+      "step": 5002
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0312312528279184e-05,
+      "loss": 1.0739,
+      "step": 5003
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.031154945499779e-05,
+      "loss": 1.1271,
+      "step": 5004
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.031078622515303e-05,
+      "loss": 1.0952,
+      "step": 5005
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0310022838759282e-05,
+      "loss": 1.1587,
+      "step": 5006
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0309259295830922e-05,
+      "loss": 1.0997,
+      "step": 5007
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0308495596382332e-05,
+      "loss": 1.1654,
+      "step": 5008
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0307731740427905e-05,
+      "loss": 1.0017,
+      "step": 5009
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.030696772798202e-05,
+      "loss": 1.1131,
+      "step": 5010
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0306203559059076e-05,
+      "loss": 1.1854,
+      "step": 5011
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0305439233673463e-05,
+      "loss": 1.0167,
+      "step": 5012
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0304674751839584e-05,
+      "loss": 1.171,
+      "step": 5013
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0303910113571832e-05,
+      "loss": 1.1272,
+      "step": 5014
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0303145318884617e-05,
+      "loss": 1.1009,
+      "step": 5015
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0302380367792337e-05,
+      "loss": 1.0097,
+      "step": 5016
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0301615260309413e-05,
+      "loss": 1.1553,
+      "step": 5017
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0300849996450245e-05,
+      "loss": 1.1334,
+      "step": 5018
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.030008457622926e-05,
+      "loss": 1.1209,
+      "step": 5019
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0299318999660864e-05,
+      "loss": 1.0844,
+      "step": 5020
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0298553266759485e-05,
+      "loss": 1.0258,
+      "step": 5021
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.029778737753955e-05,
+      "loss": 1.1303,
+      "step": 5022
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.029702133201548e-05,
+      "loss": 1.1911,
+      "step": 5023
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0296255130201708e-05,
+      "loss": 1.1395,
+      "step": 5024
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0295488772112666e-05,
+      "loss": 1.0706,
+      "step": 5025
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.029472225776279e-05,
+      "loss": 1.1145,
+      "step": 5026
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0293955587166515e-05,
+      "loss": 1.128,
+      "step": 5027
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0293188760338286e-05,
+      "loss": 1.0447,
+      "step": 5028
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.029242177729255e-05,
+      "loss": 1.1055,
+      "step": 5029
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0291654638043748e-05,
+      "loss": 1.0579,
+      "step": 5030
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.029088734260634e-05,
+      "loss": 1.1444,
+      "step": 5031
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0290119890994773e-05,
+      "loss": 1.0404,
+      "step": 5032
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0289352283223502e-05,
+      "loss": 1.0367,
+      "step": 5033
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0288584519306986e-05,
+      "loss": 1.0561,
+      "step": 5034
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0287816599259696e-05,
+      "loss": 1.0734,
+      "step": 5035
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.028704852309609e-05,
+      "loss": 1.1165,
+      "step": 5036
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0286280290830634e-05,
+      "loss": 1.0435,
+      "step": 5037
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0285511902477804e-05,
+      "loss": 1.0145,
+      "step": 5038
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0284743358052075e-05,
+      "loss": 1.1029,
+      "step": 5039
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0283974657567916e-05,
+      "loss": 1.0874,
+      "step": 5040
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0283205801039817e-05,
+      "loss": 1.0188,
+      "step": 5041
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.028243678848225e-05,
+      "loss": 1.0653,
+      "step": 5042
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.028166761990971e-05,
+      "loss": 1.1555,
+      "step": 5043
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0280898295336684e-05,
+      "loss": 1.0743,
+      "step": 5044
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.028012881477766e-05,
+      "loss": 1.0683,
+      "step": 5045
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0279359178247136e-05,
+      "loss": 1.0681,
+      "step": 5046
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0278589385759606e-05,
+      "loss": 1.1078,
+      "step": 5047
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0277819437329578e-05,
+      "loss": 1.0754,
+      "step": 5048
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0277049332971545e-05,
+      "loss": 1.0686,
+      "step": 5049
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0276279072700018e-05,
+      "loss": 1.0844,
+      "step": 5050
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.027550865652951e-05,
+      "loss": 1.1375,
+      "step": 5051
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0274738084474526e-05,
+      "loss": 1.0644,
+      "step": 5052
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0273967356549587e-05,
+      "loss": 1.1966,
+      "step": 5053
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.027319647276921e-05,
+      "loss": 1.0294,
+      "step": 5054
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0272425433147917e-05,
+      "loss": 1.0778,
+      "step": 5055
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0271654237700225e-05,
+      "loss": 1.0852,
+      "step": 5056
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.027088288644067e-05,
+      "loss": 1.0711,
+      "step": 5057
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0270111379383775e-05,
+      "loss": 1.0881,
+      "step": 5058
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.026933971654408e-05,
+      "loss": 1.1403,
+      "step": 5059
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.026856789793611e-05,
+      "loss": 1.1034,
+      "step": 5060
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0267795923574414e-05,
+      "loss": 1.1109,
+      "step": 5061
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.026702379347353e-05,
+      "loss": 1.1099,
+      "step": 5062
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0266251507648e-05,
+      "loss": 1.088,
+      "step": 5063
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0265479066112372e-05,
+      "loss": 1.0983,
+      "step": 5064
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.02647064688812e-05,
+      "loss": 1.1322,
+      "step": 5065
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0263933715969033e-05,
+      "loss": 1.0775,
+      "step": 5066
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.026316080739043e-05,
+      "loss": 1.0577,
+      "step": 5067
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.026238774315995e-05,
+      "loss": 1.0587,
+      "step": 5068
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0261614523292155e-05,
+      "loss": 1.1615,
+      "step": 5069
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.026084114780161e-05,
+      "loss": 1.039,
+      "step": 5070
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.026006761670288e-05,
+      "loss": 1.1202,
+      "step": 5071
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.025929393001054e-05,
+      "loss": 1.0619,
+      "step": 5072
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0258520087739163e-05,
+      "loss": 1.0542,
+      "step": 5073
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.025774608990332e-05,
+      "loss": 1.0794,
+      "step": 5074
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.02569719365176e-05,
+      "loss": 1.1164,
+      "step": 5075
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0256197627596583e-05,
+      "loss": 1.179,
+      "step": 5076
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.025542316315485e-05,
+      "loss": 1.1451,
+      "step": 5077
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.025464854320699e-05,
+      "loss": 1.1025,
+      "step": 5078
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0253873767767603e-05,
+      "loss": 1.1368,
+      "step": 5079
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.025309883685127e-05,
+      "loss": 1.0748,
+      "step": 5080
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.02523237504726e-05,
+      "loss": 1.0316,
+      "step": 5081
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0251548508646187e-05,
+      "loss": 1.1328,
+      "step": 5082
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0250773111386634e-05,
+      "loss": 1.0672,
+      "step": 5083
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0249997558708552e-05,
+      "loss": 1.1543,
+      "step": 5084
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0249221850626542e-05,
+      "loss": 1.0997,
+      "step": 5085
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0248445987155223e-05,
+      "loss": 1.0942,
+      "step": 5086
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0247669968309205e-05,
+      "loss": 1.0189,
+      "step": 5087
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0246893794103114e-05,
+      "loss": 1.0568,
+      "step": 5088
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0246117464551556e-05,
+      "loss": 1.1166,
+      "step": 5089
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0245340979669166e-05,
+      "loss": 1.0397,
+      "step": 5090
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0244564339470572e-05,
+      "loss": 1.0924,
+      "step": 5091
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0243787543970395e-05,
+      "loss": 1.0518,
+      "step": 5092
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.024301059318327e-05,
+      "loss": 1.0332,
+      "step": 5093
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0242233487123838e-05,
+      "loss": 1.1266,
+      "step": 5094
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.024145622580673e-05,
+      "loss": 1.1794,
+      "step": 5095
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.024067880924659e-05,
+      "loss": 1.0842,
+      "step": 5096
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0239901237458065e-05,
+      "loss": 1.0748,
+      "step": 5097
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.02391235104558e-05,
+      "loss": 1.14,
+      "step": 5098
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0238345628254446e-05,
+      "loss": 1.1302,
+      "step": 5099
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0237567590868646e-05,
+      "loss": 1.0877,
+      "step": 5100
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0236789398313068e-05,
+      "loss": 1.0453,
+      "step": 5101
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.023601105060237e-05,
+      "loss": 1.1257,
+      "step": 5102
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0235232547751206e-05,
+      "loss": 1.1668,
+      "step": 5103
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.023445388977425e-05,
+      "loss": 1.0639,
+      "step": 5104
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0233675076686156e-05,
+      "loss": 1.1645,
+      "step": 5105
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0232896108501607e-05,
+      "loss": 1.122,
+      "step": 5106
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.023211698523527e-05,
+      "loss": 1.0851,
+      "step": 5107
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0231337706901824e-05,
+      "loss": 1.1062,
+      "step": 5108
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.023055827351595e-05,
+      "loss": 1.1383,
+      "step": 5109
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0229778685092324e-05,
+      "loss": 1.0537,
+      "step": 5110
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0228998941645637e-05,
+      "loss": 1.1495,
+      "step": 5111
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0228219043190572e-05,
+      "loss": 1.1504,
+      "step": 5112
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.022743898974182e-05,
+      "loss": 1.0833,
+      "step": 5113
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.022665878131408e-05,
+      "loss": 1.0432,
+      "step": 5114
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0225878417922043e-05,
+      "loss": 1.0268,
+      "step": 5115
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.022509789958041e-05,
+      "loss": 1.1405,
+      "step": 5116
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0224317226303892e-05,
+      "loss": 1.0905,
+      "step": 5117
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0223536398107178e-05,
+      "loss": 1.0691,
+      "step": 5118
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0222755415004988e-05,
+      "loss": 1.0352,
+      "step": 5119
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.022197427701203e-05,
+      "loss": 1.102,
+      "step": 5120
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.022119298414302e-05,
+      "loss": 1.1002,
+      "step": 5121
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0220411536412674e-05,
+      "loss": 1.0722,
+      "step": 5122
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0219629933835713e-05,
+      "loss": 1.0752,
+      "step": 5123
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0218848176426858e-05,
+      "loss": 1.0212,
+      "step": 5124
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0218066264200834e-05,
+      "loss": 1.0645,
+      "step": 5125
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0217284197172373e-05,
+      "loss": 1.1422,
+      "step": 5126
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0216501975356204e-05,
+      "loss": 1.0556,
+      "step": 5127
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.021571959876706e-05,
+      "loss": 1.0548,
+      "step": 5128
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0214937067419685e-05,
+      "loss": 1.0423,
+      "step": 5129
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0214154381328816e-05,
+      "loss": 1.0997,
+      "step": 5130
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0213371540509192e-05,
+      "loss": 1.0611,
+      "step": 5131
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0212588544975568e-05,
+      "loss": 1.0104,
+      "step": 5132
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0211805394742685e-05,
+      "loss": 1.0538,
+      "step": 5133
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0211022089825296e-05,
+      "loss": 1.0242,
+      "step": 5134
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.021023863023816e-05,
+      "loss": 1.0361,
+      "step": 5135
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0209455015996035e-05,
+      "loss": 1.1195,
+      "step": 5136
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.020867124711368e-05,
+      "loss": 1.025,
+      "step": 5137
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.020788732360586e-05,
+      "loss": 1.2122,
+      "step": 5138
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0207103245487336e-05,
+      "loss": 1.0958,
+      "step": 5139
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0206319012772886e-05,
+      "loss": 1.1681,
+      "step": 5140
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0205534625477275e-05,
+      "loss": 1.1133,
+      "step": 5141
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0204750083615284e-05,
+      "loss": 1.0886,
+      "step": 5142
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.020396538720169e-05,
+      "loss": 1.0666,
+      "step": 5143
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0203180536251272e-05,
+      "loss": 1.0964,
+      "step": 5144
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0202395530778815e-05,
+      "loss": 1.0782,
+      "step": 5145
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0201610370799105e-05,
+      "loss": 1.0026,
+      "step": 5146
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.020082505632694e-05,
+      "loss": 0.9409,
+      "step": 5147
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.02000395873771e-05,
+      "loss": 1.0776,
+      "step": 5148
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.019925396396439e-05,
+      "loss": 1.1715,
+      "step": 5149
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0198468186103603e-05,
+      "loss": 1.0857,
+      "step": 5150
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0197682253809545e-05,
+      "loss": 1.0797,
+      "step": 5151
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.019689616709702e-05,
+      "loss": 1.0683,
+      "step": 5152
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0196109925980832e-05,
+      "loss": 1.1005,
+      "step": 5153
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0195323530475792e-05,
+      "loss": 1.1289,
+      "step": 5154
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0194536980596717e-05,
+      "loss": 1.1087,
+      "step": 5155
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.019375027635842e-05,
+      "loss": 1.0614,
+      "step": 5156
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.019296341777572e-05,
+      "loss": 1.1482,
+      "step": 5157
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0192176404863443e-05,
+      "loss": 1.1023,
+      "step": 5158
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.019138923763641e-05,
+      "loss": 1.1374,
+      "step": 5159
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0190601916109445e-05,
+      "loss": 1.0421,
+      "step": 5160
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0189814440297385e-05,
+      "loss": 1.1699,
+      "step": 5161
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0189026810215065e-05,
+      "loss": 1.0533,
+      "step": 5162
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0188239025877315e-05,
+      "loss": 1.1376,
+      "step": 5163
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0187451087298975e-05,
+      "loss": 1.0874,
+      "step": 5164
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0186662994494892e-05,
+      "loss": 1.1091,
+      "step": 5165
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0185874747479913e-05,
+      "loss": 1.1028,
+      "step": 5166
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0185086346268877e-05,
+      "loss": 1.1134,
+      "step": 5167
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.018429779087664e-05,
+      "loss": 1.0774,
+      "step": 5168
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.018350908131806e-05,
+      "loss": 1.1388,
+      "step": 5169
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0182720217607982e-05,
+      "loss": 1.1081,
+      "step": 5170
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0181931199761278e-05,
+      "loss": 1.091,
+      "step": 5171
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.018114202779281e-05,
+      "loss": 1.1634,
+      "step": 5172
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.018035270171743e-05,
+      "loss": 1.0698,
+      "step": 5173
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.017956322155002e-05,
+      "loss": 1.1177,
+      "step": 5174
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.017877358730545e-05,
+      "loss": 1.0864,
+      "step": 5175
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.017798379899859e-05,
+      "loss": 1.0863,
+      "step": 5176
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0177193856644316e-05,
+      "loss": 1.0401,
+      "step": 5177
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0176403760257512e-05,
+      "loss": 1.0432,
+      "step": 5178
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.017561350985306e-05,
+      "loss": 1.1885,
+      "step": 5179
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0174823105445846e-05,
+      "loss": 1.0612,
+      "step": 5180
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0174032547050755e-05,
+      "loss": 1.1234,
+      "step": 5181
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0173241834682682e-05,
+      "loss": 1.1185,
+      "step": 5182
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0172450968356526e-05,
+      "loss": 1.0874,
+      "step": 5183
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0171659948087176e-05,
+      "loss": 1.0672,
+      "step": 5184
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0170868773889534e-05,
+      "loss": 1.0962,
+      "step": 5185
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0170077445778504e-05,
+      "loss": 1.0375,
+      "step": 5186
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0169285963769e-05,
+      "loss": 1.1536,
+      "step": 5187
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.016849432787592e-05,
+      "loss": 1.0298,
+      "step": 5188
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.016770253811418e-05,
+      "loss": 1.1294,
+      "step": 5189
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0166910594498696e-05,
+      "loss": 1.1128,
+      "step": 5190
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0166118497044384e-05,
+      "loss": 1.1544,
+      "step": 5191
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0165326245766165e-05,
+      "loss": 1.1152,
+      "step": 5192
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0164533840678964e-05,
+      "loss": 1.1039,
+      "step": 5193
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0163741281797708e-05,
+      "loss": 1.1872,
+      "step": 5194
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0162948569137322e-05,
+      "loss": 1.1014,
+      "step": 5195
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.016215570271274e-05,
+      "loss": 1.1141,
+      "step": 5196
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0161362682538897e-05,
+      "loss": 1.0583,
+      "step": 5197
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0160569508630735e-05,
+      "loss": 1.0972,
+      "step": 5198
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0159776181003193e-05,
+      "loss": 1.0969,
+      "step": 5199
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.015898269967121e-05,
+      "loss": 1.0518,
+      "step": 5200
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.015818906464974e-05,
+      "loss": 1.0531,
+      "step": 5201
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0157395275953723e-05,
+      "loss": 1.0304,
+      "step": 5202
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.015660133359812e-05,
+      "loss": 0.9954,
+      "step": 5203
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0155807237597885e-05,
+      "loss": 1.0655,
+      "step": 5204
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0155012987967973e-05,
+      "loss": 1.1157,
+      "step": 5205
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.015421858472335e-05,
+      "loss": 1.1376,
+      "step": 5206
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0153424027878973e-05,
+      "loss": 1.0817,
+      "step": 5207
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0152629317449815e-05,
+      "loss": 1.0864,
+      "step": 5208
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0151834453450844e-05,
+      "loss": 1.144,
+      "step": 5209
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0151039435897033e-05,
+      "loss": 1.0006,
+      "step": 5210
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.015024426480336e-05,
+      "loss": 1.091,
+      "step": 5211
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0149448940184798e-05,
+      "loss": 1.1006,
+      "step": 5212
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.014865346205633e-05,
+      "loss": 1.0616,
+      "step": 5213
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.014785783043294e-05,
+      "loss": 1.1208,
+      "step": 5214
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.014706204532962e-05,
+      "loss": 1.1357,
+      "step": 5215
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.014626610676136e-05,
+      "loss": 1.1174,
+      "step": 5216
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0145470014743144e-05,
+      "loss": 1.0529,
+      "step": 5217
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0144673769289973e-05,
+      "loss": 1.0736,
+      "step": 5218
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.014387737041685e-05,
+      "loss": 1.1225,
+      "step": 5219
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0143080818138772e-05,
+      "loss": 1.0938,
+      "step": 5220
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0142284112470744e-05,
+      "loss": 1.0705,
+      "step": 5221
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0141487253427777e-05,
+      "loss": 1.0939,
+      "step": 5222
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0140690241024873e-05,
+      "loss": 1.1619,
+      "step": 5223
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0139893075277052e-05,
+      "loss": 1.0896,
+      "step": 5224
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0139095756199327e-05,
+      "loss": 1.107,
+      "step": 5225
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.013829828380672e-05,
+      "loss": 1.0779,
+      "step": 5226
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0137500658114254e-05,
+      "loss": 1.0435,
+      "step": 5227
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.013670287913695e-05,
+      "loss": 1.1013,
+      "step": 5228
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0135904946889837e-05,
+      "loss": 1.0303,
+      "step": 5229
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0135106861387946e-05,
+      "loss": 1.1127,
+      "step": 5230
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0134308622646306e-05,
+      "loss": 1.0932,
+      "step": 5231
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0133510230679957e-05,
+      "loss": 1.1314,
+      "step": 5232
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0132711685503942e-05,
+      "loss": 1.0975,
+      "step": 5233
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0131912987133297e-05,
+      "loss": 0.9866,
+      "step": 5234
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0131114135583072e-05,
+      "loss": 1.1122,
+      "step": 5235
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0130315130868305e-05,
+      "loss": 1.0575,
+      "step": 5236
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0129515973004062e-05,
+      "loss": 1.1423,
+      "step": 5237
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0128716662005386e-05,
+      "loss": 1.0294,
+      "step": 5238
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.012791719788733e-05,
+      "loss": 1.0401,
+      "step": 5239
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0127117580664967e-05,
+      "loss": 1.0949,
+      "step": 5240
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0126317810353348e-05,
+      "loss": 1.1469,
+      "step": 5241
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.012551788696754e-05,
+      "loss": 1.0724,
+      "step": 5242
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0124717810522616e-05,
+      "loss": 1.063,
+      "step": 5243
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0123917581033643e-05,
+      "loss": 1.1648,
+      "step": 5244
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0123117198515694e-05,
+      "loss": 1.106,
+      "step": 5245
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0122316662983848e-05,
+      "loss": 1.0109,
+      "step": 5246
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0121515974453186e-05,
+      "loss": 1.0705,
+      "step": 5247
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0120715132938785e-05,
+      "loss": 1.0081,
+      "step": 5248
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0119914138455736e-05,
+      "loss": 1.105,
+      "step": 5249
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0119112991019124e-05,
+      "loss": 1.093,
+      "step": 5250
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0118311690644038e-05,
+      "loss": 0.9704,
+      "step": 5251
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0117510237345573e-05,
+      "loss": 1.033,
+      "step": 5252
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0116708631138833e-05,
+      "loss": 1.0744,
+      "step": 5253
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0115906872038913e-05,
+      "loss": 1.0786,
+      "step": 5254
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.011510496006091e-05,
+      "loss": 1.1062,
+      "step": 5255
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0114302895219937e-05,
+      "loss": 1.0393,
+      "step": 5256
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.01135006775311e-05,
+      "loss": 1.0552,
+      "step": 5257
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0112698307009507e-05,
+      "loss": 1.1133,
+      "step": 5258
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.011189578367028e-05,
+      "loss": 1.0967,
+      "step": 5259
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0111093107528528e-05,
+      "loss": 1.0612,
+      "step": 5260
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0110290278599374e-05,
+      "loss": 1.031,
+      "step": 5261
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0109487296897946e-05,
+      "loss": 1.1005,
+      "step": 5262
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0108684162439356e-05,
+      "loss": 1.0763,
+      "step": 5263
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0107880875238748e-05,
+      "loss": 1.0618,
+      "step": 5264
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.010707743531124e-05,
+      "loss": 1.1598,
+      "step": 5265
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0106273842671975e-05,
+      "loss": 1.1025,
+      "step": 5266
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.010547009733609e-05,
+      "loss": 1.1342,
+      "step": 5267
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0104666199318722e-05,
+      "loss": 1.1277,
+      "step": 5268
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0103862148635016e-05,
+      "loss": 1.099,
+      "step": 5269
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0103057945300116e-05,
+      "loss": 1.1031,
+      "step": 5270
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0102253589329167e-05,
+      "loss": 1.0706,
+      "step": 5271
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0101449080737328e-05,
+      "loss": 1.0265,
+      "step": 5272
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.010064441953975e-05,
+      "loss": 1.0791,
+      "step": 5273
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.009983960575159e-05,
+      "loss": 1.0759,
+      "step": 5274
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.009903463938801e-05,
+      "loss": 1.1157,
+      "step": 5275
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0098229520464168e-05,
+      "loss": 1.1076,
+      "step": 5276
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.009742424899524e-05,
+      "loss": 1.185,
+      "step": 5277
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.009661882499638e-05,
+      "loss": 1.089,
+      "step": 5278
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0095813248482774e-05,
+      "loss": 1.0279,
+      "step": 5279
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0095007519469585e-05,
+      "loss": 1.0761,
+      "step": 5280
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0094201637972e-05,
+      "loss": 1.0597,
+      "step": 5281
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0093395604005198e-05,
+      "loss": 1.0356,
+      "step": 5282
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0092589417584354e-05,
+      "loss": 1.051,
+      "step": 5283
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0091783078724658e-05,
+      "loss": 1.0975,
+      "step": 5284
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0090976587441302e-05,
+      "loss": 1.1328,
+      "step": 5285
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0090169943749478e-05,
+      "loss": 1.1588,
+      "step": 5286
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.008936314766438e-05,
+      "loss": 1.1165,
+      "step": 5287
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0088556199201196e-05,
+      "loss": 1.145,
+      "step": 5288
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.008774909837514e-05,
+      "loss": 1.0825,
+      "step": 5289
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.008694184520141e-05,
+      "loss": 1.0048,
+      "step": 5290
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0086134439695206e-05,
+      "loss": 1.0769,
+      "step": 5291
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0085326881871748e-05,
+      "loss": 1.0747,
+      "step": 5292
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.008451917174624e-05,
+      "loss": 1.1724,
+      "step": 5293
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.00837113093339e-05,
+      "loss": 1.1187,
+      "step": 5294
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0082903294649947e-05,
+      "loss": 1.0597,
+      "step": 5295
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0082095127709595e-05,
+      "loss": 1.1115,
+      "step": 5296
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0081286808528075e-05,
+      "loss": 1.1211,
+      "step": 5297
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0080478337120606e-05,
+      "loss": 1.1271,
+      "step": 5298
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0079669713502427e-05,
+      "loss": 1.0735,
+      "step": 5299
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0078860937688758e-05,
+      "loss": 1.1026,
+      "step": 5300
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.007805200969484e-05,
+      "loss": 1.0314,
+      "step": 5301
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0077242929535914e-05,
+      "loss": 1.1485,
+      "step": 5302
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0076433697227212e-05,
+      "loss": 1.11,
+      "step": 5303
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0075624312783986e-05,
+      "loss": 1.1616,
+      "step": 5304
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0074814776221472e-05,
+      "loss": 1.1095,
+      "step": 5305
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.007400508755493e-05,
+      "loss": 1.1171,
+      "step": 5306
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0073195246799606e-05,
+      "loss": 1.0563,
+      "step": 5307
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0072385253970757e-05,
+      "loss": 1.1218,
+      "step": 5308
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.007157510908364e-05,
+      "loss": 1.1797,
+      "step": 5309
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.007076481215352e-05,
+      "loss": 1.1117,
+      "step": 5310
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0069954363195646e-05,
+      "loss": 1.1251,
+      "step": 5311
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.00691437622253e-05,
+      "loss": 1.0509,
+      "step": 5312
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0068333009257744e-05,
+      "loss": 1.0153,
+      "step": 5313
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0067522104308253e-05,
+      "loss": 1.0751,
+      "step": 5314
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.00667110473921e-05,
+      "loss": 1.0868,
+      "step": 5315
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.006589983852456e-05,
+      "loss": 1.1219,
+      "step": 5316
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.006508847772092e-05,
+      "loss": 1.0615,
+      "step": 5317
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.006427696499646e-05,
+      "loss": 1.167,
+      "step": 5318
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0063465300366465e-05,
+      "loss": 1.0625,
+      "step": 5319
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0062653483846224e-05,
+      "loss": 1.14,
+      "step": 5320
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0061841515451035e-05,
+      "loss": 1.0721,
+      "step": 5321
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0061029395196187e-05,
+      "loss": 1.0698,
+      "step": 5322
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.006021712309698e-05,
+      "loss": 1.1284,
+      "step": 5323
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.005940469916871e-05,
+      "loss": 1.1534,
+      "step": 5324
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.005859212342669e-05,
+      "loss": 1.0961,
+      "step": 5325
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0057779395886217e-05,
+      "loss": 1.1405,
+      "step": 5326
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0056966516562604e-05,
+      "loss": 1.0687,
+      "step": 5327
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0056153485471168e-05,
+      "loss": 1.1467,
+      "step": 5328
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0055340302627213e-05,
+      "loss": 1.0709,
+      "step": 5329
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0054526968046068e-05,
+      "loss": 1.0353,
+      "step": 5330
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.005371348174305e-05,
+      "loss": 1.0897,
+      "step": 5331
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0052899843733476e-05,
+      "loss": 1.1116,
+      "step": 5332
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0052086054032678e-05,
+      "loss": 1.0568,
+      "step": 5333
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0051272112655984e-05,
+      "loss": 1.0874,
+      "step": 5334
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.005045801961873e-05,
+      "loss": 1.0907,
+      "step": 5335
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0049643774936247e-05,
+      "loss": 1.0329,
+      "step": 5336
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0048829378623876e-05,
+      "loss": 1.0604,
+      "step": 5337
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.004801483069695e-05,
+      "loss": 1.0798,
+      "step": 5338
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.004720013117082e-05,
+      "loss": 0.9992,
+      "step": 5339
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0046385280060833e-05,
+      "loss": 1.061,
+      "step": 5340
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0045570277382333e-05,
+      "loss": 1.0888,
+      "step": 5341
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0044755123150674e-05,
+      "loss": 1.1077,
+      "step": 5342
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.004393981738121e-05,
+      "loss": 1.0437,
+      "step": 5343
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.00431243600893e-05,
+      "loss": 1.0777,
+      "step": 5344
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0042308751290307e-05,
+      "loss": 1.0693,
+      "step": 5345
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0041492990999596e-05,
+      "loss": 1.2048,
+      "step": 5346
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0040677079232522e-05,
+      "loss": 1.1183,
+      "step": 5347
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0039861016004465e-05,
+      "loss": 1.13,
+      "step": 5348
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0039044801330796e-05,
+      "loss": 1.098,
+      "step": 5349
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0038228435226888e-05,
+      "loss": 1.0986,
+      "step": 5350
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0037411917708115e-05,
+      "loss": 1.1071,
+      "step": 5351
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0036595248789864e-05,
+      "loss": 1.0518,
+      "step": 5352
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0035778428487512e-05,
+      "loss": 1.0031,
+      "step": 5353
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0034961456816453e-05,
+      "loss": 1.059,
+      "step": 5354
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.003414433379207e-05,
+      "loss": 1.1162,
+      "step": 5355
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0033327059429757e-05,
+      "loss": 1.0392,
+      "step": 5356
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0032509633744908e-05,
+      "loss": 1.0157,
+      "step": 5357
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0031692056752927e-05,
+      "loss": 1.0887,
+      "step": 5358
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0030874328469203e-05,
+      "loss": 0.966,
+      "step": 5359
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.003005644890915e-05,
+      "loss": 1.0643,
+      "step": 5360
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.002923841808817e-05,
+      "loss": 1.0926,
+      "step": 5361
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.002842023602167e-05,
+      "loss": 0.9828,
+      "step": 5362
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0027601902725066e-05,
+      "loss": 1.0623,
+      "step": 5363
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.002678341821377e-05,
+      "loss": 1.1287,
+      "step": 5364
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0025964782503198e-05,
+      "loss": 1.0931,
+      "step": 5365
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0025145995608776e-05,
+      "loss": 0.9804,
+      "step": 5366
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0024327057545926e-05,
+      "loss": 1.0107,
+      "step": 5367
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.002350796833007e-05,
+      "loss": 1.1258,
+      "step": 5368
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.002268872797664e-05,
+      "loss": 1.1181,
+      "step": 5369
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0021869336501068e-05,
+      "loss": 1.1348,
+      "step": 5370
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.002104979391879e-05,
+      "loss": 1.0202,
+      "step": 5371
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.002023010024524e-05,
+      "loss": 1.0901,
+      "step": 5372
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0019410255495855e-05,
+      "loss": 1.1087,
+      "step": 5373
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0018590259686093e-05,
+      "loss": 1.1266,
+      "step": 5374
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0017770112831387e-05,
+      "loss": 1.1175,
+      "step": 5375
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.001694981494719e-05,
+      "loss": 1.078,
+      "step": 5376
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0016129366048953e-05,
+      "loss": 1.0334,
+      "step": 5377
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0015308766152132e-05,
+      "loss": 1.1003,
+      "step": 5378
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0014488015272184e-05,
+      "loss": 1.1519,
+      "step": 5379
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0013667113424567e-05,
+      "loss": 1.0456,
+      "step": 5380
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0012846060624752e-05,
+      "loss": 0.9901,
+      "step": 5381
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0012024856888192e-05,
+      "loss": 1.1131,
+      "step": 5382
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.001120350223037e-05,
+      "loss": 1.0259,
+      "step": 5383
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0010381996666747e-05,
+      "loss": 1.0842,
+      "step": 5384
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0009560340212802e-05,
+      "loss": 1.1206,
+      "step": 5385
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0008738532884014e-05,
+      "loss": 1.0976,
+      "step": 5386
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.000791657469586e-05,
+      "loss": 1.0997,
+      "step": 5387
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0007094465663824e-05,
+      "loss": 1.0225,
+      "step": 5388
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.000627220580339e-05,
+      "loss": 1.1174,
+      "step": 5389
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0005449795130052e-05,
+      "loss": 1.0851,
+      "step": 5390
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0004627233659302e-05,
+      "loss": 1.0489,
+      "step": 5391
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0003804521406625e-05,
+      "loss": 1.0879,
+      "step": 5392
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.000298165838753e-05,
+      "loss": 1.0034,
+      "step": 5393
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.000215864461751e-05,
+      "loss": 1.0564,
+      "step": 5394
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0001335480112068e-05,
+      "loss": 1.0003,
+      "step": 5395
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0000512164886712e-05,
+      "loss": 1.0819,
+      "step": 5396
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9999688698956946e-05,
+      "loss": 1.1253,
+      "step": 5397
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.999886508233829e-05,
+      "loss": 1.0615,
+      "step": 5398
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9998041315046254e-05,
+      "loss": 1.0555,
+      "step": 5399
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9997217397096356e-05,
+      "loss": 1.0131,
+      "step": 5400
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.999639332850411e-05,
+      "loss": 1.1238,
+      "step": 5401
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9995569109285052e-05,
+      "loss": 1.1704,
+      "step": 5402
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9994744739454695e-05,
+      "loss": 1.101,
+      "step": 5403
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.999392021902857e-05,
+      "loss": 1.0236,
+      "step": 5404
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9993095548022216e-05,
+      "loss": 1.0596,
+      "step": 5405
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.999227072645116e-05,
+      "loss": 1.102,
+      "step": 5406
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.999144575433094e-05,
+      "loss": 1.1357,
+      "step": 5407
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9990620631677093e-05,
+      "loss": 1.1427,
+      "step": 5408
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9989795358505173e-05,
+      "loss": 1.0811,
+      "step": 5409
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9988969934830714e-05,
+      "loss": 1.0149,
+      "step": 5410
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9988144360669265e-05,
+      "loss": 1.1412,
+      "step": 5411
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987318636036383e-05,
+      "loss": 1.0471,
+      "step": 5412
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998649276094762e-05,
+      "loss": 1.0796,
+      "step": 5413
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985666735418533e-05,
+      "loss": 1.1223,
+      "step": 5414
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9984840559464684e-05,
+      "loss": 1.0968,
+      "step": 5415
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998401423310163e-05,
+      "loss": 1.0953,
+      "step": 5416
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9983187756344934e-05,
+      "loss": 1.1185,
+      "step": 5417
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9982361129210174e-05,
+      "loss": 1.0505,
+      "step": 5418
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998153435171292e-05,
+      "loss": 1.0731,
+      "step": 5419
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9980707423868737e-05,
+      "loss": 1.114,
+      "step": 5420
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9979880345693208e-05,
+      "loss": 0.9753,
+      "step": 5421
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.997905311720191e-05,
+      "loss": 1.049,
+      "step": 5422
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.997822573841043e-05,
+      "loss": 1.0449,
+      "step": 5423
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9977398209334348e-05,
+      "loss": 1.1186,
+      "step": 5424
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9976570529989256e-05,
+      "loss": 1.0787,
+      "step": 5425
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9975742700390742e-05,
+      "loss": 1.1216,
+      "step": 5426
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.99749147205544e-05,
+      "loss": 1.1642,
+      "step": 5427
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9974086590495827e-05,
+      "loss": 1.0484,
+      "step": 5428
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9973258310230624e-05,
+      "loss": 1.1814,
+      "step": 5429
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.997242987977439e-05,
+      "loss": 1.0716,
+      "step": 5430
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9971601299142733e-05,
+      "loss": 1.0755,
+      "step": 5431
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9970772568351256e-05,
+      "loss": 1.1002,
+      "step": 5432
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9969943687415576e-05,
+      "loss": 1.0979,
+      "step": 5433
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.996911465635131e-05,
+      "loss": 1.0274,
+      "step": 5434
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.996828547517406e-05,
+      "loss": 1.1267,
+      "step": 5435
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9967456143899454e-05,
+      "loss": 1.057,
+      "step": 5436
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9966626662543114e-05,
+      "loss": 1.0686,
+      "step": 5437
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9965797031120668e-05,
+      "loss": 1.0728,
+      "step": 5438
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9964967249647737e-05,
+      "loss": 1.0261,
+      "step": 5439
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9964137318139953e-05,
+      "loss": 1.012,
+      "step": 5440
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.996330723661295e-05,
+      "loss": 1.0948,
+      "step": 5441
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9962477005082368e-05,
+      "loss": 1.0892,
+      "step": 5442
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9961646623563837e-05,
+      "loss": 1.0487,
+      "step": 5443
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.996081609207301e-05,
+      "loss": 1.1433,
+      "step": 5444
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9959985410625524e-05,
+      "loss": 1.0811,
+      "step": 5445
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9959154579237028e-05,
+      "loss": 1.0631,
+      "step": 5446
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.995832359792317e-05,
+      "loss": 1.0941,
+      "step": 5447
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9957492466699607e-05,
+      "loss": 1.1358,
+      "step": 5448
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9956661185581996e-05,
+      "loss": 1.1023,
+      "step": 5449
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.995582975458599e-05,
+      "loss": 1.0716,
+      "step": 5450
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9954998173727256e-05,
+      "loss": 1.1044,
+      "step": 5451
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9954166443021457e-05,
+      "loss": 1.0999,
+      "step": 5452
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9953334562484254e-05,
+      "loss": 1.0976,
+      "step": 5453
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9952502532131327e-05,
+      "loss": 1.1028,
+      "step": 5454
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9951670351978343e-05,
+      "loss": 1.0521,
+      "step": 5455
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.995083802204098e-05,
+      "loss": 1.0981,
+      "step": 5456
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9950005542334912e-05,
+      "loss": 1.0887,
+      "step": 5457
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9949172912875826e-05,
+      "loss": 1.0727,
+      "step": 5458
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9948340133679403e-05,
+      "loss": 1.106,
+      "step": 5459
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9947507204761332e-05,
+      "loss": 1.0518,
+      "step": 5460
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9946674126137296e-05,
+      "loss": 1.0273,
+      "step": 5461
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9945840897823002e-05,
+      "loss": 1.0898,
+      "step": 5462
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.994500751983413e-05,
+      "loss": 1.0763,
+      "step": 5463
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9944173992186382e-05,
+      "loss": 1.0326,
+      "step": 5464
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9943340314895468e-05,
+      "loss": 1.0785,
+      "step": 5465
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.994250648797708e-05,
+      "loss": 1.0589,
+      "step": 5466
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9941672511446933e-05,
+      "loss": 1.0471,
+      "step": 5467
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9940838385320734e-05,
+      "loss": 0.9993,
+      "step": 5468
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9940004109614195e-05,
+      "loss": 0.9539,
+      "step": 5469
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.993916968434303e-05,
+      "loss": 1.0353,
+      "step": 5470
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9938335109522957e-05,
+      "loss": 1.0648,
+      "step": 5471
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9937500385169696e-05,
+      "loss": 1.0761,
+      "step": 5472
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9936665511298976e-05,
+      "loss": 1.125,
+      "step": 5473
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9935830487926518e-05,
+      "loss": 1.0981,
+      "step": 5474
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.993499531506805e-05,
+      "loss": 1.0622,
+      "step": 5475
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9934159992739315e-05,
+      "loss": 1.1586,
+      "step": 5476
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9933324520956034e-05,
+      "loss": 1.0733,
+      "step": 5477
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.993248889973395e-05,
+      "loss": 1.1049,
+      "step": 5478
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9931653129088803e-05,
+      "loss": 1.0994,
+      "step": 5479
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.993081720903634e-05,
+      "loss": 1.1213,
+      "step": 5480
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9929981139592302e-05,
+      "loss": 1.0944,
+      "step": 5481
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.992914492077244e-05,
+      "loss": 1.0609,
+      "step": 5482
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9928308552592506e-05,
+      "loss": 1.0703,
+      "step": 5483
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9927472035068254e-05,
+      "loss": 1.0647,
+      "step": 5484
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9926635368215447e-05,
+      "loss": 1.0796,
+      "step": 5485
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9925798552049833e-05,
+      "loss": 1.0923,
+      "step": 5486
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9924961586587182e-05,
+      "loss": 1.0345,
+      "step": 5487
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9924124471843264e-05,
+      "loss": 1.1252,
+      "step": 5488
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9923287207833842e-05,
+      "loss": 1.0876,
+      "step": 5489
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9922449794574687e-05,
+      "loss": 1.0977,
+      "step": 5490
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9921612232081578e-05,
+      "loss": 1.0908,
+      "step": 5491
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9920774520370283e-05,
+      "loss": 1.1677,
+      "step": 5492
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9919936659456594e-05,
+      "loss": 0.9966,
+      "step": 5493
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9919098649356282e-05,
+      "loss": 1.0708,
+      "step": 5494
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9918260490085145e-05,
+      "loss": 1.0412,
+      "step": 5495
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9917422181658955e-05,
+      "loss": 1.0646,
+      "step": 5496
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.991658372409352e-05,
+      "loss": 1.1221,
+      "step": 5497
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.991574511740462e-05,
+      "loss": 1.0497,
+      "step": 5498
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9914906361608063e-05,
+      "loss": 1.0848,
+      "step": 5499
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.991406745671964e-05,
+      "loss": 1.0703,
+      "step": 5500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.991322840275516e-05,
+      "loss": 1.0419,
+      "step": 5501
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9912389199730422e-05,
+      "loss": 0.9651,
+      "step": 5502
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9911549847661237e-05,
+      "loss": 1.1594,
+      "step": 5503
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9910710346563418e-05,
+      "loss": 1.0634,
+      "step": 5504
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.990987069645277e-05,
+      "loss": 0.9887,
+      "step": 5505
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.990903089734512e-05,
+      "loss": 1.0606,
+      "step": 5506
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.990819094925628e-05,
+      "loss": 1.0974,
+      "step": 5507
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.990735085220208e-05,
+      "loss": 1.054,
+      "step": 5508
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9906510606198333e-05,
+      "loss": 1.1654,
+      "step": 5509
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9905670211260875e-05,
+      "loss": 1.1536,
+      "step": 5510
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9904829667405535e-05,
+      "loss": 1.1083,
+      "step": 5511
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9903988974648146e-05,
+      "loss": 1.028,
+      "step": 5512
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9903148133004543e-05,
+      "loss": 1.0461,
+      "step": 5513
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9902307142490562e-05,
+      "loss": 1.1215,
+      "step": 5514
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.990146600312205e-05,
+      "loss": 1.14,
+      "step": 5515
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.990062471491485e-05,
+      "loss": 1.087,
+      "step": 5516
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989978327788481e-05,
+      "loss": 1.1138,
+      "step": 5517
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989894169204778e-05,
+      "loss": 1.0483,
+      "step": 5518
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9898099957419608e-05,
+      "loss": 1.1839,
+      "step": 5519
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9897258074016153e-05,
+      "loss": 1.0582,
+      "step": 5520
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9896416041853275e-05,
+      "loss": 1.0674,
+      "step": 5521
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989557386094683e-05,
+      "loss": 1.0934,
+      "step": 5522
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9894731531312688e-05,
+      "loss": 1.1442,
+      "step": 5523
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9893889052966714e-05,
+      "loss": 1.0798,
+      "step": 5524
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9893046425924778e-05,
+      "loss": 1.0801,
+      "step": 5525
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989220365020275e-05,
+      "loss": 1.1151,
+      "step": 5526
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9891360725816503e-05,
+      "loss": 1.0198,
+      "step": 5527
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9890517652781925e-05,
+      "loss": 1.1205,
+      "step": 5528
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9889674431114887e-05,
+      "loss": 1.0823,
+      "step": 5529
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9888831060831277e-05,
+      "loss": 1.0992,
+      "step": 5530
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988798754194698e-05,
+      "loss": 1.0268,
+      "step": 5531
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9887143874477888e-05,
+      "loss": 1.1826,
+      "step": 5532
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988630005843989e-05,
+      "loss": 1.1475,
+      "step": 5533
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9885456093848883e-05,
+      "loss": 1.075,
+      "step": 5534
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9884611980720763e-05,
+      "loss": 1.0888,
+      "step": 5535
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988376771907143e-05,
+      "loss": 1.0394,
+      "step": 5536
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988292330891679e-05,
+      "loss": 1.1309,
+      "step": 5537
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9882078750272742e-05,
+      "loss": 1.1027,
+      "step": 5538
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9881234043155205e-05,
+      "loss": 1.0751,
+      "step": 5539
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9880389187580085e-05,
+      "loss": 1.1154,
+      "step": 5540
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.98795441835633e-05,
+      "loss": 1.0674,
+      "step": 5541
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9878699031120757e-05,
+      "loss": 1.0973,
+      "step": 5542
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9877853730268388e-05,
+      "loss": 1.0916,
+      "step": 5543
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9877008281022108e-05,
+      "loss": 1.0679,
+      "step": 5544
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9876162683397848e-05,
+      "loss": 1.1389,
+      "step": 5545
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9875316937411532e-05,
+      "loss": 1.0917,
+      "step": 5546
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9874471043079094e-05,
+      "loss": 1.0791,
+      "step": 5547
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987362500041647e-05,
+      "loss": 1.0878,
+      "step": 5548
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987277880943959e-05,
+      "loss": 1.0791,
+      "step": 5549
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9871932470164397e-05,
+      "loss": 1.1461,
+      "step": 5550
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9871085982606838e-05,
+      "loss": 1.2388,
+      "step": 5551
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9870239346782853e-05,
+      "loss": 1.0847,
+      "step": 5552
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986939256270839e-05,
+      "loss": 1.1181,
+      "step": 5553
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9868545630399397e-05,
+      "loss": 1.0368,
+      "step": 5554
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9867698549871835e-05,
+      "loss": 1.0733,
+      "step": 5555
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9866851321141656e-05,
+      "loss": 1.0539,
+      "step": 5556
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986600394422482e-05,
+      "loss": 1.0817,
+      "step": 5557
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986515641913729e-05,
+      "loss": 1.113,
+      "step": 5558
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9864308745895025e-05,
+      "loss": 1.0887,
+      "step": 5559
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9863460924514e-05,
+      "loss": 1.1283,
+      "step": 5560
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986261295501018e-05,
+      "loss": 1.079,
+      "step": 5561
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9861764837399545e-05,
+      "loss": 1.0118,
+      "step": 5562
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9860916571698063e-05,
+      "loss": 1.0923,
+      "step": 5563
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9860068157921713e-05,
+      "loss": 1.1572,
+      "step": 5564
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9859219596086484e-05,
+      "loss": 1.128,
+      "step": 5565
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9858370886208354e-05,
+      "loss": 1.0436,
+      "step": 5566
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.985752202830331e-05,
+      "loss": 1.128,
+      "step": 5567
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.985667302238734e-05,
+      "loss": 1.0188,
+      "step": 5568
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.985582386847644e-05,
+      "loss": 1.0302,
+      "step": 5569
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.985497456658661e-05,
+      "loss": 1.025,
+      "step": 5570
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.985412511673384e-05,
+      "loss": 1.0591,
+      "step": 5571
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9853275518934136e-05,
+      "loss": 1.0962,
+      "step": 5572
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.98524257732035e-05,
+      "loss": 1.0372,
+      "step": 5573
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9851575879557938e-05,
+      "loss": 1.0213,
+      "step": 5574
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.985072583801346e-05,
+      "loss": 1.1187,
+      "step": 5575
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9849875648586077e-05,
+      "loss": 1.1138,
+      "step": 5576
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9849025311291805e-05,
+      "loss": 1.0572,
+      "step": 5577
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9848174826146663e-05,
+      "loss": 1.1422,
+      "step": 5578
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9847324193166664e-05,
+      "loss": 0.9942,
+      "step": 5579
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9846473412367846e-05,
+      "loss": 1.1289,
+      "step": 5580
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.984562248376622e-05,
+      "loss": 1.1105,
+      "step": 5581
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9844771407377824e-05,
+      "loss": 1.1718,
+      "step": 5582
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9843920183218684e-05,
+      "loss": 1.0056,
+      "step": 5583
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.984306881130484e-05,
+      "loss": 1.0291,
+      "step": 5584
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9842217291652323e-05,
+      "loss": 1.1697,
+      "step": 5585
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9841365624277177e-05,
+      "loss": 1.1055,
+      "step": 5586
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9840513809195443e-05,
+      "loss": 1.1434,
+      "step": 5587
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.983966184642317e-05,
+      "loss": 1.1379,
+      "step": 5588
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9838809735976406e-05,
+      "loss": 1.1607,
+      "step": 5589
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9837957477871192e-05,
+      "loss": 1.1003,
+      "step": 5590
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9837105072123596e-05,
+      "loss": 1.1148,
+      "step": 5591
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9836252518749662e-05,
+      "loss": 1.0778,
+      "step": 5592
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9835399817765462e-05,
+      "loss": 1.0,
+      "step": 5593
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9834546969187045e-05,
+      "loss": 1.1272,
+      "step": 5594
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9833693973030488e-05,
+      "loss": 1.0934,
+      "step": 5595
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9832840829311853e-05,
+      "loss": 1.1522,
+      "step": 5596
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.983198753804721e-05,
+      "loss": 1.1084,
+      "step": 5597
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9831134099252634e-05,
+      "loss": 0.9969,
+      "step": 5598
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.98302805129442e-05,
+      "loss": 1.1035,
+      "step": 5599
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9829426779137986e-05,
+      "loss": 1.0696,
+      "step": 5600
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9828572897850078e-05,
+      "loss": 1.0716,
+      "step": 5601
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9827718869096555e-05,
+      "loss": 1.2011,
+      "step": 5602
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9826864692893506e-05,
+      "loss": 1.0724,
+      "step": 5603
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9826010369257024e-05,
+      "loss": 1.0606,
+      "step": 5604
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9825155898203195e-05,
+      "loss": 1.0697,
+      "step": 5605
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9824301279748127e-05,
+      "loss": 1.1112,
+      "step": 5606
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9823446513907903e-05,
+      "loss": 1.1378,
+      "step": 5607
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9822591600698633e-05,
+      "loss": 1.0858,
+      "step": 5608
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.982173654013642e-05,
+      "loss": 1.0873,
+      "step": 5609
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9820881332237367e-05,
+      "loss": 1.1242,
+      "step": 5610
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9820025977017588e-05,
+      "loss": 1.181,
+      "step": 5611
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9819170474493196e-05,
+      "loss": 1.1204,
+      "step": 5612
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.98183148246803e-05,
+      "loss": 1.0819,
+      "step": 5613
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9817459027595025e-05,
+      "loss": 1.1407,
+      "step": 5614
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9816603083253483e-05,
+      "loss": 1.132,
+      "step": 5615
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9815746991671805e-05,
+      "loss": 1.0075,
+      "step": 5616
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.981489075286611e-05,
+      "loss": 1.1444,
+      "step": 5617
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9814034366852538e-05,
+      "loss": 1.0922,
+      "step": 5618
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9813177833647207e-05,
+      "loss": 1.052,
+      "step": 5619
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.981232115326626e-05,
+      "loss": 1.0892,
+      "step": 5620
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9811464325725832e-05,
+      "loss": 1.1796,
+      "step": 5621
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9810607351042063e-05,
+      "loss": 1.0503,
+      "step": 5622
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9809750229231098e-05,
+      "loss": 1.106,
+      "step": 5623
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9808892960309076e-05,
+      "loss": 1.1149,
+      "step": 5624
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.980803554429215e-05,
+      "loss": 1.0557,
+      "step": 5625
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.980717798119647e-05,
+      "loss": 1.1374,
+      "step": 5626
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9806320271038195e-05,
+      "loss": 1.1254,
+      "step": 5627
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9805462413833472e-05,
+      "loss": 1.0393,
+      "step": 5628
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9804604409598466e-05,
+      "loss": 1.1795,
+      "step": 5629
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.980374625834934e-05,
+      "loss": 1.1392,
+      "step": 5630
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.980288796010225e-05,
+      "loss": 1.0531,
+      "step": 5631
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.980202951487338e-05,
+      "loss": 1.0148,
+      "step": 5632
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9801170922678885e-05,
+      "loss": 1.1286,
+      "step": 5633
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9800312183534947e-05,
+      "loss": 1.0572,
+      "step": 5634
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9799453297457738e-05,
+      "loss": 1.0916,
+      "step": 5635
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9798594264463436e-05,
+      "loss": 1.0933,
+      "step": 5636
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9797735084568228e-05,
+      "loss": 1.1414,
+      "step": 5637
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.979687575778829e-05,
+      "loss": 1.1493,
+      "step": 5638
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9796016284139816e-05,
+      "loss": 1.1287,
+      "step": 5639
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9795156663638994e-05,
+      "loss": 1.1616,
+      "step": 5640
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9794296896302015e-05,
+      "loss": 1.0859,
+      "step": 5641
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9793436982145074e-05,
+      "loss": 1.0551,
+      "step": 5642
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9792576921184375e-05,
+      "loss": 1.097,
+      "step": 5643
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9791716713436108e-05,
+      "loss": 1.1524,
+      "step": 5644
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9790856358916486e-05,
+      "loss": 1.1021,
+      "step": 5645
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.978999585764171e-05,
+      "loss": 1.0875,
+      "step": 5646
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9789135209627992e-05,
+      "loss": 1.0549,
+      "step": 5647
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9788274414891544e-05,
+      "loss": 1.1317,
+      "step": 5648
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9787413473448576e-05,
+      "loss": 1.0931,
+      "step": 5649
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9786552385315315e-05,
+      "loss": 1.1059,
+      "step": 5650
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.978569115050797e-05,
+      "loss": 1.0784,
+      "step": 5651
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.978482976904277e-05,
+      "loss": 1.1327,
+      "step": 5652
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.978396824093594e-05,
+      "loss": 1.0603,
+      "step": 5653
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.978310656620371e-05,
+      "loss": 1.1059,
+      "step": 5654
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.978224474486231e-05,
+      "loss": 1.0748,
+      "step": 5655
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.978138277692797e-05,
+      "loss": 1.0917,
+      "step": 5656
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9780520662416935e-05,
+      "loss": 1.0914,
+      "step": 5657
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.977965840134544e-05,
+      "loss": 1.0244,
+      "step": 5658
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9778795993729724e-05,
+      "loss": 1.0716,
+      "step": 5659
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9777933439586037e-05,
+      "loss": 1.1418,
+      "step": 5660
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9777070738930627e-05,
+      "loss": 1.1419,
+      "step": 5661
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9776207891779737e-05,
+      "loss": 1.0913,
+      "step": 5662
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9775344898149628e-05,
+      "loss": 1.0447,
+      "step": 5663
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9774481758056554e-05,
+      "loss": 1.1279,
+      "step": 5664
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9773618471516776e-05,
+      "loss": 1.1183,
+      "step": 5665
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9772755038546553e-05,
+      "loss": 1.0802,
+      "step": 5666
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.977189145916215e-05,
+      "loss": 1.1971,
+      "step": 5667
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.977102773337983e-05,
+      "loss": 1.1198,
+      "step": 5668
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.977016386121587e-05,
+      "loss": 1.1119,
+      "step": 5669
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9769299842686538e-05,
+      "loss": 1.0996,
+      "step": 5670
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9768435677808112e-05,
+      "loss": 1.1178,
+      "step": 5671
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.976757136659687e-05,
+      "loss": 1.0937,
+      "step": 5672
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.976670690906909e-05,
+      "loss": 1.11,
+      "step": 5673
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9765842305241056e-05,
+      "loss": 1.176,
+      "step": 5674
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9764977555129054e-05,
+      "loss": 1.1095,
+      "step": 5675
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.976411265874938e-05,
+      "loss": 1.0274,
+      "step": 5676
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9763247616118318e-05,
+      "loss": 1.0549,
+      "step": 5677
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.976238242725217e-05,
+      "loss": 1.0465,
+      "step": 5678
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9761517092167224e-05,
+      "loss": 1.058,
+      "step": 5679
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9760651610879787e-05,
+      "loss": 1.0748,
+      "step": 5680
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.975978598340616e-05,
+      "loss": 1.1046,
+      "step": 5681
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9758920209762647e-05,
+      "loss": 1.1081,
+      "step": 5682
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9758054289965562e-05,
+      "loss": 1.2009,
+      "step": 5683
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9757188224031207e-05,
+      "loss": 1.1544,
+      "step": 5684
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9756322011975907e-05,
+      "loss": 1.063,
+      "step": 5685
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.975545565381597e-05,
+      "loss": 1.2123,
+      "step": 5686
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9754589149567725e-05,
+      "loss": 0.9974,
+      "step": 5687
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.975372249924748e-05,
+      "loss": 1.0794,
+      "step": 5688
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.975285570287157e-05,
+      "loss": 1.1498,
+      "step": 5689
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9751988760456318e-05,
+      "loss": 1.1144,
+      "step": 5690
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9751121672018063e-05,
+      "loss": 1.1467,
+      "step": 5691
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9750254437573128e-05,
+      "loss": 1.0851,
+      "step": 5692
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9749387057137853e-05,
+      "loss": 1.1214,
+      "step": 5693
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.974851953072858e-05,
+      "loss": 1.0889,
+      "step": 5694
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9747651858361644e-05,
+      "loss": 1.1539,
+      "step": 5695
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.974678404005339e-05,
+      "loss": 1.0752,
+      "step": 5696
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9745916075820173e-05,
+      "loss": 1.1042,
+      "step": 5697
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9745047965678333e-05,
+      "loss": 1.133,
+      "step": 5698
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.974417970964423e-05,
+      "loss": 1.1911,
+      "step": 5699
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9743311307734213e-05,
+      "loss": 1.1442,
+      "step": 5700
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9742442759964642e-05,
+      "loss": 1.1535,
+      "step": 5701
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9741574066351882e-05,
+      "loss": 1.1792,
+      "step": 5702
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.974070522691229e-05,
+      "loss": 1.142,
+      "step": 5703
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9739836241662236e-05,
+      "loss": 1.2042,
+      "step": 5704
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.973896711061809e-05,
+      "loss": 1.1715,
+      "step": 5705
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.973809783379622e-05,
+      "loss": 1.1907,
+      "step": 5706
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9737228411213003e-05,
+      "loss": 1.1186,
+      "step": 5707
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.973635884288481e-05,
+      "loss": 1.1824,
+      "step": 5708
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9735489128828034e-05,
+      "loss": 1.0681,
+      "step": 5709
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9734619269059044e-05,
+      "loss": 1.0756,
+      "step": 5710
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9733749263594235e-05,
+      "loss": 1.0913,
+      "step": 5711
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9732879112449988e-05,
+      "loss": 1.0882,
+      "step": 5712
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.97320088156427e-05,
+      "loss": 1.1422,
+      "step": 5713
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9731138373188763e-05,
+      "loss": 1.0831,
+      "step": 5714
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.973026778510457e-05,
+      "loss": 1.103,
+      "step": 5715
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9729397051406524e-05,
+      "loss": 1.2007,
+      "step": 5716
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9728526172111024e-05,
+      "loss": 1.0272,
+      "step": 5717
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.972765514723448e-05,
+      "loss": 1.0749,
+      "step": 5718
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9726783976793288e-05,
+      "loss": 1.0556,
+      "step": 5719
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.972591266080387e-05,
+      "loss": 1.0741,
+      "step": 5720
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9725041199282638e-05,
+      "loss": 1.0934,
+      "step": 5721
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9724169592245998e-05,
+      "loss": 1.1921,
+      "step": 5722
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9723297839710378e-05,
+      "loss": 1.1377,
+      "step": 5723
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9722425941692194e-05,
+      "loss": 1.0661,
+      "step": 5724
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9721553898207868e-05,
+      "loss": 1.1055,
+      "step": 5725
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.972068170927383e-05,
+      "loss": 1.0726,
+      "step": 5726
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.971980937490651e-05,
+      "loss": 1.1702,
+      "step": 5727
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9718936895122345e-05,
+      "loss": 1.1437,
+      "step": 5728
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.971806426993776e-05,
+      "loss": 1.1514,
+      "step": 5729
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9717191499369196e-05,
+      "loss": 1.2046,
+      "step": 5730
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9716318583433094e-05,
+      "loss": 1.1589,
+      "step": 5731
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9715445522145895e-05,
+      "loss": 1.1006,
+      "step": 5732
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9714572315524045e-05,
+      "loss": 1.1117,
+      "step": 5733
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9713698963583995e-05,
+      "loss": 1.1281,
+      "step": 5734
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9712825466342196e-05,
+      "loss": 1.0867,
+      "step": 5735
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.97119518238151e-05,
+      "loss": 1.117,
+      "step": 5736
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9711078036019163e-05,
+      "loss": 1.1656,
+      "step": 5737
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9710204102970845e-05,
+      "loss": 1.1175,
+      "step": 5738
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9709330024686613e-05,
+      "loss": 1.1375,
+      "step": 5739
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9708455801182928e-05,
+      "loss": 1.1603,
+      "step": 5740
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9707581432476255e-05,
+      "loss": 1.2468,
+      "step": 5741
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9706706918583066e-05,
+      "loss": 1.0944,
+      "step": 5742
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9705832259519837e-05,
+      "loss": 1.2006,
+      "step": 5743
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9704957455303043e-05,
+      "loss": 1.1791,
+      "step": 5744
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.970408250594916e-05,
+      "loss": 1.1677,
+      "step": 5745
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9703207411474667e-05,
+      "loss": 1.185,
+      "step": 5746
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9702332171896053e-05,
+      "loss": 1.1471,
+      "step": 5747
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9701456787229806e-05,
+      "loss": 1.091,
+      "step": 5748
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9700581257492416e-05,
+      "loss": 1.1567,
+      "step": 5749
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9699705582700366e-05,
+      "loss": 1.0128,
+      "step": 5750
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.969882976287016e-05,
+      "loss": 1.1272,
+      "step": 5751
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9697953798018293e-05,
+      "loss": 1.1115,
+      "step": 5752
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.969707768816126e-05,
+      "loss": 1.1239,
+      "step": 5753
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9696201433315573e-05,
+      "loss": 1.2337,
+      "step": 5754
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9695325033497737e-05,
+      "loss": 1.1698,
+      "step": 5755
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.969444848872425e-05,
+      "loss": 1.0744,
+      "step": 5756
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.969357179901164e-05,
+      "loss": 1.1432,
+      "step": 5757
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9692694964376405e-05,
+      "loss": 1.1682,
+      "step": 5758
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9691817984835074e-05,
+      "loss": 1.2061,
+      "step": 5759
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.969094086040416e-05,
+      "loss": 1.2124,
+      "step": 5760
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9690063591100186e-05,
+      "loss": 1.226,
+      "step": 5761
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9689186176939676e-05,
+      "loss": 1.1142,
+      "step": 5762
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.968830861793916e-05,
+      "loss": 1.1103,
+      "step": 5763
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.968743091411517e-05,
+      "loss": 1.169,
+      "step": 5764
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9686553065484236e-05,
+      "loss": 1.1184,
+      "step": 5765
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9685675072062895e-05,
+      "loss": 1.1097,
+      "step": 5766
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9684796933867683e-05,
+      "loss": 1.1045,
+      "step": 5767
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.968391865091515e-05,
+      "loss": 1.1257,
+      "step": 5768
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.968304022322183e-05,
+      "loss": 1.1595,
+      "step": 5769
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9682161650804273e-05,
+      "loss": 1.1808,
+      "step": 5770
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.968128293367903e-05,
+      "loss": 1.2866,
+      "step": 5771
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9680404071862654e-05,
+      "loss": 1.059,
+      "step": 5772
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9679525065371698e-05,
+      "loss": 1.1186,
+      "step": 5773
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.967864591422272e-05,
+      "loss": 1.1588,
+      "step": 5774
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.967776661843228e-05,
+      "loss": 1.1005,
+      "step": 5775
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9676887178016943e-05,
+      "loss": 1.1592,
+      "step": 5776
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9676007592993274e-05,
+      "loss": 1.1446,
+      "step": 5777
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9675127863377842e-05,
+      "loss": 1.0672,
+      "step": 5778
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9674247989187215e-05,
+      "loss": 1.134,
+      "step": 5779
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.967336797043797e-05,
+      "loss": 1.1749,
+      "step": 5780
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9672487807146687e-05,
+      "loss": 1.108,
+      "step": 5781
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.967160749932994e-05,
+      "loss": 1.1386,
+      "step": 5782
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.967072704700431e-05,
+      "loss": 1.1431,
+      "step": 5783
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9669846450186385e-05,
+      "loss": 1.1743,
+      "step": 5784
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9668965708892758e-05,
+      "loss": 1.0457,
+      "step": 5785
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9668084823140013e-05,
+      "loss": 1.2108,
+      "step": 5786
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9667203792944742e-05,
+      "loss": 1.13,
+      "step": 5787
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9666322618323547e-05,
+      "loss": 1.1301,
+      "step": 5788
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.966544129929302e-05,
+      "loss": 1.1672,
+      "step": 5789
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9664559835869764e-05,
+      "loss": 1.0891,
+      "step": 5790
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9663678228070386e-05,
+      "loss": 1.0933,
+      "step": 5791
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.966279647591149e-05,
+      "loss": 1.1418,
+      "step": 5792
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9661914579409686e-05,
+      "loss": 1.1003,
+      "step": 5793
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9661032538581588e-05,
+      "loss": 1.1522,
+      "step": 5794
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9660150353443808e-05,
+      "loss": 1.109,
+      "step": 5795
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9659268024012963e-05,
+      "loss": 1.1172,
+      "step": 5796
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9658385550305675e-05,
+      "loss": 1.1946,
+      "step": 5797
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.965750293233857e-05,
+      "loss": 1.0325,
+      "step": 5798
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9656620170128267e-05,
+      "loss": 1.0743,
+      "step": 5799
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.96557372636914e-05,
+      "loss": 1.2404,
+      "step": 5800
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.96548542130446e-05,
+      "loss": 1.0938,
+      "step": 5801
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.96539710182045e-05,
+      "loss": 1.0628,
+      "step": 5802
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9653087679187735e-05,
+      "loss": 1.1301,
+      "step": 5803
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.965220419601094e-05,
+      "loss": 1.0845,
+      "step": 5804
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9651320568690766e-05,
+      "loss": 1.1102,
+      "step": 5805
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9650436797243855e-05,
+      "loss": 1.0361,
+      "step": 5806
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9649552881686852e-05,
+      "loss": 1.141,
+      "step": 5807
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.964866882203641e-05,
+      "loss": 1.0936,
+      "step": 5808
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.964778461830918e-05,
+      "loss": 1.053,
+      "step": 5809
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9646900270521818e-05,
+      "loss": 1.1535,
+      "step": 5810
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9646015778690984e-05,
+      "loss": 1.0724,
+      "step": 5811
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9645131142833337e-05,
+      "loss": 1.1508,
+      "step": 5812
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.964424636296554e-05,
+      "loss": 1.1143,
+      "step": 5813
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.964336143910426e-05,
+      "loss": 1.1605,
+      "step": 5814
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9642476371266167e-05,
+      "loss": 1.1642,
+      "step": 5815
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9641591159467936e-05,
+      "loss": 1.1609,
+      "step": 5816
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9640705803726237e-05,
+      "loss": 1.0795,
+      "step": 5817
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9639820304057746e-05,
+      "loss": 1.0383,
+      "step": 5818
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9638934660479147e-05,
+      "loss": 1.1236,
+      "step": 5819
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9638048873007123e-05,
+      "loss": 1.2307,
+      "step": 5820
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9637162941658357e-05,
+      "loss": 1.0768,
+      "step": 5821
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9636276866449536e-05,
+      "loss": 1.1448,
+      "step": 5822
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.963539064739736e-05,
+      "loss": 1.134,
+      "step": 5823
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.963450428451851e-05,
+      "loss": 1.1663,
+      "step": 5824
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9633617777829686e-05,
+      "loss": 1.1104,
+      "step": 5825
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.963273112734759e-05,
+      "loss": 1.1351,
+      "step": 5826
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.963184433308892e-05,
+      "loss": 1.1205,
+      "step": 5827
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9630957395070388e-05,
+      "loss": 1.119,
+      "step": 5828
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9630070313308693e-05,
+      "loss": 1.1636,
+      "step": 5829
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9629183087820547e-05,
+      "loss": 1.1491,
+      "step": 5830
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9628295718622667e-05,
+      "loss": 1.1553,
+      "step": 5831
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9627408205731763e-05,
+      "loss": 1.168,
+      "step": 5832
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9626520549164552e-05,
+      "loss": 1.1229,
+      "step": 5833
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.962563274893776e-05,
+      "loss": 1.1748,
+      "step": 5834
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.962474480506811e-05,
+      "loss": 1.0924,
+      "step": 5835
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.962385671757232e-05,
+      "loss": 1.3017,
+      "step": 5836
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.962296848646713e-05,
+      "loss": 1.1093,
+      "step": 5837
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.962208011176926e-05,
+      "loss": 1.0537,
+      "step": 5838
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.962119159349545e-05,
+      "loss": 1.0944,
+      "step": 5839
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.962030293166244e-05,
+      "loss": 1.1907,
+      "step": 5840
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9619414126286967e-05,
+      "loss": 1.1556,
+      "step": 5841
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.961852517738577e-05,
+      "loss": 1.1565,
+      "step": 5842
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9617636084975597e-05,
+      "loss": 1.1424,
+      "step": 5843
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9616746849073196e-05,
+      "loss": 1.187,
+      "step": 5844
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.961585746969532e-05,
+      "loss": 1.0714,
+      "step": 5845
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.961496794685871e-05,
+      "loss": 1.1259,
+      "step": 5846
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.961407828058014e-05,
+      "loss": 1.1543,
+      "step": 5847
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9613188470876357e-05,
+      "loss": 1.1853,
+      "step": 5848
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9612298517764122e-05,
+      "loss": 1.1301,
+      "step": 5849
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.96114084212602e-05,
+      "loss": 1.1647,
+      "step": 5850
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9610518181381362e-05,
+      "loss": 1.2365,
+      "step": 5851
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9609627798144374e-05,
+      "loss": 1.1876,
+      "step": 5852
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9608737271566005e-05,
+      "loss": 1.1358,
+      "step": 5853
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.960784660166304e-05,
+      "loss": 1.1757,
+      "step": 5854
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9606955788452244e-05,
+      "loss": 1.1184,
+      "step": 5855
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9606064831950404e-05,
+      "loss": 1.1434,
+      "step": 5856
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9605173732174303e-05,
+      "loss": 1.1448,
+      "step": 5857
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.960428248914072e-05,
+      "loss": 1.1001,
+      "step": 5858
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.960339110286645e-05,
+      "loss": 1.2086,
+      "step": 5859
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9602499573368282e-05,
+      "loss": 1.3051,
+      "step": 5860
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9601607900663012e-05,
+      "loss": 1.0924,
+      "step": 5861
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9600716084767433e-05,
+      "loss": 1.1708,
+      "step": 5862
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.959982412569834e-05,
+      "loss": 1.2215,
+      "step": 5863
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9598932023472547e-05,
+      "loss": 1.2765,
+      "step": 5864
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.959803977810685e-05,
+      "loss": 1.2878,
+      "step": 5865
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.959714738961805e-05,
+      "loss": 1.3411,
+      "step": 5866
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.959625485802297e-05,
+      "loss": 1.1695,
+      "step": 5867
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9595362183338413e-05,
+      "loss": 1.2386,
+      "step": 5868
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9594469365581197e-05,
+      "loss": 1.2361,
+      "step": 5869
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9593576404768147e-05,
+      "loss": 1.2415,
+      "step": 5870
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9592683300916067e-05,
+      "loss": 1.238,
+      "step": 5871
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9591790054041797e-05,
+      "loss": 1.272,
+      "step": 5872
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9590896664162154e-05,
+      "loss": 1.3338,
+      "step": 5873
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9590003131293968e-05,
+      "loss": 1.2718,
+      "step": 5874
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9589109455454074e-05,
+      "loss": 1.3042,
+      "step": 5875
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.95882156366593e-05,
+      "loss": 1.3483,
+      "step": 5876
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9587321674926486e-05,
+      "loss": 1.204,
+      "step": 5877
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.958642757027247e-05,
+      "loss": 1.2744,
+      "step": 5878
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.95855333227141e-05,
+      "loss": 1.3122,
+      "step": 5879
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9584638932268212e-05,
+      "loss": 1.2386,
+      "step": 5880
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9583744398951658e-05,
+      "loss": 1.2706,
+      "step": 5881
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.958284972278129e-05,
+      "loss": 1.2365,
+      "step": 5882
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9581954903773957e-05,
+      "loss": 1.2081,
+      "step": 5883
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.958105994194652e-05,
+      "loss": 1.2494,
+      "step": 5884
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9580164837315827e-05,
+      "loss": 1.325,
+      "step": 5885
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.957926958989875e-05,
+      "loss": 1.2284,
+      "step": 5886
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9578374199712145e-05,
+      "loss": 1.2111,
+      "step": 5887
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9577478666772888e-05,
+      "loss": 1.2045,
+      "step": 5888
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9576582991097835e-05,
+      "loss": 1.2625,
+      "step": 5889
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9575687172703866e-05,
+      "loss": 1.2696,
+      "step": 5890
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.957479121160785e-05,
+      "loss": 1.2803,
+      "step": 5891
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9573895107826673e-05,
+      "loss": 1.32,
+      "step": 5892
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9572998861377205e-05,
+      "loss": 1.3437,
+      "step": 5893
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9572102472276336e-05,
+      "loss": 1.2685,
+      "step": 5894
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9571205940540946e-05,
+      "loss": 1.2685,
+      "step": 5895
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.957030926618793e-05,
+      "loss": 1.289,
+      "step": 5896
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9569412449234168e-05,
+      "loss": 1.2933,
+      "step": 5897
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.956851548969656e-05,
+      "loss": 1.1701,
+      "step": 5898
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9567618387592002e-05,
+      "loss": 1.2608,
+      "step": 5899
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9566721142937387e-05,
+      "loss": 1.2844,
+      "step": 5900
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9565823755749624e-05,
+      "loss": 1.1842,
+      "step": 5901
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9564926226045613e-05,
+      "loss": 1.1853,
+      "step": 5902
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9564028553842258e-05,
+      "loss": 1.1833,
+      "step": 5903
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9563130739156472e-05,
+      "loss": 1.2241,
+      "step": 5904
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.956223278200517e-05,
+      "loss": 1.2351,
+      "step": 5905
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9561334682405258e-05,
+      "loss": 1.2775,
+      "step": 5906
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.956043644037366e-05,
+      "loss": 1.2287,
+      "step": 5907
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9559538055927294e-05,
+      "loss": 1.2933,
+      "step": 5908
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9558639529083084e-05,
+      "loss": 1.1753,
+      "step": 5909
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9557740859857955e-05,
+      "loss": 1.1778,
+      "step": 5910
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9556842048268827e-05,
+      "loss": 1.3339,
+      "step": 5911
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9555943094332648e-05,
+      "loss": 1.2218,
+      "step": 5912
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9555043998066334e-05,
+      "loss": 1.2199,
+      "step": 5913
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9554144759486834e-05,
+      "loss": 1.1935,
+      "step": 5914
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9553245378611084e-05,
+      "loss": 1.1359,
+      "step": 5915
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9552345855456018e-05,
+      "loss": 1.0969,
+      "step": 5916
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9551446190038587e-05,
+      "loss": 1.1872,
+      "step": 5917
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9550546382375736e-05,
+      "loss": 1.212,
+      "step": 5918
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.954964643248441e-05,
+      "loss": 1.1891,
+      "step": 5919
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9548746340381573e-05,
+      "loss": 1.2177,
+      "step": 5920
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.954784610608417e-05,
+      "loss": 1.2472,
+      "step": 5921
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9546945729609163e-05,
+      "loss": 1.2243,
+      "step": 5922
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.954604521097351e-05,
+      "loss": 1.1417,
+      "step": 5923
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9545144550194175e-05,
+      "loss": 1.2006,
+      "step": 5924
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.954424374728812e-05,
+      "loss": 1.0968,
+      "step": 5925
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.954334280227232e-05,
+      "loss": 1.2653,
+      "step": 5926
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.954244171516374e-05,
+      "loss": 1.1859,
+      "step": 5927
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.954154048597936e-05,
+      "loss": 1.324,
+      "step": 5928
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9540639114736146e-05,
+      "loss": 1.2148,
+      "step": 5929
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9539737601451085e-05,
+      "loss": 1.2732,
+      "step": 5930
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.953883594614116e-05,
+      "loss": 1.1548,
+      "step": 5931
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9537934148823355e-05,
+      "loss": 1.2249,
+      "step": 5932
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9537032209514647e-05,
+      "loss": 1.2265,
+      "step": 5933
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9536130128232036e-05,
+      "loss": 1.0771,
+      "step": 5934
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9535227904992514e-05,
+      "loss": 1.1854,
+      "step": 5935
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9534325539813073e-05,
+      "loss": 1.1061,
+      "step": 5936
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9533423032710706e-05,
+      "loss": 1.117,
+      "step": 5937
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9532520383702426e-05,
+      "loss": 1.1783,
+      "step": 5938
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9531617592805225e-05,
+      "loss": 1.2227,
+      "step": 5939
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9530714660036114e-05,
+      "loss": 1.1509,
+      "step": 5940
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9529811585412097e-05,
+      "loss": 1.1523,
+      "step": 5941
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9528908368950188e-05,
+      "loss": 1.1678,
+      "step": 5942
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.95280050106674e-05,
+      "loss": 1.1743,
+      "step": 5943
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9527101510580755e-05,
+      "loss": 1.1575,
+      "step": 5944
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9526197868707265e-05,
+      "loss": 1.1748,
+      "step": 5945
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9525294085063952e-05,
+      "loss": 1.1376,
+      "step": 5946
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.952439015966784e-05,
+      "loss": 1.2142,
+      "step": 5947
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9523486092535963e-05,
+      "loss": 1.1476,
+      "step": 5948
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9522581883685347e-05,
+      "loss": 1.1405,
+      "step": 5949
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.952167753313302e-05,
+      "loss": 1.1121,
+      "step": 5950
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9520773040896022e-05,
+      "loss": 1.2025,
+      "step": 5951
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.951986840699139e-05,
+      "loss": 1.1866,
+      "step": 5952
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9518963631436163e-05,
+      "loss": 1.2093,
+      "step": 5953
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9518058714247384e-05,
+      "loss": 1.0892,
+      "step": 5954
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9517153655442102e-05,
+      "loss": 1.1648,
+      "step": 5955
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9516248455037364e-05,
+      "loss": 1.149,
+      "step": 5956
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9515343113050218e-05,
+      "loss": 1.1531,
+      "step": 5957
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.951443762949772e-05,
+      "loss": 1.1741,
+      "step": 5958
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.951353200439693e-05,
+      "loss": 1.1641,
+      "step": 5959
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9512626237764904e-05,
+      "loss": 1.1845,
+      "step": 5960
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.95117203296187e-05,
+      "loss": 1.1924,
+      "step": 5961
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.951081427997539e-05,
+      "loss": 1.1625,
+      "step": 5962
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9509908088852035e-05,
+      "loss": 1.096,
+      "step": 5963
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9509001756265705e-05,
+      "loss": 1.1092,
+      "step": 5964
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9508095282233475e-05,
+      "loss": 1.1222,
+      "step": 5965
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9507188666772422e-05,
+      "loss": 1.1608,
+      "step": 5966
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.950628190989962e-05,
+      "loss": 1.097,
+      "step": 5967
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.950537501163215e-05,
+      "loss": 1.2114,
+      "step": 5968
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9504467971987098e-05,
+      "loss": 1.1802,
+      "step": 5969
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9503560790981546e-05,
+      "loss": 1.1125,
+      "step": 5970
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9502653468632583e-05,
+      "loss": 1.0811,
+      "step": 5971
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9501746004957304e-05,
+      "loss": 1.1913,
+      "step": 5972
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9500838399972795e-05,
+      "loss": 1.1226,
+      "step": 5973
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.949993065369616e-05,
+      "loss": 1.1078,
+      "step": 5974
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9499022766144493e-05,
+      "loss": 1.0912,
+      "step": 5975
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9498114737334903e-05,
+      "loss": 1.1997,
+      "step": 5976
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9497206567284487e-05,
+      "loss": 1.0382,
+      "step": 5977
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9496298256010355e-05,
+      "loss": 1.1768,
+      "step": 5978
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9495389803529616e-05,
+      "loss": 1.1548,
+      "step": 5979
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.949448120985938e-05,
+      "loss": 1.1021,
+      "step": 5980
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.949357247501677e-05,
+      "loss": 1.1913,
+      "step": 5981
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9492663599018894e-05,
+      "loss": 1.0818,
+      "step": 5982
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9491754581882876e-05,
+      "loss": 1.1617,
+      "step": 5983
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9490845423625842e-05,
+      "loss": 1.1422,
+      "step": 5984
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9489936124264917e-05,
+      "loss": 1.1333,
+      "step": 5985
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9489026683817227e-05,
+      "loss": 1.1165,
+      "step": 5986
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.94881171022999e-05,
+      "loss": 1.1177,
+      "step": 5987
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.948720737973008e-05,
+      "loss": 1.1811,
+      "step": 5988
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.948629751612489e-05,
+      "loss": 1.1956,
+      "step": 5989
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.948538751150148e-05,
+      "loss": 1.0597,
+      "step": 5990
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9484477365876986e-05,
+      "loss": 1.1532,
+      "step": 5991
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9483567079268556e-05,
+      "loss": 1.1393,
+      "step": 5992
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9482656651693328e-05,
+      "loss": 1.179,
+      "step": 5993
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.948174608316846e-05,
+      "loss": 1.1379,
+      "step": 5994
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9480835373711106e-05,
+      "loss": 1.1545,
+      "step": 5995
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9479924523338414e-05,
+      "loss": 1.1146,
+      "step": 5996
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9479013532067547e-05,
+      "loss": 1.1224,
+      "step": 5997
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.947810239991566e-05,
+      "loss": 1.1101,
+      "step": 5998
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9477191126899924e-05,
+      "loss": 1.1314,
+      "step": 5999
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9476279713037493e-05,
+      "loss": 1.1591,
+      "step": 6000
+    },
+    {
+      "epoch": 0.25,
+      "eval_text_loss": 0.8401471972465515,
+      "eval_text_runtime": 1.3956,
+      "eval_text_samples_per_second": 53.025,
+      "eval_text_steps_per_second": 1.433,
+      "step": 6000
+    },
+    {
+      "epoch": 0.25,
+      "eval_svit_loss": 1.2028470039367676,
+      "eval_svit_runtime": 4.8524,
+      "eval_svit_samples_per_second": 140.344,
+      "eval_svit_steps_per_second": 2.267,
+      "step": 6000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9475368158345544e-05,
+      "loss": 1.1368,
+      "step": 6001
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9474456462841246e-05,
+      "loss": 1.076,
+      "step": 6002
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.947354462654177e-05,
+      "loss": 1.1497,
+      "step": 6003
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9472632649464296e-05,
+      "loss": 1.1894,
+      "step": 6004
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9471720531626e-05,
+      "loss": 1.0365,
+      "step": 6005
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.947080827304406e-05,
+      "loss": 1.1586,
+      "step": 6006
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9469895873735672e-05,
+      "loss": 1.1505,
+      "step": 6007
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9468983333718008e-05,
+      "loss": 1.0474,
+      "step": 6008
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.946807065300827e-05,
+      "loss": 1.0858,
+      "step": 6009
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9467157831623637e-05,
+      "loss": 1.184,
+      "step": 6010
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9466244869581317e-05,
+      "loss": 1.0728,
+      "step": 6011
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.94653317668985e-05,
+      "loss": 1.0861,
+      "step": 6012
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9464418523592385e-05,
+      "loss": 1.1032,
+      "step": 6013
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.946350513968018e-05,
+      "loss": 1.1415,
+      "step": 6014
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9462591615179085e-05,
+      "loss": 1.0899,
+      "step": 6015
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9461677950106307e-05,
+      "loss": 1.1257,
+      "step": 6016
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9460764144479064e-05,
+      "loss": 1.1259,
+      "step": 6017
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9459850198314563e-05,
+      "loss": 1.1094,
+      "step": 6018
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.945893611163002e-05,
+      "loss": 1.0793,
+      "step": 6019
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9458021884442662e-05,
+      "loss": 1.0002,
+      "step": 6020
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9457107516769698e-05,
+      "loss": 1.1026,
+      "step": 6021
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9456193008628355e-05,
+      "loss": 1.1366,
+      "step": 6022
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9455278360035864e-05,
+      "loss": 1.1094,
+      "step": 6023
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9454363571009453e-05,
+      "loss": 1.1269,
+      "step": 6024
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9453448641566353e-05,
+      "loss": 1.0347,
+      "step": 6025
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9452533571723794e-05,
+      "loss": 1.1064,
+      "step": 6026
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.945161836149902e-05,
+      "loss": 1.1107,
+      "step": 6027
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9450703010909264e-05,
+      "loss": 1.0933,
+      "step": 6028
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9449787519971777e-05,
+      "loss": 1.0073,
+      "step": 6029
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9448871888703794e-05,
+      "loss": 1.0804,
+      "step": 6030
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.944795611712257e-05,
+      "loss": 1.1162,
+      "step": 6031
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.944704020524535e-05,
+      "loss": 1.0994,
+      "step": 6032
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9446124153089393e-05,
+      "loss": 1.1195,
+      "step": 6033
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9445207960671952e-05,
+      "loss": 1.1461,
+      "step": 6034
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.944429162801028e-05,
+      "loss": 1.1025,
+      "step": 6035
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9443375155121642e-05,
+      "loss": 1.062,
+      "step": 6036
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9442458542023306e-05,
+      "loss": 1.1161,
+      "step": 6037
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.944154178873253e-05,
+      "loss": 1.027,
+      "step": 6038
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9440624895266585e-05,
+      "loss": 1.0278,
+      "step": 6039
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.943970786164275e-05,
+      "loss": 1.0677,
+      "step": 6040
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9438790687878286e-05,
+      "loss": 1.0566,
+      "step": 6041
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.943787337399048e-05,
+      "loss": 1.083,
+      "step": 6042
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9436955919996603e-05,
+      "loss": 1.186,
+      "step": 6043
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9436038325913947e-05,
+      "loss": 1.0658,
+      "step": 6044
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.943512059175979e-05,
+      "loss": 1.1607,
+      "step": 6045
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9434202717551418e-05,
+      "loss": 1.1588,
+      "step": 6046
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9433284703306125e-05,
+      "loss": 1.163,
+      "step": 6047
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9432366549041204e-05,
+      "loss": 1.0441,
+      "step": 6048
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9431448254773944e-05,
+      "loss": 1.0686,
+      "step": 6049
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.943052982052165e-05,
+      "loss": 1.1277,
+      "step": 6050
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9429611246301617e-05,
+      "loss": 1.1529,
+      "step": 6051
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.942869253213115e-05,
+      "loss": 1.0356,
+      "step": 6052
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9427773678027557e-05,
+      "loss": 1.1375,
+      "step": 6053
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.942685468400814e-05,
+      "loss": 1.0363,
+      "step": 6054
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.942593555009022e-05,
+      "loss": 1.1748,
+      "step": 6055
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.94250162762911e-05,
+      "loss": 1.0737,
+      "step": 6056
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.94240968626281e-05,
+      "loss": 1.065,
+      "step": 6057
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9423177309118543e-05,
+      "loss": 1.0395,
+      "step": 6058
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9422257615779747e-05,
+      "loss": 1.0498,
+      "step": 6059
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9421337782629032e-05,
+      "loss": 1.1069,
+      "step": 6060
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9420417809683732e-05,
+      "loss": 1.0428,
+      "step": 6061
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9419497696961173e-05,
+      "loss": 1.0673,
+      "step": 6062
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.941857744447869e-05,
+      "loss": 1.0736,
+      "step": 6063
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9417657052253613e-05,
+      "loss": 1.1318,
+      "step": 6064
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.941673652030328e-05,
+      "loss": 1.0729,
+      "step": 6065
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9415815848645033e-05,
+      "loss": 1.2141,
+      "step": 6066
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9414895037296215e-05,
+      "loss": 1.1622,
+      "step": 6067
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.941397408627417e-05,
+      "loss": 1.1798,
+      "step": 6068
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9413052995596246e-05,
+      "loss": 1.079,
+      "step": 6069
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9412131765279792e-05,
+      "loss": 1.1177,
+      "step": 6070
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9411210395342164e-05,
+      "loss": 1.1453,
+      "step": 6071
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9410288885800717e-05,
+      "loss": 1.0196,
+      "step": 6072
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9409367236672807e-05,
+      "loss": 1.0633,
+      "step": 6073
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9408445447975795e-05,
+      "loss": 1.0513,
+      "step": 6074
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9407523519727046e-05,
+      "loss": 1.1367,
+      "step": 6075
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.940660145194393e-05,
+      "loss": 1.1951,
+      "step": 6076
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9405679244643805e-05,
+      "loss": 1.0756,
+      "step": 6077
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9404756897844056e-05,
+      "loss": 1.1764,
+      "step": 6078
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.940383441156205e-05,
+      "loss": 1.1712,
+      "step": 6079
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9402911785815163e-05,
+      "loss": 1.1153,
+      "step": 6080
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.940198902062077e-05,
+      "loss": 1.0364,
+      "step": 6081
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9401066115996266e-05,
+      "loss": 1.0782,
+      "step": 6082
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.940014307195903e-05,
+      "loss": 1.1351,
+      "step": 6083
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.939921988852644e-05,
+      "loss": 1.1276,
+      "step": 6084
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9398296565715897e-05,
+      "loss": 1.1093,
+      "step": 6085
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.939737310354479e-05,
+      "loss": 1.0476,
+      "step": 6086
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9396449502030512e-05,
+      "loss": 1.1229,
+      "step": 6087
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9395525761190467e-05,
+      "loss": 1.0933,
+      "step": 6088
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9394601881042047e-05,
+      "loss": 1.111,
+      "step": 6089
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.939367786160266e-05,
+      "loss": 1.0918,
+      "step": 6090
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9392753702889715e-05,
+      "loss": 1.0923,
+      "step": 6091
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9391829404920613e-05,
+      "loss": 1.1153,
+      "step": 6092
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9390904967712766e-05,
+      "loss": 1.0943,
+      "step": 6093
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.938998039128359e-05,
+      "loss": 1.1026,
+      "step": 6094
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.93890556756505e-05,
+      "loss": 1.0634,
+      "step": 6095
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9388130820830915e-05,
+      "loss": 1.0635,
+      "step": 6096
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9387205826842257e-05,
+      "loss": 1.081,
+      "step": 6097
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9386280693701952e-05,
+      "loss": 1.0169,
+      "step": 6098
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9385355421427422e-05,
+      "loss": 1.1487,
+      "step": 6099
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.93844300100361e-05,
+      "loss": 1.1279,
+      "step": 6100
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9383504459545412e-05,
+      "loss": 1.0866,
+      "step": 6101
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.93825787699728e-05,
+      "loss": 1.2172,
+      "step": 6102
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9381652941335697e-05,
+      "loss": 1.0863,
+      "step": 6103
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9380726973651542e-05,
+      "loss": 1.174,
+      "step": 6104
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9379800866937775e-05,
+      "loss": 0.9938,
+      "step": 6105
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.937887462121185e-05,
+      "loss": 1.0588,
+      "step": 6106
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9377948236491207e-05,
+      "loss": 1.1591,
+      "step": 6107
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.93770217127933e-05,
+      "loss": 1.1272,
+      "step": 6108
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9376095050135577e-05,
+      "loss": 1.037,
+      "step": 6109
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.93751682485355e-05,
+      "loss": 1.08,
+      "step": 6110
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9374241308010518e-05,
+      "loss": 1.078,
+      "step": 6111
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.93733142285781e-05,
+      "loss": 1.082,
+      "step": 6112
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9372387010255708e-05,
+      "loss": 1.0907,
+      "step": 6113
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9371459653060807e-05,
+      "loss": 1.2186,
+      "step": 6114
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.937053215701086e-05,
+      "loss": 1.1652,
+      "step": 6115
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.936960452212335e-05,
+      "loss": 1.1477,
+      "step": 6116
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9368676748415738e-05,
+      "loss": 1.0861,
+      "step": 6117
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9367748835905505e-05,
+      "loss": 1.1192,
+      "step": 6118
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.936682078461013e-05,
+      "loss": 1.0974,
+      "step": 6119
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9365892594547098e-05,
+      "loss": 1.1292,
+      "step": 6120
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.936496426573389e-05,
+      "loss": 1.0761,
+      "step": 6121
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.936403579818799e-05,
+      "loss": 1.25,
+      "step": 6122
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.93631071919269e-05,
+      "loss": 1.1209,
+      "step": 6123
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9362178446968094e-05,
+      "loss": 0.9749,
+      "step": 6124
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9361249563329075e-05,
+      "loss": 1.1572,
+      "step": 6125
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9360320541027343e-05,
+      "loss": 1.1717,
+      "step": 6126
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9359391380080395e-05,
+      "loss": 1.0753,
+      "step": 6127
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9358462080505735e-05,
+      "loss": 1.0528,
+      "step": 6128
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.935753264232087e-05,
+      "loss": 1.0596,
+      "step": 6129
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.93566030655433e-05,
+      "loss": 1.1137,
+      "step": 6130
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9355673350190537e-05,
+      "loss": 1.117,
+      "step": 6131
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9354743496280104e-05,
+      "loss": 1.1843,
+      "step": 6132
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.93538135038295e-05,
+      "loss": 1.032,
+      "step": 6133
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9352883372856262e-05,
+      "loss": 1.1405,
+      "step": 6134
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9351953103377896e-05,
+      "loss": 1.1669,
+      "step": 6135
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.935102269541193e-05,
+      "loss": 1.1084,
+      "step": 6136
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9350092148975895e-05,
+      "loss": 1.0453,
+      "step": 6137
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9349161464087313e-05,
+      "loss": 1.1371,
+      "step": 6138
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9348230640763715e-05,
+      "loss": 1.0709,
+      "step": 6139
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9347299679022638e-05,
+      "loss": 1.1163,
+      "step": 6140
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9346368578881616e-05,
+      "loss": 1.1926,
+      "step": 6141
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9345437340358195e-05,
+      "loss": 1.1468,
+      "step": 6142
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9344505963469906e-05,
+      "loss": 1.0389,
+      "step": 6143
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9343574448234295e-05,
+      "loss": 1.0826,
+      "step": 6144
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9342642794668918e-05,
+      "loss": 1.1521,
+      "step": 6145
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9341711002791317e-05,
+      "loss": 1.2301,
+      "step": 6146
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9340779072619044e-05,
+      "loss": 1.1386,
+      "step": 6147
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9339847004169654e-05,
+      "loss": 1.0884,
+      "step": 6148
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9338914797460707e-05,
+      "loss": 1.0364,
+      "step": 6149
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.933798245250976e-05,
+      "loss": 1.2013,
+      "step": 6150
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9337049969334376e-05,
+      "loss": 1.1292,
+      "step": 6151
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.933611734795212e-05,
+      "loss": 1.2025,
+      "step": 6152
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9335184588380565e-05,
+      "loss": 1.055,
+      "step": 6153
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9334251690637273e-05,
+      "loss": 1.125,
+      "step": 6154
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.933331865473982e-05,
+      "loss": 1.0783,
+      "step": 6155
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.933238548070578e-05,
+      "loss": 1.0867,
+      "step": 6156
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9331452168552738e-05,
+      "loss": 1.0547,
+      "step": 6157
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9330518718298264e-05,
+      "loss": 1.1675,
+      "step": 6158
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9329585129959954e-05,
+      "loss": 1.0616,
+      "step": 6159
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.932865140355538e-05,
+      "loss": 1.2245,
+      "step": 6160
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9327717539102137e-05,
+      "loss": 1.1554,
+      "step": 6161
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9326783536617818e-05,
+      "loss": 1.0682,
+      "step": 6162
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9325849396120014e-05,
+      "loss": 1.0904,
+      "step": 6163
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.932491511762632e-05,
+      "loss": 1.1298,
+      "step": 6164
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.932398070115434e-05,
+      "loss": 1.0385,
+      "step": 6165
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9323046146721672e-05,
+      "loss": 1.077,
+      "step": 6166
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9322111454345918e-05,
+      "loss": 1.2105,
+      "step": 6167
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.932117662404469e-05,
+      "loss": 1.1455,
+      "step": 6168
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9320241655835592e-05,
+      "loss": 1.1306,
+      "step": 6169
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.931930654973624e-05,
+      "loss": 1.1292,
+      "step": 6170
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9318371305764246e-05,
+      "loss": 1.0359,
+      "step": 6171
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9317435923937225e-05,
+      "loss": 1.115,
+      "step": 6172
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.93165004042728e-05,
+      "loss": 1.0712,
+      "step": 6173
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9315564746788594e-05,
+      "loss": 1.1722,
+      "step": 6174
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9314628951502233e-05,
+      "loss": 1.0735,
+      "step": 6175
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9313693018431337e-05,
+      "loss": 1.0811,
+      "step": 6176
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.931275694759354e-05,
+      "loss": 1.1225,
+      "step": 6177
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.931182073900648e-05,
+      "loss": 1.1242,
+      "step": 6178
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9310884392687786e-05,
+      "loss": 1.1002,
+      "step": 6179
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9309947908655097e-05,
+      "loss": 1.1004,
+      "step": 6180
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9309011286926056e-05,
+      "loss": 1.1052,
+      "step": 6181
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.93080745275183e-05,
+      "loss": 1.0615,
+      "step": 6182
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.930713763044948e-05,
+      "loss": 1.0469,
+      "step": 6183
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.930620059573724e-05,
+      "loss": 1.1165,
+      "step": 6184
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.930526342339924e-05,
+      "loss": 1.1036,
+      "step": 6185
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.930432611345312e-05,
+      "loss": 1.0442,
+      "step": 6186
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9303388665916543e-05,
+      "loss": 1.1884,
+      "step": 6187
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.930245108080717e-05,
+      "loss": 1.1234,
+      "step": 6188
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.930151335814266e-05,
+      "loss": 1.1435,
+      "step": 6189
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9300575497940676e-05,
+      "loss": 1.0241,
+      "step": 6190
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9299637500218887e-05,
+      "loss": 1.2237,
+      "step": 6191
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9298699364994953e-05,
+      "loss": 1.0955,
+      "step": 6192
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9297761092286554e-05,
+      "loss": 1.1667,
+      "step": 6193
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.929682268211136e-05,
+      "loss": 1.1833,
+      "step": 6194
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9295884134487055e-05,
+      "loss": 1.1241,
+      "step": 6195
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.929494544943131e-05,
+      "loss": 1.1873,
+      "step": 6196
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.929400662696181e-05,
+      "loss": 1.0914,
+      "step": 6197
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.929306766709624e-05,
+      "loss": 1.1285,
+      "step": 6198
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9292128569852285e-05,
+      "loss": 1.209,
+      "step": 6199
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9291189335247637e-05,
+      "loss": 1.1412,
+      "step": 6200
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9290249963299987e-05,
+      "loss": 1.1161,
+      "step": 6201
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.928931045402703e-05,
+      "loss": 1.19,
+      "step": 6202
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9288370807446464e-05,
+      "loss": 1.2236,
+      "step": 6203
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.928743102357599e-05,
+      "loss": 1.1229,
+      "step": 6204
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9286491102433305e-05,
+      "loss": 1.094,
+      "step": 6205
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9285551044036118e-05,
+      "loss": 1.1627,
+      "step": 6206
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9284610848402138e-05,
+      "loss": 1.1344,
+      "step": 6207
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9283670515549074e-05,
+      "loss": 1.1616,
+      "step": 6208
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.928273004549464e-05,
+      "loss": 1.1142,
+      "step": 6209
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.928178943825655e-05,
+      "loss": 1.231,
+      "step": 6210
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9280848693852522e-05,
+      "loss": 1.1109,
+      "step": 6211
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9279907812300277e-05,
+      "loss": 1.0493,
+      "step": 6212
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9278966793617542e-05,
+      "loss": 1.1357,
+      "step": 6213
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9278025637822036e-05,
+      "loss": 1.0665,
+      "step": 6214
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.927708434493149e-05,
+      "loss": 1.1779,
+      "step": 6215
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9276142914963636e-05,
+      "loss": 1.1221,
+      "step": 6216
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9275201347936206e-05,
+      "loss": 1.0695,
+      "step": 6217
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9274259643866942e-05,
+      "loss": 1.1029,
+      "step": 6218
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9273317802773572e-05,
+      "loss": 1.1269,
+      "step": 6219
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.927237582467385e-05,
+      "loss": 1.1178,
+      "step": 6220
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9271433709585507e-05,
+      "loss": 1.1961,
+      "step": 6221
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.92704914575263e-05,
+      "loss": 1.1375,
+      "step": 6222
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9269549068513974e-05,
+      "loss": 1.0753,
+      "step": 6223
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9268606542566277e-05,
+      "loss": 1.1325,
+      "step": 6224
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.926766387970097e-05,
+      "loss": 1.1487,
+      "step": 6225
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9266721079935804e-05,
+      "loss": 1.0777,
+      "step": 6226
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9265778143288543e-05,
+      "loss": 1.0317,
+      "step": 6227
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9264835069776946e-05,
+      "loss": 1.2337,
+      "step": 6228
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.926389185941878e-05,
+      "loss": 1.1163,
+      "step": 6229
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9262948512231807e-05,
+      "loss": 1.1062,
+      "step": 6230
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9262005028233804e-05,
+      "loss": 1.0584,
+      "step": 6231
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9261061407442537e-05,
+      "loss": 1.108,
+      "step": 6232
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9260117649875785e-05,
+      "loss": 1.1394,
+      "step": 6233
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9259173755551322e-05,
+      "loss": 1.0787,
+      "step": 6234
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9258229724486928e-05,
+      "loss": 1.0332,
+      "step": 6235
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9257285556700392e-05,
+      "loss": 1.1063,
+      "step": 6236
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9256341252209487e-05,
+      "loss": 1.1284,
+      "step": 6237
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9255396811032015e-05,
+      "loss": 1.121,
+      "step": 6238
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9254452233185757e-05,
+      "loss": 1.1451,
+      "step": 6239
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.925350751868851e-05,
+      "loss": 1.118,
+      "step": 6240
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9252562667558065e-05,
+      "loss": 1.1732,
+      "step": 6241
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9251617679812225e-05,
+      "loss": 1.1127,
+      "step": 6242
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9250672555468786e-05,
+      "loss": 1.0581,
+      "step": 6243
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9249727294545553e-05,
+      "loss": 1.1235,
+      "step": 6244
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.924878189706034e-05,
+      "loss": 1.1863,
+      "step": 6245
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9247836363030936e-05,
+      "loss": 1.0907,
+      "step": 6246
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.924689069247517e-05,
+      "loss": 1.1207,
+      "step": 6247
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9245944885410846e-05,
+      "loss": 1.1215,
+      "step": 6248
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9244998941855786e-05,
+      "loss": 1.1847,
+      "step": 6249
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9244052861827804e-05,
+      "loss": 1.1748,
+      "step": 6250
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.924310664534472e-05,
+      "loss": 1.1393,
+      "step": 6251
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9242160292424363e-05,
+      "loss": 1.1416,
+      "step": 6252
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.924121380308456e-05,
+      "loss": 1.1553,
+      "step": 6253
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9240267177343128e-05,
+      "loss": 1.1322,
+      "step": 6254
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.923932041521791e-05,
+      "loss": 1.1413,
+      "step": 6255
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.923837351672674e-05,
+      "loss": 1.1276,
+      "step": 6256
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9237426481887453e-05,
+      "loss": 1.1412,
+      "step": 6257
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.923647931071788e-05,
+      "loss": 1.0779,
+      "step": 6258
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9235532003235876e-05,
+      "loss": 1.0482,
+      "step": 6259
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9234584559459275e-05,
+      "loss": 1.1297,
+      "step": 6260
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9233636979405928e-05,
+      "loss": 1.0769,
+      "step": 6261
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.923268926309368e-05,
+      "loss": 1.1408,
+      "step": 6262
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9231741410540396e-05,
+      "loss": 1.1533,
+      "step": 6263
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9230793421763915e-05,
+      "loss": 1.1569,
+      "step": 6264
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9229845296782103e-05,
+      "loss": 1.1617,
+      "step": 6265
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9228897035612816e-05,
+      "loss": 1.1487,
+      "step": 6266
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.922794863827392e-05,
+      "loss": 1.0759,
+      "step": 6267
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9227000104783275e-05,
+      "loss": 1.1661,
+      "step": 6268
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.922605143515875e-05,
+      "loss": 1.1495,
+      "step": 6269
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.922510262941822e-05,
+      "loss": 1.1325,
+      "step": 6270
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9224153687579543e-05,
+      "loss": 1.0744,
+      "step": 6271
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9223204609660614e-05,
+      "loss": 1.2466,
+      "step": 6272
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.92222553956793e-05,
+      "loss": 1.1587,
+      "step": 6273
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9221306045653478e-05,
+      "loss": 1.0954,
+      "step": 6274
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9220356559601037e-05,
+      "loss": 1.0368,
+      "step": 6275
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9219406937539862e-05,
+      "loss": 1.1198,
+      "step": 6276
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9218457179487835e-05,
+      "loss": 1.127,
+      "step": 6277
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9217507285462856e-05,
+      "loss": 1.1137,
+      "step": 6278
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9216557255482807e-05,
+      "loss": 1.128,
+      "step": 6279
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.921560708956559e-05,
+      "loss": 1.0791,
+      "step": 6280
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9214656787729105e-05,
+      "loss": 1.1024,
+      "step": 6281
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9213706349991244e-05,
+      "loss": 1.067,
+      "step": 6282
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9212755776369922e-05,
+      "loss": 1.1642,
+      "step": 6283
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.921180506688304e-05,
+      "loss": 1.2249,
+      "step": 6284
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.92108542215485e-05,
+      "loss": 1.077,
+      "step": 6285
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9209903240384222e-05,
+      "loss": 1.151,
+      "step": 6286
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9208952123408113e-05,
+      "loss": 1.0903,
+      "step": 6287
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9208000870638095e-05,
+      "loss": 1.169,
+      "step": 6288
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.920704948209208e-05,
+      "loss": 1.0481,
+      "step": 6289
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9206097957787995e-05,
+      "loss": 1.0928,
+      "step": 6290
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9205146297743763e-05,
+      "loss": 1.1412,
+      "step": 6291
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.92041945019773e-05,
+      "loss": 1.0551,
+      "step": 6292
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.920324257050655e-05,
+      "loss": 1.2174,
+      "step": 6293
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9202290503349437e-05,
+      "loss": 1.0733,
+      "step": 6294
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9201338300523897e-05,
+      "loss": 1.1755,
+      "step": 6295
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9200385962047864e-05,
+      "loss": 1.1986,
+      "step": 6296
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9199433487939277e-05,
+      "loss": 1.102,
+      "step": 6297
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.919848087821608e-05,
+      "loss": 1.1032,
+      "step": 6298
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9197528132896217e-05,
+      "loss": 1.1147,
+      "step": 6299
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.919657525199763e-05,
+      "loss": 1.1139,
+      "step": 6300
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9195622235538277e-05,
+      "loss": 1.1102,
+      "step": 6301
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.91946690835361e-05,
+      "loss": 1.1135,
+      "step": 6302
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9193715796009058e-05,
+      "loss": 1.1627,
+      "step": 6303
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.919276237297511e-05,
+      "loss": 1.2193,
+      "step": 6304
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9191808814452218e-05,
+      "loss": 1.1918,
+      "step": 6305
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9190855120458334e-05,
+      "loss": 1.1626,
+      "step": 6306
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.918990129101143e-05,
+      "loss": 1.1236,
+      "step": 6307
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.918894732612947e-05,
+      "loss": 1.1753,
+      "step": 6308
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9187993225830426e-05,
+      "loss": 1.2306,
+      "step": 6309
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9187038990132267e-05,
+      "loss": 1.1262,
+      "step": 6310
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9186084619052974e-05,
+      "loss": 1.2233,
+      "step": 6311
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.918513011261052e-05,
+      "loss": 1.2236,
+      "step": 6312
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9184175470822884e-05,
+      "loss": 1.2018,
+      "step": 6313
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.918322069370805e-05,
+      "loss": 1.1325,
+      "step": 6314
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9182265781284002e-05,
+      "loss": 1.2398,
+      "step": 6315
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9181310733568726e-05,
+      "loss": 1.128,
+      "step": 6316
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.918035555058022e-05,
+      "loss": 1.178,
+      "step": 6317
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9179400232336463e-05,
+      "loss": 1.1028,
+      "step": 6318
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9178444778855465e-05,
+      "loss": 1.1564,
+      "step": 6319
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9177489190155214e-05,
+      "loss": 1.2141,
+      "step": 6320
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9176533466253713e-05,
+      "loss": 1.0947,
+      "step": 6321
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9175577607168965e-05,
+      "loss": 1.2276,
+      "step": 6322
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.917462161291898e-05,
+      "loss": 1.1311,
+      "step": 6323
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9173665483521758e-05,
+      "loss": 1.1765,
+      "step": 6324
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9172709218995308e-05,
+      "loss": 1.1942,
+      "step": 6325
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9171752819357654e-05,
+      "loss": 1.1666,
+      "step": 6326
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9170796284626803e-05,
+      "loss": 1.1061,
+      "step": 6327
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9169839614820776e-05,
+      "loss": 1.2083,
+      "step": 6328
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.916888280995759e-05,
+      "loss": 1.1686,
+      "step": 6329
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9167925870055274e-05,
+      "loss": 1.1171,
+      "step": 6330
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.916696879513185e-05,
+      "loss": 1.098,
+      "step": 6331
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.916601158520535e-05,
+      "loss": 1.1551,
+      "step": 6332
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.91650542402938e-05,
+      "loss": 1.1729,
+      "step": 6333
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9164096760415233e-05,
+      "loss": 1.0563,
+      "step": 6334
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.916313914558769e-05,
+      "loss": 1.0945,
+      "step": 6335
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9162181395829205e-05,
+      "loss": 1.1214,
+      "step": 6336
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9161223511157822e-05,
+      "loss": 1.0891,
+      "step": 6337
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.916026549159158e-05,
+      "loss": 1.101,
+      "step": 6338
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9159307337148536e-05,
+      "loss": 1.1438,
+      "step": 6339
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9158349047846723e-05,
+      "loss": 1.1822,
+      "step": 6340
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9157390623704206e-05,
+      "loss": 1.0908,
+      "step": 6341
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9156432064739026e-05,
+      "loss": 1.0768,
+      "step": 6342
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9155473370969253e-05,
+      "loss": 1.153,
+      "step": 6343
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9154514542412936e-05,
+      "loss": 1.1388,
+      "step": 6344
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9153555579088138e-05,
+      "loss": 1.1294,
+      "step": 6345
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9152596481012924e-05,
+      "loss": 1.1382,
+      "step": 6346
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.915163724820536e-05,
+      "loss": 1.1422,
+      "step": 6347
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9150677880683516e-05,
+      "loss": 1.0724,
+      "step": 6348
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.914971837846546e-05,
+      "loss": 1.1443,
+      "step": 6349
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9148758741569272e-05,
+      "loss": 1.0443,
+      "step": 6350
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9147798970013025e-05,
+      "loss": 1.1435,
+      "step": 6351
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9146839063814796e-05,
+      "loss": 1.1385,
+      "step": 6352
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.914587902299267e-05,
+      "loss": 1.0945,
+      "step": 6353
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.914491884756473e-05,
+      "loss": 1.166,
+      "step": 6354
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9143958537549064e-05,
+      "loss": 1.1641,
+      "step": 6355
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9142998092963757e-05,
+      "loss": 1.0934,
+      "step": 6356
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9142037513826905e-05,
+      "loss": 1.1894,
+      "step": 6357
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9141076800156602e-05,
+      "loss": 1.0967,
+      "step": 6358
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9140115951970942e-05,
+      "loss": 1.2038,
+      "step": 6359
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9139154969288027e-05,
+      "loss": 1.1109,
+      "step": 6360
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.913819385212596e-05,
+      "loss": 1.0961,
+      "step": 6361
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9137232600502837e-05,
+      "loss": 1.0881,
+      "step": 6362
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9136271214436778e-05,
+      "loss": 1.1358,
+      "step": 6363
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9135309693945878e-05,
+      "loss": 1.1554,
+      "step": 6364
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9134348039048265e-05,
+      "loss": 1.1647,
+      "step": 6365
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.913338624976204e-05,
+      "loss": 1.1468,
+      "step": 6366
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9132424326105326e-05,
+      "loss": 1.1498,
+      "step": 6367
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.913146226809625e-05,
+      "loss": 1.1633,
+      "step": 6368
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9130500075752914e-05,
+      "loss": 1.0415,
+      "step": 6369
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.912953774909346e-05,
+      "loss": 1.1167,
+      "step": 6370
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.912857528813601e-05,
+      "loss": 1.2134,
+      "step": 6371
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9127612692898696e-05,
+      "loss": 1.0707,
+      "step": 6372
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9126649963399646e-05,
+      "loss": 1.0896,
+      "step": 6373
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9125687099656994e-05,
+      "loss": 1.1598,
+      "step": 6374
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9124724101688886e-05,
+      "loss": 1.1177,
+      "step": 6375
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9123760969513453e-05,
+      "loss": 1.1465,
+      "step": 6376
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9122797703148837e-05,
+      "loss": 1.1481,
+      "step": 6377
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.912183430261319e-05,
+      "loss": 1.0459,
+      "step": 6378
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9120870767924654e-05,
+      "loss": 1.0373,
+      "step": 6379
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9119907099101383e-05,
+      "loss": 1.1513,
+      "step": 6380
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9118943296161524e-05,
+      "loss": 1.1353,
+      "step": 6381
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9117979359123238e-05,
+      "loss": 1.1607,
+      "step": 6382
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9117015288004678e-05,
+      "loss": 1.045,
+      "step": 6383
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9116051082824004e-05,
+      "loss": 1.2218,
+      "step": 6384
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9115086743599383e-05,
+      "loss": 1.1221,
+      "step": 6385
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9114122270348974e-05,
+      "loss": 1.1071,
+      "step": 6386
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.911315766309095e-05,
+      "loss": 1.1583,
+      "step": 6387
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9112192921843478e-05,
+      "loss": 1.1773,
+      "step": 6388
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9111228046624733e-05,
+      "loss": 1.143,
+      "step": 6389
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.911026303745289e-05,
+      "loss": 1.1702,
+      "step": 6390
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9109297894346126e-05,
+      "loss": 1.159,
+      "step": 6391
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.910833261732262e-05,
+      "loss": 1.1386,
+      "step": 6392
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9107367206400556e-05,
+      "loss": 1.171,
+      "step": 6393
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9106401661598116e-05,
+      "loss": 1.0986,
+      "step": 6394
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9105435982933496e-05,
+      "loss": 1.2183,
+      "step": 6395
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.910447017042488e-05,
+      "loss": 1.1071,
+      "step": 6396
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.910350422409046e-05,
+      "loss": 1.1976,
+      "step": 6397
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.910253814394843e-05,
+      "loss": 1.2274,
+      "step": 6398
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9101571930017e-05,
+      "loss": 1.13,
+      "step": 6399
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9100605582314357e-05,
+      "loss": 1.1186,
+      "step": 6400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.909963910085871e-05,
+      "loss": 1.1327,
+      "step": 6401
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9098672485668262e-05,
+      "loss": 1.193,
+      "step": 6402
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9097705736761223e-05,
+      "loss": 1.1936,
+      "step": 6403
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.90967388541558e-05,
+      "loss": 1.0848,
+      "step": 6404
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.909577183787021e-05,
+      "loss": 1.2362,
+      "step": 6405
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.909480468792267e-05,
+      "loss": 1.0406,
+      "step": 6406
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9093837404331392e-05,
+      "loss": 1.0943,
+      "step": 6407
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9092869987114602e-05,
+      "loss": 1.1061,
+      "step": 6408
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9091902436290518e-05,
+      "loss": 1.117,
+      "step": 6409
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.909093475187737e-05,
+      "loss": 1.1259,
+      "step": 6410
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9089966933893386e-05,
+      "loss": 1.082,
+      "step": 6411
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9088998982356795e-05,
+      "loss": 1.112,
+      "step": 6412
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9088030897285833e-05,
+      "loss": 1.0977,
+      "step": 6413
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9087062678698727e-05,
+      "loss": 1.1902,
+      "step": 6414
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9086094326613726e-05,
+      "loss": 1.1137,
+      "step": 6415
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9085125841049062e-05,
+      "loss": 1.1176,
+      "step": 6416
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9084157222022987e-05,
+      "loss": 1.0608,
+      "step": 6417
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.908318846955374e-05,
+      "loss": 1.089,
+      "step": 6418
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9082219583659572e-05,
+      "loss": 1.0659,
+      "step": 6419
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.908125056435873e-05,
+      "loss": 1.0936,
+      "step": 6420
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.908028141166947e-05,
+      "loss": 1.1052,
+      "step": 6421
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.907931212561005e-05,
+      "loss": 1.0503,
+      "step": 6422
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9078342706198725e-05,
+      "loss": 1.0708,
+      "step": 6423
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.907737315345376e-05,
+      "loss": 1.0888,
+      "step": 6424
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9076403467393415e-05,
+      "loss": 1.1252,
+      "step": 6425
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9075433648035954e-05,
+      "loss": 1.1172,
+      "step": 6426
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9074463695399647e-05,
+      "loss": 1.0746,
+      "step": 6427
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9073493609502768e-05,
+      "loss": 1.1063,
+      "step": 6428
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9072523390363584e-05,
+      "loss": 1.1874,
+      "step": 6429
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9071553038000382e-05,
+      "loss": 1.1436,
+      "step": 6430
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9070582552431423e-05,
+      "loss": 1.1584,
+      "step": 6431
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9069611933675007e-05,
+      "loss": 1.0745,
+      "step": 6432
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9068641181749402e-05,
+      "loss": 1.1755,
+      "step": 6433
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9067670296672905e-05,
+      "loss": 1.146,
+      "step": 6434
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9066699278463795e-05,
+      "loss": 1.0896,
+      "step": 6435
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9065728127140366e-05,
+      "loss": 1.1058,
+      "step": 6436
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.906475684272092e-05,
+      "loss": 1.123,
+      "step": 6437
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.906378542522374e-05,
+      "loss": 1.1683,
+      "step": 6438
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9062813874667135e-05,
+      "loss": 1.1275,
+      "step": 6439
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.90618421910694e-05,
+      "loss": 1.029,
+      "step": 6440
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9060870374448835e-05,
+      "loss": 1.1167,
+      "step": 6441
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9059898424823755e-05,
+      "loss": 1.1303,
+      "step": 6442
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9058926342212464e-05,
+      "loss": 1.1768,
+      "step": 6443
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.905795412663327e-05,
+      "loss": 1.0901,
+      "step": 6444
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9056981778104495e-05,
+      "loss": 1.1379,
+      "step": 6445
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9056009296644446e-05,
+      "loss": 1.0763,
+      "step": 6446
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9055036682271444e-05,
+      "loss": 1.1597,
+      "step": 6447
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9054063935003814e-05,
+      "loss": 1.1582,
+      "step": 6448
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9053091054859875e-05,
+      "loss": 1.1171,
+      "step": 6449
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9052118041857955e-05,
+      "loss": 1.0977,
+      "step": 6450
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9051144896016383e-05,
+      "loss": 1.1881,
+      "step": 6451
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9050171617353485e-05,
+      "loss": 1.0493,
+      "step": 6452
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9049198205887602e-05,
+      "loss": 1.0967,
+      "step": 6453
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9048224661637067e-05,
+      "loss": 1.101,
+      "step": 6454
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9047250984620215e-05,
+      "loss": 1.026,
+      "step": 6455
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.904627717485539e-05,
+      "loss": 1.162,
+      "step": 6456
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.904530323236094e-05,
+      "loss": 1.1156,
+      "step": 6457
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9044329157155202e-05,
+      "loss": 1.0386,
+      "step": 6458
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.904335494925653e-05,
+      "loss": 1.1443,
+      "step": 6459
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9042380608683278e-05,
+      "loss": 1.1002,
+      "step": 6460
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.904140613545379e-05,
+      "loss": 1.1936,
+      "step": 6461
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9040431529586428e-05,
+      "loss": 1.1704,
+      "step": 6462
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9039456791099553e-05,
+      "loss": 1.1376,
+      "step": 6463
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9038481920011523e-05,
+      "loss": 1.0717,
+      "step": 6464
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9037506916340698e-05,
+      "loss": 1.0482,
+      "step": 6465
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.903653178010545e-05,
+      "loss": 1.1611,
+      "step": 6466
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9035556511324145e-05,
+      "loss": 1.056,
+      "step": 6467
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9034581110015157e-05,
+      "loss": 1.0922,
+      "step": 6468
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.903360557619685e-05,
+      "loss": 1.1167,
+      "step": 6469
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.903262990988761e-05,
+      "loss": 1.0258,
+      "step": 6470
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9031654111105813e-05,
+      "loss": 1.1132,
+      "step": 6471
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9030678179869835e-05,
+      "loss": 1.1433,
+      "step": 6472
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9029702116198067e-05,
+      "loss": 1.1594,
+      "step": 6473
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.902872592010889e-05,
+      "loss": 0.9998,
+      "step": 6474
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9027749591620694e-05,
+      "loss": 1.1559,
+      "step": 6475
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.902677313075187e-05,
+      "loss": 1.1381,
+      "step": 6476
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.902579653752081e-05,
+      "loss": 1.039,
+      "step": 6477
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.902481981194591e-05,
+      "loss": 1.2142,
+      "step": 6478
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9023842954045573e-05,
+      "loss": 1.1406,
+      "step": 6479
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9022865963838196e-05,
+      "loss": 1.1071,
+      "step": 6480
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9021888841342182e-05,
+      "loss": 1.1306,
+      "step": 6481
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9020911586575936e-05,
+      "loss": 1.1114,
+      "step": 6482
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.901993419955787e-05,
+      "loss": 1.1059,
+      "step": 6483
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.901895668030639e-05,
+      "loss": 1.0581,
+      "step": 6484
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.901797902883992e-05,
+      "loss": 1.0773,
+      "step": 6485
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.901700124517686e-05,
+      "loss": 1.1254,
+      "step": 6486
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9016023329335638e-05,
+      "loss": 1.1187,
+      "step": 6487
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9015045281334678e-05,
+      "loss": 1.0819,
+      "step": 6488
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9014067101192395e-05,
+      "loss": 1.1957,
+      "step": 6489
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.901308878892722e-05,
+      "loss": 1.1425,
+      "step": 6490
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9012110344557578e-05,
+      "loss": 1.1706,
+      "step": 6491
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9011131768101907e-05,
+      "loss": 1.0565,
+      "step": 6492
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9010153059578632e-05,
+      "loss": 1.1631,
+      "step": 6493
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.900917421900619e-05,
+      "loss": 1.1414,
+      "step": 6494
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.900819524640302e-05,
+      "loss": 1.0676,
+      "step": 6495
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.900721614178757e-05,
+      "loss": 1.132,
+      "step": 6496
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9006236905178272e-05,
+      "loss": 1.1204,
+      "step": 6497
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9005257536593578e-05,
+      "loss": 1.1068,
+      "step": 6498
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9004278036051938e-05,
+      "loss": 1.066,
+      "step": 6499
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9003298403571795e-05,
+      "loss": 1.1163,
+      "step": 6500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9002318639171613e-05,
+      "loss": 1.0813,
+      "step": 6501
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9001338742869837e-05,
+      "loss": 1.1688,
+      "step": 6502
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.900035871468493e-05,
+      "loss": 1.063,
+      "step": 6503
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.899937855463535e-05,
+      "loss": 1.0902,
+      "step": 6504
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.899839826273957e-05,
+      "loss": 1.0606,
+      "step": 6505
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.899741783901604e-05,
+      "loss": 1.1043,
+      "step": 6506
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8996437283483243e-05,
+      "loss": 1.1359,
+      "step": 6507
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8995456596159638e-05,
+      "loss": 1.0282,
+      "step": 6508
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8994475777063703e-05,
+      "loss": 1.2258,
+      "step": 6509
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8993494826213918e-05,
+      "loss": 1.0952,
+      "step": 6510
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.899251374362875e-05,
+      "loss": 1.1227,
+      "step": 6511
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.899153252932669e-05,
+      "loss": 1.1456,
+      "step": 6512
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8990551183326217e-05,
+      "loss": 1.1062,
+      "step": 6513
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.898956970564582e-05,
+      "loss": 1.0944,
+      "step": 6514
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8988588096303977e-05,
+      "loss": 1.0375,
+      "step": 6515
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8987606355319185e-05,
+      "loss": 1.1376,
+      "step": 6516
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.898662448270994e-05,
+      "loss": 1.1336,
+      "step": 6517
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.898564247849473e-05,
+      "loss": 1.0542,
+      "step": 6518
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.898466034269206e-05,
+      "loss": 1.088,
+      "step": 6519
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8983678075320425e-05,
+      "loss": 1.0253,
+      "step": 6520
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8982695676398333e-05,
+      "loss": 1.1209,
+      "step": 6521
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8981713145944285e-05,
+      "loss": 1.1318,
+      "step": 6522
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.898073048397679e-05,
+      "loss": 1.1876,
+      "step": 6523
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8979747690514356e-05,
+      "loss": 0.97,
+      "step": 6524
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.89787647655755e-05,
+      "loss": 1.0717,
+      "step": 6525
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8977781709178736e-05,
+      "loss": 1.1475,
+      "step": 6526
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.897679852134258e-05,
+      "loss": 1.081,
+      "step": 6527
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8975815202085557e-05,
+      "loss": 1.1059,
+      "step": 6528
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8974831751426182e-05,
+      "loss": 1.0844,
+      "step": 6529
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.897384816938298e-05,
+      "loss": 1.0951,
+      "step": 6530
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.897286445597449e-05,
+      "loss": 1.0659,
+      "step": 6531
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.897188061121923e-05,
+      "loss": 0.9802,
+      "step": 6532
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8970896635135737e-05,
+      "loss": 1.0355,
+      "step": 6533
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.896991252774255e-05,
+      "loss": 1.1185,
+      "step": 6534
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.89689282890582e-05,
+      "loss": 1.1357,
+      "step": 6535
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.896794391910123e-05,
+      "loss": 1.1276,
+      "step": 6536
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8966959417890177e-05,
+      "loss": 1.1201,
+      "step": 6537
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8965974785443596e-05,
+      "loss": 1.0724,
+      "step": 6538
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8964990021780027e-05,
+      "loss": 1.1268,
+      "step": 6539
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.896400512691802e-05,
+      "loss": 1.0818,
+      "step": 6540
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8963020100876138e-05,
+      "loss": 1.0759,
+      "step": 6541
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.896203494367292e-05,
+      "loss": 1.1248,
+      "step": 6542
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8961049655326932e-05,
+      "loss": 1.1554,
+      "step": 6543
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.896006423585673e-05,
+      "loss": 1.1386,
+      "step": 6544
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.895907868528088e-05,
+      "loss": 1.0604,
+      "step": 6545
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8958093003617944e-05,
+      "loss": 1.0846,
+      "step": 6546
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8957107190886492e-05,
+      "loss": 1.131,
+      "step": 6547
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8956121247105088e-05,
+      "loss": 1.1545,
+      "step": 6548
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8955135172292312e-05,
+      "loss": 1.1276,
+      "step": 6549
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8954148966466734e-05,
+      "loss": 1.0118,
+      "step": 6550
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.895316262964693e-05,
+      "loss": 1.0675,
+      "step": 6551
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.895217616185148e-05,
+      "loss": 1.1359,
+      "step": 6552
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8951189563098968e-05,
+      "loss": 1.0238,
+      "step": 6553
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8950202833407978e-05,
+      "loss": 1.0962,
+      "step": 6554
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8949215972797095e-05,
+      "loss": 1.1059,
+      "step": 6555
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8948228981284907e-05,
+      "loss": 1.1461,
+      "step": 6556
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.894724185889001e-05,
+      "loss": 1.0152,
+      "step": 6557
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8946254605630996e-05,
+      "loss": 1.0575,
+      "step": 6558
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.894526722152646e-05,
+      "loss": 1.0394,
+      "step": 6559
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8944279706595007e-05,
+      "loss": 1.0363,
+      "step": 6560
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8943292060855235e-05,
+      "loss": 1.1207,
+      "step": 6561
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8942304284325746e-05,
+      "loss": 1.1274,
+      "step": 6562
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8941316377025147e-05,
+      "loss": 1.0767,
+      "step": 6563
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8940328338972054e-05,
+      "loss": 1.1388,
+      "step": 6564
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.893934017018507e-05,
+      "loss": 1.0906,
+      "step": 6565
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8938351870682813e-05,
+      "loss": 1.0983,
+      "step": 6566
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8937363440483893e-05,
+      "loss": 1.0882,
+      "step": 6567
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.893637487960694e-05,
+      "loss": 1.148,
+      "step": 6568
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.893538618807057e-05,
+      "loss": 1.1188,
+      "step": 6569
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.893439736589341e-05,
+      "loss": 1.1244,
+      "step": 6570
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.893340841309408e-05,
+      "loss": 1.0937,
+      "step": 6571
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.893241932969121e-05,
+      "loss": 1.0783,
+      "step": 6572
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8931430115703435e-05,
+      "loss": 1.0832,
+      "step": 6573
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8930440771149387e-05,
+      "loss": 1.1361,
+      "step": 6574
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8929451296047703e-05,
+      "loss": 1.0715,
+      "step": 6575
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.892846169041702e-05,
+      "loss": 1.0479,
+      "step": 6576
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8927471954275977e-05,
+      "loss": 1.1178,
+      "step": 6577
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8926482087643224e-05,
+      "loss": 1.1606,
+      "step": 6578
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8925492090537395e-05,
+      "loss": 1.1293,
+      "step": 6579
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8924501962977152e-05,
+      "loss": 1.1236,
+      "step": 6580
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.892351170498114e-05,
+      "loss": 1.0781,
+      "step": 6581
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8922521316568015e-05,
+      "loss": 1.239,
+      "step": 6582
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8921530797756427e-05,
+      "loss": 1.1499,
+      "step": 6583
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8920540148565038e-05,
+      "loss": 1.0383,
+      "step": 6584
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.891954936901251e-05,
+      "loss": 1.1544,
+      "step": 6585
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8918558459117503e-05,
+      "loss": 1.1453,
+      "step": 6586
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8917567418898683e-05,
+      "loss": 1.152,
+      "step": 6587
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.891657624837472e-05,
+      "loss": 1.0627,
+      "step": 6588
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8915584947564288e-05,
+      "loss": 1.1431,
+      "step": 6589
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.891459351648605e-05,
+      "loss": 1.0391,
+      "step": 6590
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.891360195515869e-05,
+      "loss": 1.0595,
+      "step": 6591
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.891261026360088e-05,
+      "loss": 1.0643,
+      "step": 6592
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8911618441831308e-05,
+      "loss": 1.0897,
+      "step": 6593
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.891062648986865e-05,
+      "loss": 1.1473,
+      "step": 6594
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.890963440773159e-05,
+      "loss": 1.096,
+      "step": 6595
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8908642195438824e-05,
+      "loss": 1.1428,
+      "step": 6596
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8907649853009035e-05,
+      "loss": 1.07,
+      "step": 6597
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.890665738046092e-05,
+      "loss": 1.1513,
+      "step": 6598
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.890566477781317e-05,
+      "loss": 1.106,
+      "step": 6599
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8904672045084486e-05,
+      "loss": 1.1889,
+      "step": 6600
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8903679182293565e-05,
+      "loss": 1.1361,
+      "step": 6601
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.890268618945911e-05,
+      "loss": 1.1196,
+      "step": 6602
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8901693066599825e-05,
+      "loss": 1.099,
+      "step": 6603
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8900699813734423e-05,
+      "loss": 1.129,
+      "step": 6604
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.889970643088161e-05,
+      "loss": 1.1247,
+      "step": 6605
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8898712918060094e-05,
+      "loss": 1.1541,
+      "step": 6606
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8897719275288597e-05,
+      "loss": 1.0711,
+      "step": 6607
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.889672550258583e-05,
+      "loss": 1.0993,
+      "step": 6608
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8895731599970517e-05,
+      "loss": 0.9301,
+      "step": 6609
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8894737567461376e-05,
+      "loss": 1.0588,
+      "step": 6610
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8893743405077135e-05,
+      "loss": 1.1175,
+      "step": 6611
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8892749112836522e-05,
+      "loss": 1.0858,
+      "step": 6612
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.889175469075826e-05,
+      "loss": 1.1238,
+      "step": 6613
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8890760138861087e-05,
+      "loss": 1.1571,
+      "step": 6614
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.888976545716373e-05,
+      "loss": 1.0542,
+      "step": 6615
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8888770645684938e-05,
+      "loss": 1.137,
+      "step": 6616
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8887775704443435e-05,
+      "loss": 1.1237,
+      "step": 6617
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8886780633457976e-05,
+      "loss": 1.1411,
+      "step": 6618
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8885785432747297e-05,
+      "loss": 1.1299,
+      "step": 6619
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8884790102330144e-05,
+      "loss": 1.0134,
+      "step": 6620
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.888379464222527e-05,
+      "loss": 1.0672,
+      "step": 6621
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8882799052451426e-05,
+      "loss": 1.0476,
+      "step": 6622
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8881803333027362e-05,
+      "loss": 1.0911,
+      "step": 6623
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.888080748397184e-05,
+      "loss": 1.1001,
+      "step": 6624
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8879811505303615e-05,
+      "loss": 1.0066,
+      "step": 6625
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8878815397041446e-05,
+      "loss": 1.0623,
+      "step": 6626
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.88778191592041e-05,
+      "loss": 1.1308,
+      "step": 6627
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.887682279181034e-05,
+      "loss": 1.0651,
+      "step": 6628
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8875826294878943e-05,
+      "loss": 1.0829,
+      "step": 6629
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8874829668428668e-05,
+      "loss": 1.0969,
+      "step": 6630
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8873832912478294e-05,
+      "loss": 1.0687,
+      "step": 6631
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8872836027046597e-05,
+      "loss": 1.0174,
+      "step": 6632
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8871839012152356e-05,
+      "loss": 1.0573,
+      "step": 6633
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8870841867814346e-05,
+      "loss": 1.1182,
+      "step": 6634
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.886984459405136e-05,
+      "loss": 1.0547,
+      "step": 6635
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.886884719088217e-05,
+      "loss": 1.0117,
+      "step": 6636
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.886784965832558e-05,
+      "loss": 1.0973,
+      "step": 6637
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8866851996400367e-05,
+      "loss": 1.1037,
+      "step": 6638
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.886585420512533e-05,
+      "loss": 1.0808,
+      "step": 6639
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8864856284519264e-05,
+      "loss": 1.1266,
+      "step": 6640
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8863858234600963e-05,
+      "loss": 1.0423,
+      "step": 6641
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8862860055389232e-05,
+      "loss": 1.0153,
+      "step": 6642
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8861861746902868e-05,
+      "loss": 1.0157,
+      "step": 6643
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8860863309160683e-05,
+      "loss": 1.1623,
+      "step": 6644
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.885986474218148e-05,
+      "loss": 1.0957,
+      "step": 6645
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8858866045984067e-05,
+      "loss": 1.0657,
+      "step": 6646
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8857867220587262e-05,
+      "loss": 1.0909,
+      "step": 6647
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8856868266009875e-05,
+      "loss": 1.0275,
+      "step": 6648
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8855869182270728e-05,
+      "loss": 1.1053,
+      "step": 6649
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8854869969388632e-05,
+      "loss": 1.0638,
+      "step": 6650
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8853870627382417e-05,
+      "loss": 1.1375,
+      "step": 6651
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8852871156270905e-05,
+      "loss": 1.0727,
+      "step": 6652
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.885187155607292e-05,
+      "loss": 1.1232,
+      "step": 6653
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8850871826807298e-05,
+      "loss": 1.0444,
+      "step": 6654
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8849871968492865e-05,
+      "loss": 1.089,
+      "step": 6655
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8848871981148454e-05,
+      "loss": 1.0951,
+      "step": 6656
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.884787186479291e-05,
+      "loss": 1.1225,
+      "step": 6657
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8846871619445062e-05,
+      "loss": 1.1481,
+      "step": 6658
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.884587124512376e-05,
+      "loss": 1.0338,
+      "step": 6659
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.884487074184784e-05,
+      "loss": 1.0868,
+      "step": 6660
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8843870109636152e-05,
+      "loss": 1.1653,
+      "step": 6661
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.884286934850755e-05,
+      "loss": 1.1156,
+      "step": 6662
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8841868458480875e-05,
+      "loss": 1.0392,
+      "step": 6663
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.884086743957499e-05,
+      "loss": 1.0701,
+      "step": 6664
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.883986629180874e-05,
+      "loss": 1.0982,
+      "step": 6665
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8838865015200997e-05,
+      "loss": 1.1201,
+      "step": 6666
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8837863609770613e-05,
+      "loss": 1.1171,
+      "step": 6667
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8836862075536455e-05,
+      "loss": 1.113,
+      "step": 6668
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8835860412517386e-05,
+      "loss": 1.1536,
+      "step": 6669
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8834858620732273e-05,
+      "loss": 1.1339,
+      "step": 6670
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.883385670019999e-05,
+      "loss": 1.0644,
+      "step": 6671
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.883285465093941e-05,
+      "loss": 1.0299,
+      "step": 6672
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.883185247296941e-05,
+      "loss": 1.0159,
+      "step": 6673
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.883085016630886e-05,
+      "loss": 1.1093,
+      "step": 6674
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.882984773097665e-05,
+      "loss": 1.1465,
+      "step": 6675
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8828845166991656e-05,
+      "loss": 1.0793,
+      "step": 6676
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.882784247437277e-05,
+      "loss": 1.2222,
+      "step": 6677
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8826839653138873e-05,
+      "loss": 1.0824,
+      "step": 6678
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8825836703308855e-05,
+      "loss": 1.1838,
+      "step": 6679
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.882483362490161e-05,
+      "loss": 1.0699,
+      "step": 6680
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8823830417936034e-05,
+      "loss": 1.1438,
+      "step": 6681
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8822827082431023e-05,
+      "loss": 1.1186,
+      "step": 6682
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8821823618405476e-05,
+      "loss": 1.059,
+      "step": 6683
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.88208200258783e-05,
+      "loss": 1.0721,
+      "step": 6684
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8819816304868394e-05,
+      "loss": 1.0668,
+      "step": 6685
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8818812455394667e-05,
+      "loss": 1.0348,
+      "step": 6686
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.881780847747603e-05,
+      "loss": 1.1171,
+      "step": 6687
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.881680437113139e-05,
+      "loss": 1.0403,
+      "step": 6688
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8815800136379664e-05,
+      "loss": 1.1229,
+      "step": 6689
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8814795773239767e-05,
+      "loss": 0.9738,
+      "step": 6690
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8813791281730624e-05,
+      "loss": 1.1638,
+      "step": 6691
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.881278666187115e-05,
+      "loss": 1.1415,
+      "step": 6692
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8811781913680274e-05,
+      "loss": 1.1276,
+      "step": 6693
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8810777037176916e-05,
+      "loss": 1.1783,
+      "step": 6694
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8809772032380007e-05,
+      "loss": 1.1207,
+      "step": 6695
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8808766899308482e-05,
+      "loss": 1.0559,
+      "step": 6696
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8807761637981272e-05,
+      "loss": 1.0959,
+      "step": 6697
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.880675624841731e-05,
+      "loss": 1.013,
+      "step": 6698
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8805750730635542e-05,
+      "loss": 1.117,
+      "step": 6699
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.88047450846549e-05,
+      "loss": 1.0822,
+      "step": 6700
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.880373931049433e-05,
+      "loss": 1.0981,
+      "step": 6701
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8802733408172782e-05,
+      "loss": 1.0188,
+      "step": 6702
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8801727377709197e-05,
+      "loss": 1.073,
+      "step": 6703
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.880072121912253e-05,
+      "loss": 1.0213,
+      "step": 6704
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8799714932431735e-05,
+      "loss": 1.101,
+      "step": 6705
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.879870851765576e-05,
+      "loss": 1.0946,
+      "step": 6706
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8797701974813574e-05,
+      "loss": 1.1119,
+      "step": 6707
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.879669530392413e-05,
+      "loss": 1.1571,
+      "step": 6708
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.879568850500639e-05,
+      "loss": 1.2096,
+      "step": 6709
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8794681578079322e-05,
+      "loss": 1.1596,
+      "step": 6710
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.879367452316189e-05,
+      "loss": 1.0969,
+      "step": 6711
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8792667340273064e-05,
+      "loss": 1.0935,
+      "step": 6712
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.879166002943182e-05,
+      "loss": 1.09,
+      "step": 6713
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8790652590657126e-05,
+      "loss": 1.1571,
+      "step": 6714
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.878964502396797e-05,
+      "loss": 1.1896,
+      "step": 6715
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.878863732938332e-05,
+      "loss": 1.0587,
+      "step": 6716
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8787629506922164e-05,
+      "loss": 1.1227,
+      "step": 6717
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.878662155660348e-05,
+      "loss": 1.1428,
+      "step": 6718
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8785613478446263e-05,
+      "loss": 1.1062,
+      "step": 6719
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8784605272469503e-05,
+      "loss": 1.1364,
+      "step": 6720
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8783596938692178e-05,
+      "loss": 1.1384,
+      "step": 6721
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8782588477133296e-05,
+      "loss": 1.1008,
+      "step": 6722
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8781579887811843e-05,
+      "loss": 1.0705,
+      "step": 6723
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8780571170746824e-05,
+      "loss": 1.1875,
+      "step": 6724
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.877956232595724e-05,
+      "loss": 1.0913,
+      "step": 6725
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8778553353462093e-05,
+      "loss": 1.1033,
+      "step": 6726
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8777544253280386e-05,
+      "loss": 1.1237,
+      "step": 6727
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8776535025431132e-05,
+      "loss": 1.1346,
+      "step": 6728
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.877552566993334e-05,
+      "loss": 1.0353,
+      "step": 6729
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.877451618680602e-05,
+      "loss": 1.0814,
+      "step": 6730
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.877350657606819e-05,
+      "loss": 1.0738,
+      "step": 6731
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8772496837738867e-05,
+      "loss": 1.0284,
+      "step": 6732
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.877148697183708e-05,
+      "loss": 1.14,
+      "step": 6733
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8770476978381835e-05,
+      "loss": 1.1192,
+      "step": 6734
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8769466857392168e-05,
+      "loss": 1.1431,
+      "step": 6735
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8768456608887104e-05,
+      "loss": 1.0729,
+      "step": 6736
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.876744623288568e-05,
+      "loss": 1.1178,
+      "step": 6737
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8766435729406915e-05,
+      "loss": 1.1328,
+      "step": 6738
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8765425098469854e-05,
+      "loss": 1.1964,
+      "step": 6739
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8764414340093528e-05,
+      "loss": 1.1034,
+      "step": 6740
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8763403454296976e-05,
+      "loss": 1.0418,
+      "step": 6741
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8762392441099248e-05,
+      "loss": 1.066,
+      "step": 6742
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8761381300519383e-05,
+      "loss": 1.0292,
+      "step": 6743
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8760370032576425e-05,
+      "loss": 1.1203,
+      "step": 6744
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.875935863728943e-05,
+      "loss": 1.1636,
+      "step": 6745
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.875834711467744e-05,
+      "loss": 1.16,
+      "step": 6746
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8757335464759517e-05,
+      "loss": 1.1069,
+      "step": 6747
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8756323687554714e-05,
+      "loss": 1.1285,
+      "step": 6748
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8755311783082092e-05,
+      "loss": 1.1434,
+      "step": 6749
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.875429975136071e-05,
+      "loss": 1.1731,
+      "step": 6750
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8753287592409632e-05,
+      "loss": 1.0241,
+      "step": 6751
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8752275306247925e-05,
+      "loss": 1.1447,
+      "step": 6752
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8751262892894656e-05,
+      "loss": 1.0887,
+      "step": 6753
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8750250352368893e-05,
+      "loss": 1.0332,
+      "step": 6754
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8749237684689713e-05,
+      "loss": 1.1057,
+      "step": 6755
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.874822488987619e-05,
+      "loss": 1.0564,
+      "step": 6756
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.87472119679474e-05,
+      "loss": 1.0987,
+      "step": 6757
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8746198918922427e-05,
+      "loss": 1.1223,
+      "step": 6758
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8745185742820356e-05,
+      "loss": 1.169,
+      "step": 6759
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8744172439660264e-05,
+      "loss": 1.1067,
+      "step": 6760
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8743159009461244e-05,
+      "loss": 1.1042,
+      "step": 6761
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8742145452242384e-05,
+      "loss": 1.0974,
+      "step": 6762
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.874113176802278e-05,
+      "loss": 1.0831,
+      "step": 6763
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8740117956821516e-05,
+      "loss": 1.1329,
+      "step": 6764
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.87391040186577e-05,
+      "loss": 1.1081,
+      "step": 6765
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.873808995355043e-05,
+      "loss": 1.1774,
+      "step": 6766
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8737075761518797e-05,
+      "loss": 1.0488,
+      "step": 6767
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8736061442581923e-05,
+      "loss": 1.0907,
+      "step": 6768
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8735046996758897e-05,
+      "loss": 1.0615,
+      "step": 6769
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8734032424068842e-05,
+      "loss": 1.0552,
+      "step": 6770
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.873301772453086e-05,
+      "loss": 1.0643,
+      "step": 6771
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8732002898164073e-05,
+      "loss": 1.1209,
+      "step": 6772
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.873098794498759e-05,
+      "loss": 1.1081,
+      "step": 6773
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.872997286502053e-05,
+      "loss": 0.9779,
+      "step": 6774
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8728957658282015e-05,
+      "loss": 1.1606,
+      "step": 6775
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.872794232479117e-05,
+      "loss": 1.0662,
+      "step": 6776
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8726926864567122e-05,
+      "loss": 1.131,
+      "step": 6777
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8725911277628997e-05,
+      "loss": 1.1137,
+      "step": 6778
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.872489556399592e-05,
+      "loss": 0.9784,
+      "step": 6779
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.872387972368703e-05,
+      "loss": 1.0951,
+      "step": 6780
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8722863756721466e-05,
+      "loss": 1.1315,
+      "step": 6781
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.872184766311836e-05,
+      "loss": 1.082,
+      "step": 6782
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8720831442896852e-05,
+      "loss": 1.0446,
+      "step": 6783
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8719815096076085e-05,
+      "loss": 0.9823,
+      "step": 6784
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8718798622675203e-05,
+      "loss": 1.0871,
+      "step": 6785
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8717782022713354e-05,
+      "loss": 1.0516,
+      "step": 6786
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.871676529620969e-05,
+      "loss": 1.1684,
+      "step": 6787
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8715748443183355e-05,
+      "loss": 1.1057,
+      "step": 6788
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8714731463653513e-05,
+      "loss": 1.0961,
+      "step": 6789
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8713714357639318e-05,
+      "loss": 1.0311,
+      "step": 6790
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8712697125159926e-05,
+      "loss": 1.0441,
+      "step": 6791
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.87116797662345e-05,
+      "loss": 1.1465,
+      "step": 6792
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8710662280882207e-05,
+      "loss": 1.1186,
+      "step": 6793
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8709644669122206e-05,
+      "loss": 1.1427,
+      "step": 6794
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.870862693097367e-05,
+      "loss": 1.1787,
+      "step": 6795
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.870760906645577e-05,
+      "loss": 1.0315,
+      "step": 6796
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.870659107558768e-05,
+      "loss": 1.0819,
+      "step": 6797
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8705572958388577e-05,
+      "loss": 1.0789,
+      "step": 6798
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8704554714877635e-05,
+      "loss": 1.0401,
+      "step": 6799
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8703536345074036e-05,
+      "loss": 1.0729,
+      "step": 6800
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.870251784899696e-05,
+      "loss": 1.0516,
+      "step": 6801
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.87014992266656e-05,
+      "loss": 1.0993,
+      "step": 6802
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.870048047809914e-05,
+      "loss": 1.0165,
+      "step": 6803
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8699461603316767e-05,
+      "loss": 1.0843,
+      "step": 6804
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8698442602337677e-05,
+      "loss": 1.0917,
+      "step": 6805
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8697423475181063e-05,
+      "loss": 1.1008,
+      "step": 6806
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.869640422186612e-05,
+      "loss": 1.0392,
+      "step": 6807
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8695384842412054e-05,
+      "loss": 0.9516,
+      "step": 6808
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.869436533683806e-05,
+      "loss": 1.0039,
+      "step": 6809
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8693345705163344e-05,
+      "loss": 1.0718,
+      "step": 6810
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8692325947407116e-05,
+      "loss": 1.0726,
+      "step": 6811
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8691306063588584e-05,
+      "loss": 1.1083,
+      "step": 6812
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.869028605372696e-05,
+      "loss": 1.0745,
+      "step": 6813
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.868926591784145e-05,
+      "loss": 1.0702,
+      "step": 6814
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.868824565595128e-05,
+      "loss": 1.0876,
+      "step": 6815
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8687225268075666e-05,
+      "loss": 1.055,
+      "step": 6816
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8686204754233825e-05,
+      "loss": 1.0863,
+      "step": 6817
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.868518411444498e-05,
+      "loss": 1.0823,
+      "step": 6818
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8684163348728366e-05,
+      "loss": 1.2039,
+      "step": 6819
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.86831424571032e-05,
+      "loss": 1.1207,
+      "step": 6820
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8682121439588716e-05,
+      "loss": 1.1866,
+      "step": 6821
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8681100296204152e-05,
+      "loss": 1.1049,
+      "step": 6822
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8680079026968733e-05,
+      "loss": 1.1352,
+      "step": 6823
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8679057631901708e-05,
+      "loss": 1.0938,
+      "step": 6824
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8678036111022304e-05,
+      "loss": 1.0475,
+      "step": 6825
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8677014464349776e-05,
+      "loss": 1.1642,
+      "step": 6826
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8675992691903363e-05,
+      "loss": 1.1611,
+      "step": 6827
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.867497079370231e-05,
+      "loss": 1.0781,
+      "step": 6828
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.867394876976587e-05,
+      "loss": 1.1292,
+      "step": 6829
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8672926620113288e-05,
+      "loss": 1.0844,
+      "step": 6830
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.867190434476383e-05,
+      "loss": 1.0623,
+      "step": 6831
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.867088194373674e-05,
+      "loss": 1.1135,
+      "step": 6832
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.866985941705128e-05,
+      "loss": 1.0597,
+      "step": 6833
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.866883676472672e-05,
+      "loss": 1.0477,
+      "step": 6834
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8667813986782314e-05,
+      "loss": 1.0491,
+      "step": 6835
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8666791083237332e-05,
+      "loss": 1.1261,
+      "step": 6836
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.866576805411104e-05,
+      "loss": 1.084,
+      "step": 6837
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.866474489942271e-05,
+      "loss": 1.1322,
+      "step": 6838
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8663721619191614e-05,
+      "loss": 1.0959,
+      "step": 6839
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.866269821343703e-05,
+      "loss": 1.1434,
+      "step": 6840
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.866167468217823e-05,
+      "loss": 1.1424,
+      "step": 6841
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.86606510254345e-05,
+      "loss": 1.0996,
+      "step": 6842
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8659627243225125e-05,
+      "loss": 1.1201,
+      "step": 6843
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.865860333556938e-05,
+      "loss": 1.1712,
+      "step": 6844
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.865757930248656e-05,
+      "loss": 1.1612,
+      "step": 6845
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8656555143995947e-05,
+      "loss": 1.1237,
+      "step": 6846
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.865553086011684e-05,
+      "loss": 1.143,
+      "step": 6847
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.865450645086854e-05,
+      "loss": 1.155,
+      "step": 6848
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8653481916270323e-05,
+      "loss": 1.104,
+      "step": 6849
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8652457256341505e-05,
+      "loss": 1.1387,
+      "step": 6850
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8651432471101382e-05,
+      "loss": 1.1041,
+      "step": 6851
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.865040756056926e-05,
+      "loss": 1.086,
+      "step": 6852
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8649382524764438e-05,
+      "loss": 1.1356,
+      "step": 6853
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8648357363706232e-05,
+      "loss": 1.1345,
+      "step": 6854
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.864733207741395e-05,
+      "loss": 1.1433,
+      "step": 6855
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8646306665906906e-05,
+      "loss": 1.0942,
+      "step": 6856
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8645281129204414e-05,
+      "loss": 1.26,
+      "step": 6857
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8644255467325794e-05,
+      "loss": 1.095,
+      "step": 6858
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8643229680290365e-05,
+      "loss": 1.2092,
+      "step": 6859
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8642203768117446e-05,
+      "loss": 1.1018,
+      "step": 6860
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.864117773082637e-05,
+      "loss": 1.1776,
+      "step": 6861
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.864015156843646e-05,
+      "loss": 1.1282,
+      "step": 6862
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8639125280967043e-05,
+      "loss": 1.079,
+      "step": 6863
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8638098868437454e-05,
+      "loss": 1.0665,
+      "step": 6864
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8637072330867026e-05,
+      "loss": 1.0687,
+      "step": 6865
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.86360456682751e-05,
+      "loss": 1.0731,
+      "step": 6866
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.863501888068101e-05,
+      "loss": 1.1651,
+      "step": 6867
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8633991968104096e-05,
+      "loss": 1.1341,
+      "step": 6868
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.863296493056371e-05,
+      "loss": 1.1323,
+      "step": 6869
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8631937768079192e-05,
+      "loss": 1.0615,
+      "step": 6870
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.863091048066989e-05,
+      "loss": 1.1611,
+      "step": 6871
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8629883068355155e-05,
+      "loss": 1.1993,
+      "step": 6872
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8628855531154345e-05,
+      "loss": 1.2376,
+      "step": 6873
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8627827869086807e-05,
+      "loss": 1.1343,
+      "step": 6874
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.862680008217191e-05,
+      "loss": 1.0778,
+      "step": 6875
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8625772170429006e-05,
+      "loss": 1.0839,
+      "step": 6876
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8624744133877462e-05,
+      "loss": 1.1391,
+      "step": 6877
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.862371597253664e-05,
+      "loss": 1.1371,
+      "step": 6878
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8622687686425907e-05,
+      "loss": 1.1387,
+      "step": 6879
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8621659275564632e-05,
+      "loss": 1.2186,
+      "step": 6880
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8620630739972192e-05,
+      "loss": 1.1615,
+      "step": 6881
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8619602079667957e-05,
+      "loss": 1.1459,
+      "step": 6882
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8618573294671305e-05,
+      "loss": 1.0878,
+      "step": 6883
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8617544385001617e-05,
+      "loss": 1.119,
+      "step": 6884
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8616515350678272e-05,
+      "loss": 1.1309,
+      "step": 6885
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8615486191720653e-05,
+      "loss": 1.1359,
+      "step": 6886
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.861445690814815e-05,
+      "loss": 1.126,
+      "step": 6887
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8613427499980144e-05,
+      "loss": 1.138,
+      "step": 6888
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8612397967236033e-05,
+      "loss": 1.0644,
+      "step": 6889
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8611368309935212e-05,
+      "loss": 1.1479,
+      "step": 6890
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8610338528097068e-05,
+      "loss": 1.0771,
+      "step": 6891
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8609308621741004e-05,
+      "loss": 1.1688,
+      "step": 6892
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8608278590886416e-05,
+      "loss": 1.1463,
+      "step": 6893
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8607248435552715e-05,
+      "loss": 1.1423,
+      "step": 6894
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8606218155759297e-05,
+      "loss": 1.1543,
+      "step": 6895
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.860518775152558e-05,
+      "loss": 1.1035,
+      "step": 6896
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8604157222870956e-05,
+      "loss": 1.1216,
+      "step": 6897
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8603126569814855e-05,
+      "loss": 1.125,
+      "step": 6898
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.860209579237668e-05,
+      "loss": 1.1461,
+      "step": 6899
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8601064890575854e-05,
+      "loss": 1.1324,
+      "step": 6900
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8600033864431792e-05,
+      "loss": 1.0853,
+      "step": 6901
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8599002713963917e-05,
+      "loss": 1.0535,
+      "step": 6902
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.859797143919165e-05,
+      "loss": 1.063,
+      "step": 6903
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8596940040134418e-05,
+      "loss": 1.1126,
+      "step": 6904
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8595908516811655e-05,
+      "loss": 1.1157,
+      "step": 6905
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8594876869242786e-05,
+      "loss": 1.0778,
+      "step": 6906
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8593845097447242e-05,
+      "loss": 1.0402,
+      "step": 6907
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8592813201444464e-05,
+      "loss": 1.1875,
+      "step": 6908
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8591781181253885e-05,
+      "loss": 1.15,
+      "step": 6909
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.859074903689495e-05,
+      "loss": 1.1462,
+      "step": 6910
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8589716768387097e-05,
+      "loss": 1.1144,
+      "step": 6911
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.858868437574977e-05,
+      "loss": 1.1336,
+      "step": 6912
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8587651859002418e-05,
+      "loss": 1.1319,
+      "step": 6913
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.858661921816449e-05,
+      "loss": 1.1073,
+      "step": 6914
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.858558645325544e-05,
+      "loss": 1.1238,
+      "step": 6915
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.858455356429472e-05,
+      "loss": 1.128,
+      "step": 6916
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8583520551301786e-05,
+      "loss": 1.1647,
+      "step": 6917
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8582487414296098e-05,
+      "loss": 1.1592,
+      "step": 6918
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.858145415329712e-05,
+      "loss": 1.0656,
+      "step": 6919
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8580420768324305e-05,
+      "loss": 1.1311,
+      "step": 6920
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.857938725939713e-05,
+      "loss": 1.1167,
+      "step": 6921
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8578353626535058e-05,
+      "loss": 1.1459,
+      "step": 6922
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8577319869757558e-05,
+      "loss": 1.2138,
+      "step": 6923
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8576285989084102e-05,
+      "loss": 1.1342,
+      "step": 6924
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.857525198453417e-05,
+      "loss": 1.0574,
+      "step": 6925
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8574217856127237e-05,
+      "loss": 1.0198,
+      "step": 6926
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8573183603882782e-05,
+      "loss": 1.1112,
+      "step": 6927
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8572149227820287e-05,
+      "loss": 1.075,
+      "step": 6928
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.857111472795924e-05,
+      "loss": 1.1857,
+      "step": 6929
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8570080104319123e-05,
+      "loss": 1.1661,
+      "step": 6930
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8569045356919424e-05,
+      "loss": 1.0944,
+      "step": 6931
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8568010485779643e-05,
+      "loss": 1.1617,
+      "step": 6932
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.856697549091926e-05,
+      "loss": 1.1352,
+      "step": 6933
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.856594037235779e-05,
+      "loss": 1.1464,
+      "step": 6934
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8564905130114712e-05,
+      "loss": 1.1897,
+      "step": 6935
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.856386976420954e-05,
+      "loss": 1.1218,
+      "step": 6936
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8562834274661767e-05,
+      "loss": 1.1389,
+      "step": 6937
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8561798661490906e-05,
+      "loss": 1.1362,
+      "step": 6938
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.856076292471646e-05,
+      "loss": 1.1936,
+      "step": 6939
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8559727064357945e-05,
+      "loss": 1.1172,
+      "step": 6940
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8558691080434866e-05,
+      "loss": 1.1237,
+      "step": 6941
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8557654972966744e-05,
+      "loss": 1.134,
+      "step": 6942
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.855661874197309e-05,
+      "loss": 1.1374,
+      "step": 6943
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8555582387473426e-05,
+      "loss": 1.1088,
+      "step": 6944
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8554545909487278e-05,
+      "loss": 1.0883,
+      "step": 6945
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8553509308034162e-05,
+      "loss": 1.1123,
+      "step": 6946
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.855247258313361e-05,
+      "loss": 1.0406,
+      "step": 6947
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.855143573480515e-05,
+      "loss": 1.0853,
+      "step": 6948
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.855039876306831e-05,
+      "loss": 1.1178,
+      "step": 6949
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.854936166794263e-05,
+      "loss": 1.1637,
+      "step": 6950
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8548324449447634e-05,
+      "loss": 1.0565,
+      "step": 6951
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.854728710760287e-05,
+      "loss": 1.1291,
+      "step": 6952
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8546249642427873e-05,
+      "loss": 1.064,
+      "step": 6953
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.854521205394219e-05,
+      "loss": 1.1192,
+      "step": 6954
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8544174342165363e-05,
+      "loss": 1.0779,
+      "step": 6955
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.854313650711694e-05,
+      "loss": 1.1198,
+      "step": 6956
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.854209854881647e-05,
+      "loss": 1.0718,
+      "step": 6957
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8541060467283505e-05,
+      "loss": 1.1249,
+      "step": 6958
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8540022262537597e-05,
+      "loss": 1.0448,
+      "step": 6959
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8538983934598305e-05,
+      "loss": 1.162,
+      "step": 6960
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.853794548348519e-05,
+      "loss": 1.1818,
+      "step": 6961
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8536906909217812e-05,
+      "loss": 1.1245,
+      "step": 6962
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.853586821181573e-05,
+      "loss": 1.0641,
+      "step": 6963
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8534829391298512e-05,
+      "loss": 1.1561,
+      "step": 6964
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8533790447685727e-05,
+      "loss": 1.1104,
+      "step": 6965
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.853275138099695e-05,
+      "loss": 1.0643,
+      "step": 6966
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8531712191251746e-05,
+      "loss": 1.1067,
+      "step": 6967
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8530672878469693e-05,
+      "loss": 1.1289,
+      "step": 6968
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8529633442670365e-05,
+      "loss": 1.1573,
+      "step": 6969
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8528593883873353e-05,
+      "loss": 1.1428,
+      "step": 6970
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.852755420209823e-05,
+      "loss": 1.2113,
+      "step": 6971
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8526514397364577e-05,
+      "loss": 1.1362,
+      "step": 6972
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8525474469691986e-05,
+      "loss": 1.1471,
+      "step": 6973
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.852443441910005e-05,
+      "loss": 1.1382,
+      "step": 6974
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.852339424560835e-05,
+      "loss": 1.0984,
+      "step": 6975
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8522353949236487e-05,
+      "loss": 1.1132,
+      "step": 6976
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8521313530004054e-05,
+      "loss": 1.0432,
+      "step": 6977
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8520272987930654e-05,
+      "loss": 1.1503,
+      "step": 6978
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.851923232303588e-05,
+      "loss": 1.0378,
+      "step": 6979
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.851819153533934e-05,
+      "loss": 1.1379,
+      "step": 6980
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8517150624860636e-05,
+      "loss": 1.1716,
+      "step": 6981
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.851610959161938e-05,
+      "loss": 1.136,
+      "step": 6982
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8515068435635178e-05,
+      "loss": 1.1294,
+      "step": 6983
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8514027156927646e-05,
+      "loss": 1.0263,
+      "step": 6984
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.851298575551639e-05,
+      "loss": 1.166,
+      "step": 6985
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.851194423142103e-05,
+      "loss": 1.1227,
+      "step": 6986
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8510902584661197e-05,
+      "loss": 1.0629,
+      "step": 6987
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8509860815256497e-05,
+      "loss": 1.0559,
+      "step": 6988
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.850881892322656e-05,
+      "loss": 1.117,
+      "step": 6989
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.850777690859101e-05,
+      "loss": 1.139,
+      "step": 6990
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8506734771369476e-05,
+      "loss": 1.14,
+      "step": 6991
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.850569251158159e-05,
+      "loss": 1.1152,
+      "step": 6992
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8504650129246985e-05,
+      "loss": 1.1121,
+      "step": 6993
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8503607624385295e-05,
+      "loss": 1.1381,
+      "step": 6994
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8502564997016156e-05,
+      "loss": 1.1483,
+      "step": 6995
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8501522247159212e-05,
+      "loss": 1.1096,
+      "step": 6996
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8500479374834098e-05,
+      "loss": 1.0993,
+      "step": 6997
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8499436380060468e-05,
+      "loss": 1.1447,
+      "step": 6998
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.849839326285796e-05,
+      "loss": 1.0927,
+      "step": 6999
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8497350023246226e-05,
+      "loss": 1.1073,
+      "step": 7000
+    },
+    {
+      "epoch": 0.3,
+      "eval_text_loss": 0.8335211277008057,
+      "eval_text_runtime": 1.3928,
+      "eval_text_samples_per_second": 53.129,
+      "eval_text_steps_per_second": 1.436,
+      "step": 7000
+    },
+    {
+      "epoch": 0.3,
+      "eval_svit_loss": 1.1838340759277344,
+      "eval_svit_runtime": 4.8601,
+      "eval_svit_samples_per_second": 140.12,
+      "eval_svit_steps_per_second": 2.263,
+      "step": 7000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.849630666124492e-05,
+      "loss": 1.0728,
+      "step": 7001
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8495263176873694e-05,
+      "loss": 1.0802,
+      "step": 7002
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8494219570152204e-05,
+      "loss": 1.101,
+      "step": 7003
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8493175841100108e-05,
+      "loss": 1.1345,
+      "step": 7004
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8492131989737067e-05,
+      "loss": 1.0993,
+      "step": 7005
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.849108801608274e-05,
+      "loss": 1.128,
+      "step": 7006
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.84900439201568e-05,
+      "loss": 1.0476,
+      "step": 7007
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8488999701978907e-05,
+      "loss": 1.1305,
+      "step": 7008
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8487955361568733e-05,
+      "loss": 1.0895,
+      "step": 7009
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.848691089894595e-05,
+      "loss": 1.0741,
+      "step": 7010
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8485866314130238e-05,
+      "loss": 1.0673,
+      "step": 7011
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8484821607141267e-05,
+      "loss": 0.9842,
+      "step": 7012
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8483776777998717e-05,
+      "loss": 1.1675,
+      "step": 7013
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.848273182672227e-05,
+      "loss": 1.0326,
+      "step": 7014
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.848168675333161e-05,
+      "loss": 0.9919,
+      "step": 7015
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8480641557846425e-05,
+      "loss": 1.0683,
+      "step": 7016
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.84795962402864e-05,
+      "loss": 1.0552,
+      "step": 7017
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8478550800671225e-05,
+      "loss": 1.2212,
+      "step": 7018
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8477505239020593e-05,
+      "loss": 1.127,
+      "step": 7019
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.84764595553542e-05,
+      "loss": 1.0527,
+      "step": 7020
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8475413749691746e-05,
+      "loss": 1.0814,
+      "step": 7021
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8474367822052926e-05,
+      "loss": 1.0932,
+      "step": 7022
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8473321772457445e-05,
+      "loss": 1.1042,
+      "step": 7023
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8472275600925005e-05,
+      "loss": 1.1072,
+      "step": 7024
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.847122930747532e-05,
+      "loss": 1.0201,
+      "step": 7025
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8470182892128086e-05,
+      "loss": 1.1044,
+      "step": 7026
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8469136354903025e-05,
+      "loss": 1.0796,
+      "step": 7027
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8468089695819846e-05,
+      "loss": 1.1021,
+      "step": 7028
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8467042914898263e-05,
+      "loss": 1.0684,
+      "step": 7029
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8465996012157997e-05,
+      "loss": 1.1097,
+      "step": 7030
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8464948987618774e-05,
+      "loss": 1.0743,
+      "step": 7031
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.84639018413003e-05,
+      "loss": 1.0709,
+      "step": 7032
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8462854573222317e-05,
+      "loss": 1.1569,
+      "step": 7033
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8461807183404547e-05,
+      "loss": 1.1534,
+      "step": 7034
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8460759671866712e-05,
+      "loss": 1.0829,
+      "step": 7035
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8459712038628553e-05,
+      "loss": 1.1305,
+      "step": 7036
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.84586642837098e-05,
+      "loss": 1.0866,
+      "step": 7037
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8457616407130192e-05,
+      "loss": 1.0578,
+      "step": 7038
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8456568408909466e-05,
+      "loss": 1.0358,
+      "step": 7039
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8455520289067362e-05,
+      "loss": 1.0702,
+      "step": 7040
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8454472047623622e-05,
+      "loss": 1.0569,
+      "step": 7041
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8453423684597996e-05,
+      "loss": 1.1799,
+      "step": 7042
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8452375200010233e-05,
+      "loss": 1.1392,
+      "step": 7043
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8451326593880073e-05,
+      "loss": 1.0988,
+      "step": 7044
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8450277866227282e-05,
+      "loss": 1.1185,
+      "step": 7045
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8449229017071606e-05,
+      "loss": 1.1355,
+      "step": 7046
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.84481800464328e-05,
+      "loss": 1.2065,
+      "step": 7047
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8447130954330633e-05,
+      "loss": 1.1175,
+      "step": 7048
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.844608174078486e-05,
+      "loss": 1.1263,
+      "step": 7049
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.844503240581524e-05,
+      "loss": 1.0571,
+      "step": 7050
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.844398294944155e-05,
+      "loss": 1.1163,
+      "step": 7051
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.844293337168355e-05,
+      "loss": 1.0313,
+      "step": 7052
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8441883672561017e-05,
+      "loss": 1.0939,
+      "step": 7053
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.844083385209372e-05,
+      "loss": 1.0747,
+      "step": 7054
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8439783910301436e-05,
+      "loss": 1.1603,
+      "step": 7055
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.843873384720394e-05,
+      "loss": 1.0865,
+      "step": 7056
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8437683662821016e-05,
+      "loss": 1.0975,
+      "step": 7057
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8436633357172444e-05,
+      "loss": 1.0834,
+      "step": 7058
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8435582930278008e-05,
+      "loss": 1.0198,
+      "step": 7059
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8434532382157497e-05,
+      "loss": 1.146,
+      "step": 7060
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8433481712830694e-05,
+      "loss": 1.108,
+      "step": 7061
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8432430922317397e-05,
+      "loss": 1.1067,
+      "step": 7062
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8431380010637395e-05,
+      "loss": 1.0816,
+      "step": 7063
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8430328977810493e-05,
+      "loss": 1.0657,
+      "step": 7064
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8429277823856473e-05,
+      "loss": 1.1392,
+      "step": 7065
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8428226548795152e-05,
+      "loss": 1.0615,
+      "step": 7066
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8427175152646322e-05,
+      "loss": 1.0898,
+      "step": 7067
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8426123635429788e-05,
+      "loss": 1.0405,
+      "step": 7068
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8425071997165364e-05,
+      "loss": 1.0535,
+      "step": 7069
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8424020237872855e-05,
+      "loss": 1.1113,
+      "step": 7070
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8422968357572078e-05,
+      "loss": 1.0465,
+      "step": 7071
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8421916356282835e-05,
+      "loss": 1.1591,
+      "step": 7072
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8420864234024954e-05,
+      "loss": 1.1081,
+      "step": 7073
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8419811990818253e-05,
+      "loss": 1.0696,
+      "step": 7074
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.841875962668255e-05,
+      "loss": 1.1496,
+      "step": 7075
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8417707141637662e-05,
+      "loss": 1.1139,
+      "step": 7076
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8416654535703426e-05,
+      "loss": 1.0534,
+      "step": 7077
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.841560180889966e-05,
+      "loss": 1.0914,
+      "step": 7078
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.84145489612462e-05,
+      "loss": 1.1319,
+      "step": 7079
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.841349599276288e-05,
+      "loss": 1.0855,
+      "step": 7080
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8412442903469528e-05,
+      "loss": 0.9868,
+      "step": 7081
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8411389693385986e-05,
+      "loss": 1.0787,
+      "step": 7082
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.841033636253209e-05,
+      "loss": 1.001,
+      "step": 7083
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8409282910927686e-05,
+      "loss": 1.1113,
+      "step": 7084
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8408229338592612e-05,
+      "loss": 1.1511,
+      "step": 7085
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.840717564554672e-05,
+      "loss": 1.0849,
+      "step": 7086
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.840612183180985e-05,
+      "loss": 1.1505,
+      "step": 7087
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.840506789740186e-05,
+      "loss": 1.1038,
+      "step": 7088
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.84040138423426e-05,
+      "loss": 1.1348,
+      "step": 7089
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8402959666651926e-05,
+      "loss": 1.1806,
+      "step": 7090
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8401905370349694e-05,
+      "loss": 1.1326,
+      "step": 7091
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.840085095345576e-05,
+      "loss": 1.1149,
+      "step": 7092
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.839979641599e-05,
+      "loss": 1.1249,
+      "step": 7093
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8398741757972257e-05,
+      "loss": 1.1255,
+      "step": 7094
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8397686979422415e-05,
+      "loss": 1.2111,
+      "step": 7095
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8396632080360333e-05,
+      "loss": 1.1079,
+      "step": 7096
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8395577060805887e-05,
+      "loss": 1.1521,
+      "step": 7097
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.839452192077895e-05,
+      "loss": 1.0859,
+      "step": 7098
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8393466660299397e-05,
+      "loss": 1.1151,
+      "step": 7099
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.83924112793871e-05,
+      "loss": 1.187,
+      "step": 7100
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8391355778061946e-05,
+      "loss": 1.1301,
+      "step": 7101
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8390300156343816e-05,
+      "loss": 1.1288,
+      "step": 7102
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8389244414252593e-05,
+      "loss": 1.1486,
+      "step": 7103
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8388188551808167e-05,
+      "loss": 1.0546,
+      "step": 7104
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.838713256903042e-05,
+      "loss": 1.0806,
+      "step": 7105
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8386076465939252e-05,
+      "loss": 1.1104,
+      "step": 7106
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.838502024255455e-05,
+      "loss": 1.1789,
+      "step": 7107
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8383963898896217e-05,
+      "loss": 1.1167,
+      "step": 7108
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8382907434984143e-05,
+      "loss": 1.1552,
+      "step": 7109
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8381850850838233e-05,
+      "loss": 1.0863,
+      "step": 7110
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8380794146478392e-05,
+      "loss": 1.1137,
+      "step": 7111
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.837973732192452e-05,
+      "loss": 1.0818,
+      "step": 7112
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8378680377196527e-05,
+      "loss": 1.1478,
+      "step": 7113
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8377623312314317e-05,
+      "loss": 1.1389,
+      "step": 7114
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.837656612729781e-05,
+      "loss": 1.0664,
+      "step": 7115
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.837550882216692e-05,
+      "loss": 1.1451,
+      "step": 7116
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8374451396941558e-05,
+      "loss": 1.1323,
+      "step": 7117
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8373393851641644e-05,
+      "loss": 1.0062,
+      "step": 7118
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8372336186287094e-05,
+      "loss": 1.0523,
+      "step": 7119
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8371278400897842e-05,
+      "loss": 1.0592,
+      "step": 7120
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8370220495493807e-05,
+      "loss": 1.0839,
+      "step": 7121
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8369162470094917e-05,
+      "loss": 1.1691,
+      "step": 7122
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.83681043247211e-05,
+      "loss": 1.1517,
+      "step": 7123
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.836704605939229e-05,
+      "loss": 1.1222,
+      "step": 7124
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8365987674128428e-05,
+      "loss": 1.1124,
+      "step": 7125
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.836492916894944e-05,
+      "loss": 1.154,
+      "step": 7126
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8363870543875265e-05,
+      "loss": 1.1844,
+      "step": 7127
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8362811798925853e-05,
+      "loss": 1.1189,
+      "step": 7128
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8361752934121144e-05,
+      "loss": 1.1249,
+      "step": 7129
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8360693949481077e-05,
+      "loss": 1.0996,
+      "step": 7130
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8359634845025614e-05,
+      "loss": 1.1927,
+      "step": 7131
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8358575620774688e-05,
+      "loss": 1.1373,
+      "step": 7132
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8357516276748265e-05,
+      "loss": 1.0811,
+      "step": 7133
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.835645681296629e-05,
+      "loss": 1.1017,
+      "step": 7134
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.835539722944873e-05,
+      "loss": 1.149,
+      "step": 7135
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.835433752621553e-05,
+      "loss": 1.1298,
+      "step": 7136
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8353277703286672e-05,
+      "loss": 1.1459,
+      "step": 7137
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.83522177606821e-05,
+      "loss": 1.0304,
+      "step": 7138
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.835115769842179e-05,
+      "loss": 1.062,
+      "step": 7139
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8350097516525707e-05,
+      "loss": 1.0072,
+      "step": 7140
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8349037215013823e-05,
+      "loss": 1.1123,
+      "step": 7141
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8347976793906115e-05,
+      "loss": 1.1675,
+      "step": 7142
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.834691625322255e-05,
+      "loss": 1.0893,
+      "step": 7143
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8345855592983108e-05,
+      "loss": 1.1034,
+      "step": 7144
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.834479481320777e-05,
+      "loss": 1.2132,
+      "step": 7145
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8343733913916518e-05,
+      "loss": 1.1715,
+      "step": 7146
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8342672895129328e-05,
+      "loss": 1.0837,
+      "step": 7147
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8341611756866203e-05,
+      "loss": 1.1159,
+      "step": 7148
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8340550499147116e-05,
+      "loss": 1.1696,
+      "step": 7149
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8339489121992063e-05,
+      "loss": 1.0287,
+      "step": 7150
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8338427625421038e-05,
+      "loss": 1.0985,
+      "step": 7151
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8337366009454038e-05,
+      "loss": 1.0313,
+      "step": 7152
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8336304274111056e-05,
+      "loss": 1.0959,
+      "step": 7153
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8335242419412096e-05,
+      "loss": 1.1663,
+      "step": 7154
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8334180445377157e-05,
+      "loss": 1.1958,
+      "step": 7155
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.833311835202624e-05,
+      "loss": 1.1301,
+      "step": 7156
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.833205613937936e-05,
+      "loss": 1.1095,
+      "step": 7157
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.833099380745652e-05,
+      "loss": 1.0607,
+      "step": 7158
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8329931356277735e-05,
+      "loss": 1.121,
+      "step": 7159
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8328868785863012e-05,
+      "loss": 1.0729,
+      "step": 7160
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8327806096232373e-05,
+      "loss": 1.1588,
+      "step": 7161
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8326743287405827e-05,
+      "loss": 1.0921,
+      "step": 7162
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8325680359403406e-05,
+      "loss": 1.0539,
+      "step": 7163
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8324617312245124e-05,
+      "loss": 1.1595,
+      "step": 7164
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8323554145951003e-05,
+      "loss": 1.0682,
+      "step": 7165
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.832249086054108e-05,
+      "loss": 1.0452,
+      "step": 7166
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8321427456035374e-05,
+      "loss": 1.1276,
+      "step": 7167
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8320363932453923e-05,
+      "loss": 1.1841,
+      "step": 7168
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8319300289816754e-05,
+      "loss": 1.1302,
+      "step": 7169
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.831823652814391e-05,
+      "loss": 1.1253,
+      "step": 7170
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8317172647455422e-05,
+      "loss": 1.087,
+      "step": 7171
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8316108647771334e-05,
+      "loss": 1.1189,
+      "step": 7172
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.831504452911169e-05,
+      "loss": 1.1655,
+      "step": 7173
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.831398029149653e-05,
+      "loss": 1.1385,
+      "step": 7174
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8312915934945907e-05,
+      "loss": 1.1574,
+      "step": 7175
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8311851459479862e-05,
+      "loss": 1.1325,
+      "step": 7176
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8310786865118455e-05,
+      "loss": 1.0528,
+      "step": 7177
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.830972215188173e-05,
+      "loss": 1.1406,
+      "step": 7178
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8308657319789753e-05,
+      "loss": 1.0679,
+      "step": 7179
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8307592368862573e-05,
+      "loss": 1.1734,
+      "step": 7180
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.830652729912026e-05,
+      "loss": 1.1251,
+      "step": 7181
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8305462110582864e-05,
+      "loss": 1.044,
+      "step": 7182
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.830439680327046e-05,
+      "loss": 1.1746,
+      "step": 7183
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8303331377203114e-05,
+      "loss": 1.0241,
+      "step": 7184
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8302265832400894e-05,
+      "loss": 1.0323,
+      "step": 7185
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.830120016888387e-05,
+      "loss": 1.1035,
+      "step": 7186
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8300134386672113e-05,
+      "loss": 1.104,
+      "step": 7187
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8299068485785702e-05,
+      "loss": 1.0448,
+      "step": 7188
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8298002466244717e-05,
+      "loss": 1.1017,
+      "step": 7189
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.829693632806924e-05,
+      "loss": 1.1348,
+      "step": 7190
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8295870071279347e-05,
+      "loss": 1.0712,
+      "step": 7191
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.829480369589513e-05,
+      "loss": 1.0714,
+      "step": 7192
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8293737201936668e-05,
+      "loss": 1.0256,
+      "step": 7193
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8292670589424058e-05,
+      "loss": 1.055,
+      "step": 7194
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.829160385837739e-05,
+      "loss": 1.1256,
+      "step": 7195
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8290537008816754e-05,
+      "loss": 1.1227,
+      "step": 7196
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.828947004076225e-05,
+      "loss": 1.1592,
+      "step": 7197
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8288402954233977e-05,
+      "loss": 1.1937,
+      "step": 7198
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8287335749252032e-05,
+      "loss": 1.0463,
+      "step": 7199
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8286268425836518e-05,
+      "loss": 1.0692,
+      "step": 7200
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8285200984007545e-05,
+      "loss": 1.067,
+      "step": 7201
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8284133423785214e-05,
+      "loss": 1.0695,
+      "step": 7202
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8283065745189635e-05,
+      "loss": 1.231,
+      "step": 7203
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8281997948240927e-05,
+      "loss": 1.1013,
+      "step": 7204
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8280930032959192e-05,
+      "loss": 1.0245,
+      "step": 7205
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.827986199936456e-05,
+      "loss": 1.0545,
+      "step": 7206
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8278793847477134e-05,
+      "loss": 1.1079,
+      "step": 7207
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.827772557731705e-05,
+      "loss": 1.1243,
+      "step": 7208
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.827665718890442e-05,
+      "loss": 1.1219,
+      "step": 7209
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8275588682259376e-05,
+      "loss": 1.176,
+      "step": 7210
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8274520057402038e-05,
+      "loss": 1.0731,
+      "step": 7211
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.827345131435254e-05,
+      "loss": 1.0843,
+      "step": 7212
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8272382453131018e-05,
+      "loss": 1.0083,
+      "step": 7213
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8271313473757597e-05,
+      "loss": 1.0987,
+      "step": 7214
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.827024437625242e-05,
+      "loss": 1.0456,
+      "step": 7215
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.826917516063562e-05,
+      "loss": 1.0604,
+      "step": 7216
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8268105826927342e-05,
+      "loss": 1.1335,
+      "step": 7217
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.826703637514773e-05,
+      "loss": 1.0267,
+      "step": 7218
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.826596680531692e-05,
+      "loss": 1.121,
+      "step": 7219
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.826489711745507e-05,
+      "loss": 1.1309,
+      "step": 7220
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8263827311582323e-05,
+      "loss": 1.0658,
+      "step": 7221
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8262757387718835e-05,
+      "loss": 1.1133,
+      "step": 7222
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.826168734588476e-05,
+      "loss": 1.0549,
+      "step": 7223
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8260617186100253e-05,
+      "loss": 1.1571,
+      "step": 7224
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8259546908385467e-05,
+      "loss": 1.0981,
+      "step": 7225
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8258476512760572e-05,
+      "loss": 1.1445,
+      "step": 7226
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8257405999245723e-05,
+      "loss": 1.128,
+      "step": 7227
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.825633536786109e-05,
+      "loss": 1.1469,
+      "step": 7228
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8255264618626836e-05,
+      "loss": 1.0207,
+      "step": 7229
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.825419375156314e-05,
+      "loss": 1.173,
+      "step": 7230
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8253122766690158e-05,
+      "loss": 1.1797,
+      "step": 7231
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8252051664028078e-05,
+      "loss": 1.0668,
+      "step": 7232
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.825098044359707e-05,
+      "loss": 1.1185,
+      "step": 7233
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8249909105417317e-05,
+      "loss": 1.1494,
+      "step": 7234
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8248837649508992e-05,
+      "loss": 1.0642,
+      "step": 7235
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8247766075892284e-05,
+      "loss": 1.1118,
+      "step": 7236
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8246694384587374e-05,
+      "loss": 1.0893,
+      "step": 7237
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8245622575614453e-05,
+      "loss": 1.1519,
+      "step": 7238
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.824455064899371e-05,
+      "loss": 1.1266,
+      "step": 7239
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8243478604745335e-05,
+      "loss": 1.1043,
+      "step": 7240
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.824240644288952e-05,
+      "loss": 1.1638,
+      "step": 7241
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8241334163446467e-05,
+      "loss": 1.1102,
+      "step": 7242
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8240261766436372e-05,
+      "loss": 1.1461,
+      "step": 7243
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8239189251879433e-05,
+      "loss": 1.0992,
+      "step": 7244
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8238116619795852e-05,
+      "loss": 1.0865,
+      "step": 7245
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.823704387020584e-05,
+      "loss": 1.1425,
+      "step": 7246
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8235971003129597e-05,
+      "loss": 1.1305,
+      "step": 7247
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8234898018587337e-05,
+      "loss": 1.0253,
+      "step": 7248
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8233824916599272e-05,
+      "loss": 1.1812,
+      "step": 7249
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8232751697185618e-05,
+      "loss": 1.1427,
+      "step": 7250
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8231678360366583e-05,
+      "loss": 1.0272,
+      "step": 7251
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8230604906162388e-05,
+      "loss": 1.1407,
+      "step": 7252
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8229531334593258e-05,
+      "loss": 1.0347,
+      "step": 7253
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8228457645679415e-05,
+      "loss": 1.0894,
+      "step": 7254
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8227383839441084e-05,
+      "loss": 1.1437,
+      "step": 7255
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8226309915898484e-05,
+      "loss": 1.04,
+      "step": 7256
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8225235875071854e-05,
+      "loss": 1.0044,
+      "step": 7257
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.822416171698142e-05,
+      "loss": 1.0708,
+      "step": 7258
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.822308744164742e-05,
+      "loss": 1.0928,
+      "step": 7259
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8222013049090087e-05,
+      "loss": 1.1186,
+      "step": 7260
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8220938539329663e-05,
+      "loss": 1.0546,
+      "step": 7261
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8219863912386378e-05,
+      "loss": 1.1698,
+      "step": 7262
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.821878916828049e-05,
+      "loss": 1.0919,
+      "step": 7263
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.821771430703223e-05,
+      "loss": 1.1178,
+      "step": 7264
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8216639328661853e-05,
+      "loss": 1.1196,
+      "step": 7265
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8215564233189607e-05,
+      "loss": 1.1415,
+      "step": 7266
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8214489020635738e-05,
+      "loss": 1.0592,
+      "step": 7267
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8213413691020506e-05,
+      "loss": 1.122,
+      "step": 7268
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8212338244364163e-05,
+      "loss": 1.122,
+      "step": 7269
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.821126268068697e-05,
+      "loss": 1.0633,
+      "step": 7270
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.821018700000919e-05,
+      "loss": 1.0814,
+      "step": 7271
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8209111202351077e-05,
+      "loss": 1.1234,
+      "step": 7272
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.82080352877329e-05,
+      "loss": 1.1023,
+      "step": 7273
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.820695925617493e-05,
+      "loss": 1.1385,
+      "step": 7274
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.820588310769743e-05,
+      "loss": 1.1248,
+      "step": 7275
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8204806842320668e-05,
+      "loss": 1.1256,
+      "step": 7276
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8203730460064927e-05,
+      "loss": 1.1219,
+      "step": 7277
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8202653960950475e-05,
+      "loss": 1.119,
+      "step": 7278
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.82015773449976e-05,
+      "loss": 1.0757,
+      "step": 7279
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8200500612226568e-05,
+      "loss": 1.1599,
+      "step": 7280
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.819942376265767e-05,
+      "loss": 1.1261,
+      "step": 7281
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.819834679631119e-05,
+      "loss": 1.0323,
+      "step": 7282
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8197269713207415e-05,
+      "loss": 1.1694,
+      "step": 7283
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.819619251336663e-05,
+      "loss": 1.157,
+      "step": 7284
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8195115196809128e-05,
+      "loss": 1.0443,
+      "step": 7285
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8194037763555204e-05,
+      "loss": 1.0012,
+      "step": 7286
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.819296021362515e-05,
+      "loss": 1.1847,
+      "step": 7287
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.819188254703927e-05,
+      "loss": 1.1062,
+      "step": 7288
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8190804763817855e-05,
+      "loss": 1.0795,
+      "step": 7289
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8189726863981213e-05,
+      "loss": 1.1334,
+      "step": 7290
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8188648847549646e-05,
+      "loss": 1.123,
+      "step": 7291
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8187570714543462e-05,
+      "loss": 1.0323,
+      "step": 7292
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.818649246498297e-05,
+      "loss": 1.1357,
+      "step": 7293
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8185414098888476e-05,
+      "loss": 1.0155,
+      "step": 7294
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8184335616280302e-05,
+      "loss": 1.1321,
+      "step": 7295
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8183257017178755e-05,
+      "loss": 1.1309,
+      "step": 7296
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8182178301604153e-05,
+      "loss": 1.1898,
+      "step": 7297
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8181099469576823e-05,
+      "loss": 1.0845,
+      "step": 7298
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.818002052111708e-05,
+      "loss": 1.1199,
+      "step": 7299
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8178941456245242e-05,
+      "loss": 1.1391,
+      "step": 7300
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.817786227498165e-05,
+      "loss": 1.057,
+      "step": 7301
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8176782977346628e-05,
+      "loss": 1.0988,
+      "step": 7302
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.81757035633605e-05,
+      "loss": 1.0777,
+      "step": 7303
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8174624033043602e-05,
+      "loss": 1.1128,
+      "step": 7304
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.817354438641627e-05,
+      "loss": 1.1124,
+      "step": 7305
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8172464623498842e-05,
+      "loss": 1.0776,
+      "step": 7306
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8171384744311656e-05,
+      "loss": 1.1029,
+      "step": 7307
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.817030474887505e-05,
+      "loss": 1.1994,
+      "step": 7308
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8169224637209372e-05,
+      "loss": 1.0369,
+      "step": 7309
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8168144409334967e-05,
+      "loss": 1.0355,
+      "step": 7310
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8167064065272185e-05,
+      "loss": 1.1019,
+      "step": 7311
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8165983605041376e-05,
+      "loss": 1.0489,
+      "step": 7312
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8164903028662888e-05,
+      "loss": 1.1761,
+      "step": 7313
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8163822336157077e-05,
+      "loss": 1.0424,
+      "step": 7314
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8162741527544304e-05,
+      "loss": 1.0869,
+      "step": 7315
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8161660602844926e-05,
+      "loss": 1.1138,
+      "step": 7316
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8160579562079303e-05,
+      "loss": 1.1798,
+      "step": 7317
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8159498405267795e-05,
+      "loss": 1.1767,
+      "step": 7318
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8158417132430778e-05,
+      "loss": 1.0259,
+      "step": 7319
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8157335743588607e-05,
+      "loss": 1.0511,
+      "step": 7320
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8156254238761666e-05,
+      "loss": 1.0776,
+      "step": 7321
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.815517261797031e-05,
+      "loss": 1.1027,
+      "step": 7322
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.815409088123493e-05,
+      "loss": 1.1473,
+      "step": 7323
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8153009028575895e-05,
+      "loss": 1.1708,
+      "step": 7324
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.815192706001358e-05,
+      "loss": 1.1049,
+      "step": 7325
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.815084497556837e-05,
+      "loss": 1.1039,
+      "step": 7326
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.814976277526065e-05,
+      "loss": 1.1476,
+      "step": 7327
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8148680459110804e-05,
+      "loss": 1.0043,
+      "step": 7328
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8147598027139217e-05,
+      "loss": 1.0903,
+      "step": 7329
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8146515479366283e-05,
+      "loss": 1.0567,
+      "step": 7330
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8145432815812385e-05,
+      "loss": 1.0939,
+      "step": 7331
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8144350036497927e-05,
+      "loss": 1.2029,
+      "step": 7332
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8143267141443297e-05,
+      "loss": 1.1055,
+      "step": 7333
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8142184130668903e-05,
+      "loss": 1.1397,
+      "step": 7334
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8141101004195135e-05,
+      "loss": 1.1385,
+      "step": 7335
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8140017762042402e-05,
+      "loss": 1.0558,
+      "step": 7336
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.813893440423111e-05,
+      "loss": 1.115,
+      "step": 7337
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.813785093078166e-05,
+      "loss": 1.1134,
+      "step": 7338
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.813676734171447e-05,
+      "loss": 1.0931,
+      "step": 7339
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.813568363704994e-05,
+      "loss": 1.0914,
+      "step": 7340
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8134599816808492e-05,
+      "loss": 1.0787,
+      "step": 7341
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.813351588101054e-05,
+      "loss": 1.1525,
+      "step": 7342
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8132431829676498e-05,
+      "loss": 1.2261,
+      "step": 7343
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8131347662826794e-05,
+      "loss": 1.1055,
+      "step": 7344
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8130263380481844e-05,
+      "loss": 1.143,
+      "step": 7345
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8129178982662074e-05,
+      "loss": 1.0851,
+      "step": 7346
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.812809446938791e-05,
+      "loss": 1.1126,
+      "step": 7347
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8127009840679782e-05,
+      "loss": 1.1439,
+      "step": 7348
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8125925096558124e-05,
+      "loss": 1.1003,
+      "step": 7349
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8124840237043362e-05,
+      "loss": 1.0023,
+      "step": 7350
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8123755262155936e-05,
+      "loss": 1.0041,
+      "step": 7351
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.812267017191628e-05,
+      "loss": 1.0008,
+      "step": 7352
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8121584966344838e-05,
+      "loss": 1.163,
+      "step": 7353
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.812049964546205e-05,
+      "loss": 1.1186,
+      "step": 7354
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8119414209288363e-05,
+      "loss": 1.1113,
+      "step": 7355
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8118328657844217e-05,
+      "loss": 1.0719,
+      "step": 7356
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8117242991150065e-05,
+      "loss": 1.1037,
+      "step": 7357
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8116157209226357e-05,
+      "loss": 1.1082,
+      "step": 7358
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8115071312093544e-05,
+      "loss": 1.1971,
+      "step": 7359
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8113985299772084e-05,
+      "loss": 1.0827,
+      "step": 7360
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.811289917228243e-05,
+      "loss": 1.0817,
+      "step": 7361
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.811181292964504e-05,
+      "loss": 1.0828,
+      "step": 7362
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8110726571880383e-05,
+      "loss": 1.1157,
+      "step": 7363
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8109640099008915e-05,
+      "loss": 1.0933,
+      "step": 7364
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8108553511051108e-05,
+      "loss": 1.1047,
+      "step": 7365
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.810746680802742e-05,
+      "loss": 1.0087,
+      "step": 7366
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8106379989958335e-05,
+      "loss": 1.075,
+      "step": 7367
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8105293056864315e-05,
+      "loss": 1.1245,
+      "step": 7368
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8104206008765837e-05,
+      "loss": 1.0042,
+      "step": 7369
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8103118845683382e-05,
+      "loss": 1.1267,
+      "step": 7370
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8102031567637417e-05,
+      "loss": 1.1827,
+      "step": 7371
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8100944174648434e-05,
+      "loss": 1.1242,
+      "step": 7372
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8099856666736915e-05,
+      "loss": 1.1221,
+      "step": 7373
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.809876904392334e-05,
+      "loss": 1.1838,
+      "step": 7374
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.80976813062282e-05,
+      "loss": 1.1208,
+      "step": 7375
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8096593453671985e-05,
+      "loss": 1.0683,
+      "step": 7376
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.809550548627518e-05,
+      "loss": 1.0239,
+      "step": 7377
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.809441740405829e-05,
+      "loss": 1.1173,
+      "step": 7378
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8093329207041806e-05,
+      "loss": 1.0612,
+      "step": 7379
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.809224089524622e-05,
+      "loss": 1.0533,
+      "step": 7380
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.809115246869204e-05,
+      "loss": 1.1174,
+      "step": 7381
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8090063927399766e-05,
+      "loss": 1.1443,
+      "step": 7382
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8088975271389906e-05,
+      "loss": 1.0646,
+      "step": 7383
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.808788650068296e-05,
+      "loss": 1.0813,
+      "step": 7384
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8086797615299444e-05,
+      "loss": 1.1634,
+      "step": 7385
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8085708615259865e-05,
+      "loss": 1.1492,
+      "step": 7386
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8084619500584737e-05,
+      "loss": 1.0248,
+      "step": 7387
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8083530271294573e-05,
+      "loss": 1.0863,
+      "step": 7388
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8082440927409897e-05,
+      "loss": 1.0806,
+      "step": 7389
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8081351468951225e-05,
+      "loss": 1.1666,
+      "step": 7390
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.808026189593908e-05,
+      "loss": 1.1129,
+      "step": 7391
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8079172208393986e-05,
+      "loss": 1.0993,
+      "step": 7392
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8078082406336467e-05,
+      "loss": 1.0503,
+      "step": 7393
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8076992489787052e-05,
+      "loss": 1.1563,
+      "step": 7394
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8075902458766275e-05,
+      "loss": 1.1358,
+      "step": 7395
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8074812313294666e-05,
+      "loss": 1.1853,
+      "step": 7396
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8073722053392755e-05,
+      "loss": 1.1945,
+      "step": 7397
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.807263167908109e-05,
+      "loss": 1.1161,
+      "step": 7398
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8071541190380206e-05,
+      "loss": 1.119,
+      "step": 7399
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8070450587310642e-05,
+      "loss": 1.0958,
+      "step": 7400
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8069359869892938e-05,
+      "loss": 1.1151,
+      "step": 7401
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8068269038147646e-05,
+      "loss": 1.1306,
+      "step": 7402
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8067178092095313e-05,
+      "loss": 1.0932,
+      "step": 7403
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8066087031756486e-05,
+      "loss": 1.1445,
+      "step": 7404
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8064995857151725e-05,
+      "loss": 1.108,
+      "step": 7405
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8063904568301573e-05,
+      "loss": 1.1563,
+      "step": 7406
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8062813165226595e-05,
+      "loss": 1.0396,
+      "step": 7407
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8061721647947344e-05,
+      "loss": 1.0851,
+      "step": 7408
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8060630016484383e-05,
+      "loss": 1.123,
+      "step": 7409
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8059538270858276e-05,
+      "loss": 1.0707,
+      "step": 7410
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8058446411089588e-05,
+      "loss": 1.1218,
+      "step": 7411
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8057354437198883e-05,
+      "loss": 1.1103,
+      "step": 7412
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8056262349206735e-05,
+      "loss": 1.0727,
+      "step": 7413
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.805517014713371e-05,
+      "loss": 1.0759,
+      "step": 7414
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.805407783100039e-05,
+      "loss": 1.2135,
+      "step": 7415
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.805298540082734e-05,
+      "loss": 1.066,
+      "step": 7416
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8051892856635147e-05,
+      "loss": 1.0376,
+      "step": 7417
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8050800198444387e-05,
+      "loss": 1.0329,
+      "step": 7418
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.804970742627564e-05,
+      "loss": 1.0386,
+      "step": 7419
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8048614540149495e-05,
+      "loss": 1.0703,
+      "step": 7420
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8047521540086537e-05,
+      "loss": 1.0323,
+      "step": 7421
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8046428426107355e-05,
+      "loss": 1.1878,
+      "step": 7422
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.804533519823254e-05,
+      "loss": 1.1105,
+      "step": 7423
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.804424185648268e-05,
+      "loss": 1.1074,
+      "step": 7424
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8043148400878378e-05,
+      "loss": 1.1433,
+      "step": 7425
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8042054831440228e-05,
+      "loss": 1.1289,
+      "step": 7426
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8040961148188826e-05,
+      "loss": 1.0945,
+      "step": 7427
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.803986735114478e-05,
+      "loss": 1.0858,
+      "step": 7428
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8038773440328685e-05,
+      "loss": 1.194,
+      "step": 7429
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8037679415761156e-05,
+      "loss": 1.1858,
+      "step": 7430
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8036585277462798e-05,
+      "loss": 1.1311,
+      "step": 7431
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8035491025454217e-05,
+      "loss": 1.0851,
+      "step": 7432
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8034396659756026e-05,
+      "loss": 1.1696,
+      "step": 7433
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.803330218038885e-05,
+      "loss": 1.103,
+      "step": 7434
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8032207587373292e-05,
+      "loss": 1.111,
+      "step": 7435
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8031112880729975e-05,
+      "loss": 1.1529,
+      "step": 7436
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8030018060479522e-05,
+      "loss": 1.0778,
+      "step": 7437
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8028923126642555e-05,
+      "loss": 1.0863,
+      "step": 7438
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8027828079239695e-05,
+      "loss": 1.1017,
+      "step": 7439
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8026732918291578e-05,
+      "loss": 1.1159,
+      "step": 7440
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.802563764381883e-05,
+      "loss": 1.2237,
+      "step": 7441
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8024542255842078e-05,
+      "loss": 1.0644,
+      "step": 7442
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.802344675438196e-05,
+      "loss": 1.1445,
+      "step": 7443
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8022351139459107e-05,
+      "loss": 1.1826,
+      "step": 7444
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8021255411094163e-05,
+      "loss": 1.1668,
+      "step": 7445
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8020159569307763e-05,
+      "loss": 1.198,
+      "step": 7446
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.801906361412055e-05,
+      "loss": 1.1349,
+      "step": 7447
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8017967545553175e-05,
+      "loss": 1.1237,
+      "step": 7448
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8016871363626278e-05,
+      "loss": 1.0795,
+      "step": 7449
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.801577506836051e-05,
+      "loss": 1.1251,
+      "step": 7450
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.801467865977652e-05,
+      "loss": 1.1747,
+      "step": 7451
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.801358213789496e-05,
+      "loss": 1.2224,
+      "step": 7452
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8012485502736487e-05,
+      "loss": 1.0886,
+      "step": 7453
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8011388754321758e-05,
+      "loss": 1.1773,
+      "step": 7454
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8010291892671433e-05,
+      "loss": 1.0887,
+      "step": 7455
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8009194917806172e-05,
+      "loss": 1.1177,
+      "step": 7456
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8008097829746638e-05,
+      "loss": 1.1229,
+      "step": 7457
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.80070006285135e-05,
+      "loss": 1.1091,
+      "step": 7458
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8005903314127418e-05,
+      "loss": 1.1718,
+      "step": 7459
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8004805886609075e-05,
+      "loss": 1.1244,
+      "step": 7460
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8003708345979132e-05,
+      "loss": 1.1724,
+      "step": 7461
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8002610692258266e-05,
+      "loss": 1.0806,
+      "step": 7462
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8001512925467154e-05,
+      "loss": 1.1468,
+      "step": 7463
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8000415045626476e-05,
+      "loss": 1.1522,
+      "step": 7464
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.799931705275691e-05,
+      "loss": 1.1708,
+      "step": 7465
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.799821894687914e-05,
+      "loss": 1.0545,
+      "step": 7466
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.799712072801385e-05,
+      "loss": 1.1456,
+      "step": 7467
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.799602239618173e-05,
+      "loss": 1.1779,
+      "step": 7468
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7994923951403465e-05,
+      "loss": 1.0663,
+      "step": 7469
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7993825393699747e-05,
+      "loss": 0.9816,
+      "step": 7470
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7992726723091274e-05,
+      "loss": 1.1552,
+      "step": 7471
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7991627939598733e-05,
+      "loss": 1.1369,
+      "step": 7472
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.799052904324283e-05,
+      "loss": 1.1266,
+      "step": 7473
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.798943003404426e-05,
+      "loss": 1.1,
+      "step": 7474
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7988330912023726e-05,
+      "loss": 1.1404,
+      "step": 7475
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.798723167720193e-05,
+      "loss": 1.1164,
+      "step": 7476
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7986132329599586e-05,
+      "loss": 1.0941,
+      "step": 7477
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7985032869237393e-05,
+      "loss": 1.1052,
+      "step": 7478
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.798393329613606e-05,
+      "loss": 1.0926,
+      "step": 7479
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7982833610316312e-05,
+      "loss": 1.0811,
+      "step": 7480
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7981733811798855e-05,
+      "loss": 1.1791,
+      "step": 7481
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7980633900604402e-05,
+      "loss": 1.0975,
+      "step": 7482
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7979533876753676e-05,
+      "loss": 1.0975,
+      "step": 7483
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7978433740267403e-05,
+      "loss": 1.0923,
+      "step": 7484
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.79773334911663e-05,
+      "loss": 1.0543,
+      "step": 7485
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7976233129471093e-05,
+      "loss": 1.1007,
+      "step": 7486
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.797513265520251e-05,
+      "loss": 1.131,
+      "step": 7487
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.797403206838128e-05,
+      "loss": 1.1848,
+      "step": 7488
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.797293136902814e-05,
+      "loss": 1.1318,
+      "step": 7489
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7971830557163814e-05,
+      "loss": 1.0572,
+      "step": 7490
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7970729632809045e-05,
+      "loss": 1.0354,
+      "step": 7491
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7969628595984567e-05,
+      "loss": 1.0896,
+      "step": 7492
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7968527446711122e-05,
+      "loss": 1.062,
+      "step": 7493
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.796742618500945e-05,
+      "loss": 1.0135,
+      "step": 7494
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.79663248109003e-05,
+      "loss": 1.1359,
+      "step": 7495
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7965223324404415e-05,
+      "loss": 1.1537,
+      "step": 7496
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7964121725542545e-05,
+      "loss": 1.0936,
+      "step": 7497
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.796302001433544e-05,
+      "loss": 1.0892,
+      "step": 7498
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.796191819080385e-05,
+      "loss": 1.1586,
+      "step": 7499
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.796081625496853e-05,
+      "loss": 1.0205,
+      "step": 7500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7959714206850245e-05,
+      "loss": 1.0813,
+      "step": 7501
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7958612046469746e-05,
+      "loss": 1.082,
+      "step": 7502
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.79575097738478e-05,
+      "loss": 1.2338,
+      "step": 7503
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7956407389005162e-05,
+      "loss": 1.1258,
+      "step": 7504
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7955304891962602e-05,
+      "loss": 1.0593,
+      "step": 7505
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7954202282740893e-05,
+      "loss": 1.0624,
+      "step": 7506
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7953099561360796e-05,
+      "loss": 1.1881,
+      "step": 7507
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7951996727843088e-05,
+      "loss": 1.1348,
+      "step": 7508
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7950893782208542e-05,
+      "loss": 1.0531,
+      "step": 7509
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7949790724477934e-05,
+      "loss": 1.0765,
+      "step": 7510
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.794868755467204e-05,
+      "loss": 1.1024,
+      "step": 7511
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.794758427281164e-05,
+      "loss": 1.1454,
+      "step": 7512
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.794648087891752e-05,
+      "loss": 1.1126,
+      "step": 7513
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7945377373010464e-05,
+      "loss": 1.1337,
+      "step": 7514
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7944273755111254e-05,
+      "loss": 1.0571,
+      "step": 7515
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7943170025240687e-05,
+      "loss": 1.0713,
+      "step": 7516
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7942066183419546e-05,
+      "loss": 1.0224,
+      "step": 7517
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7940962229668626e-05,
+      "loss": 1.1047,
+      "step": 7518
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7939858164008722e-05,
+      "loss": 1.0908,
+      "step": 7519
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7938753986460635e-05,
+      "loss": 1.1596,
+      "step": 7520
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7937649697045157e-05,
+      "loss": 1.0318,
+      "step": 7521
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7936545295783095e-05,
+      "loss": 1.0668,
+      "step": 7522
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.793544078269525e-05,
+      "loss": 1.081,
+      "step": 7523
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.793433615780243e-05,
+      "loss": 1.0986,
+      "step": 7524
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.793323142112544e-05,
+      "loss": 1.1761,
+      "step": 7525
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.793212657268509e-05,
+      "loss": 1.1426,
+      "step": 7526
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.793102161250219e-05,
+      "loss": 1.1185,
+      "step": 7527
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.792991654059756e-05,
+      "loss": 1.115,
+      "step": 7528
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7928811356992014e-05,
+      "loss": 1.0578,
+      "step": 7529
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7927706061706365e-05,
+      "loss": 1.1103,
+      "step": 7530
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7926600654761437e-05,
+      "loss": 1.1177,
+      "step": 7531
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7925495136178056e-05,
+      "loss": 1.0958,
+      "step": 7532
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.792438950597704e-05,
+      "loss": 1.1579,
+      "step": 7533
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7923283764179215e-05,
+      "loss": 1.112,
+      "step": 7534
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7922177910805417e-05,
+      "loss": 1.0985,
+      "step": 7535
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7921071945876472e-05,
+      "loss": 1.0862,
+      "step": 7536
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7919965869413217e-05,
+      "loss": 1.1019,
+      "step": 7537
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.791885968143648e-05,
+      "loss": 1.1991,
+      "step": 7538
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7917753381967103e-05,
+      "loss": 1.0282,
+      "step": 7539
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7916646971025926e-05,
+      "loss": 1.0771,
+      "step": 7540
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7915540448633786e-05,
+      "loss": 1.0655,
+      "step": 7541
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7914433814811525e-05,
+      "loss": 1.1163,
+      "step": 7542
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.791332706958e-05,
+      "loss": 1.1204,
+      "step": 7543
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7912220212960043e-05,
+      "loss": 1.113,
+      "step": 7544
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7911113244972517e-05,
+      "loss": 1.0881,
+      "step": 7545
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7910006165638268e-05,
+      "loss": 1.1443,
+      "step": 7546
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7908898974978147e-05,
+      "loss": 1.0933,
+      "step": 7547
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7907791673013017e-05,
+      "loss": 1.1185,
+      "step": 7548
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.790668425976373e-05,
+      "loss": 1.1566,
+      "step": 7549
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.790557673525115e-05,
+      "loss": 1.1272,
+      "step": 7550
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7904469099496135e-05,
+      "loss": 1.1597,
+      "step": 7551
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7903361352519553e-05,
+      "loss": 1.1184,
+      "step": 7552
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.790225349434227e-05,
+      "loss": 1.1252,
+      "step": 7553
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7901145524985156e-05,
+      "loss": 1.0748,
+      "step": 7554
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7900037444469073e-05,
+      "loss": 1.0506,
+      "step": 7555
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7898929252814908e-05,
+      "loss": 1.0621,
+      "step": 7556
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7897820950043522e-05,
+      "loss": 1.1114,
+      "step": 7557
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.78967125361758e-05,
+      "loss": 1.1019,
+      "step": 7558
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7895604011232622e-05,
+      "loss": 1.0458,
+      "step": 7559
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7894495375234866e-05,
+      "loss": 1.0792,
+      "step": 7560
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.789338662820341e-05,
+      "loss": 1.1686,
+      "step": 7561
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7892277770159153e-05,
+      "loss": 1.1382,
+      "step": 7562
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7891168801122967e-05,
+      "loss": 1.0542,
+      "step": 7563
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7890059721115755e-05,
+      "loss": 1.1311,
+      "step": 7564
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.78889505301584e-05,
+      "loss": 1.1679,
+      "step": 7565
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7887841228271795e-05,
+      "loss": 1.0322,
+      "step": 7566
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.788673181547684e-05,
+      "loss": 1.067,
+      "step": 7567
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.788562229179443e-05,
+      "loss": 1.0277,
+      "step": 7568
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7884512657245472e-05,
+      "loss": 1.0223,
+      "step": 7569
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7883402911850857e-05,
+      "loss": 1.0307,
+      "step": 7570
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.78822930556315e-05,
+      "loss": 1.1153,
+      "step": 7571
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7881183088608294e-05,
+      "loss": 1.133,
+      "step": 7572
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7880073010802158e-05,
+      "loss": 1.0901,
+      "step": 7573
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7878962822234003e-05,
+      "loss": 1.1974,
+      "step": 7574
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7877852522924734e-05,
+      "loss": 1.0641,
+      "step": 7575
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.787674211289527e-05,
+      "loss": 1.064,
+      "step": 7576
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7875631592166527e-05,
+      "loss": 1.0859,
+      "step": 7577
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7874520960759424e-05,
+      "loss": 1.0161,
+      "step": 7578
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7873410218694882e-05,
+      "loss": 1.087,
+      "step": 7579
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.787229936599382e-05,
+      "loss": 1.1405,
+      "step": 7580
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7871188402677173e-05,
+      "loss": 1.0449,
+      "step": 7581
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7870077328765858e-05,
+      "loss": 1.0641,
+      "step": 7582
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7868966144280805e-05,
+      "loss": 1.0378,
+      "step": 7583
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.786785484924295e-05,
+      "loss": 1.1315,
+      "step": 7584
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7866743443673222e-05,
+      "loss": 1.1072,
+      "step": 7585
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7865631927592562e-05,
+      "loss": 1.1139,
+      "step": 7586
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.78645203010219e-05,
+      "loss": 1.1708,
+      "step": 7587
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.786340856398218e-05,
+      "loss": 1.0975,
+      "step": 7588
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7862296716494344e-05,
+      "loss": 1.1021,
+      "step": 7589
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7861184758579333e-05,
+      "loss": 1.0835,
+      "step": 7590
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7860072690258098e-05,
+      "loss": 1.1136,
+      "step": 7591
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.785896051155158e-05,
+      "loss": 1.0669,
+      "step": 7592
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7857848222480734e-05,
+      "loss": 1.0157,
+      "step": 7593
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.785673582306651e-05,
+      "loss": 1.1406,
+      "step": 7594
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7855623313329863e-05,
+      "loss": 1.0916,
+      "step": 7595
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7854510693291745e-05,
+      "loss": 1.1841,
+      "step": 7596
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.785339796297312e-05,
+      "loss": 1.0635,
+      "step": 7597
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7852285122394944e-05,
+      "loss": 1.0894,
+      "step": 7598
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7851172171578185e-05,
+      "loss": 1.0116,
+      "step": 7599
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.78500591105438e-05,
+      "loss": 1.0838,
+      "step": 7600
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7848945939312763e-05,
+      "loss": 1.1444,
+      "step": 7601
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.784783265790604e-05,
+      "loss": 1.0695,
+      "step": 7602
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.78467192663446e-05,
+      "loss": 1.0678,
+      "step": 7603
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7845605764649416e-05,
+      "loss": 1.0503,
+      "step": 7604
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.784449215284146e-05,
+      "loss": 1.0996,
+      "step": 7605
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7843378430941716e-05,
+      "loss": 1.1295,
+      "step": 7606
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7842264598971158e-05,
+      "loss": 1.0455,
+      "step": 7607
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7841150656950773e-05,
+      "loss": 1.0699,
+      "step": 7608
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7840036604901536e-05,
+      "loss": 1.0782,
+      "step": 7609
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7838922442844437e-05,
+      "loss": 1.0476,
+      "step": 7610
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.783780817080046e-05,
+      "loss": 1.0546,
+      "step": 7611
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7836693788790598e-05,
+      "loss": 1.1257,
+      "step": 7612
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.783557929683584e-05,
+      "loss": 1.036,
+      "step": 7613
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7834464694957182e-05,
+      "loss": 1.2083,
+      "step": 7614
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.783334998317562e-05,
+      "loss": 1.1149,
+      "step": 7615
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7832235161512145e-05,
+      "loss": 1.0655,
+      "step": 7616
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7831120229987762e-05,
+      "loss": 1.0887,
+      "step": 7617
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7830005188623477e-05,
+      "loss": 1.045,
+      "step": 7618
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7828890037440287e-05,
+      "loss": 1.0564,
+      "step": 7619
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7827774776459197e-05,
+      "loss": 1.1774,
+      "step": 7620
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.782665940570122e-05,
+      "loss": 1.0617,
+      "step": 7621
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7825543925187364e-05,
+      "loss": 1.0243,
+      "step": 7622
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7824428334938642e-05,
+      "loss": 1.0721,
+      "step": 7623
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.782331263497607e-05,
+      "loss": 1.1403,
+      "step": 7624
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7822196825320656e-05,
+      "loss": 1.1262,
+      "step": 7625
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7821080905993427e-05,
+      "loss": 1.1447,
+      "step": 7626
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.78199648770154e-05,
+      "loss": 1.0925,
+      "step": 7627
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.78188487384076e-05,
+      "loss": 1.1472,
+      "step": 7628
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.781773249019105e-05,
+      "loss": 1.0841,
+      "step": 7629
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7816616132386772e-05,
+      "loss": 1.0964,
+      "step": 7630
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.78154996650158e-05,
+      "loss": 1.1186,
+      "step": 7631
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7814383088099165e-05,
+      "loss": 1.0705,
+      "step": 7632
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.78132664016579e-05,
+      "loss": 1.1305,
+      "step": 7633
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.781214960571303e-05,
+      "loss": 1.0955,
+      "step": 7634
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7811032700285607e-05,
+      "loss": 1.142,
+      "step": 7635
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.780991568539666e-05,
+      "loss": 1.0937,
+      "step": 7636
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7808798561067234e-05,
+      "loss": 1.18,
+      "step": 7637
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7807681327318373e-05,
+      "loss": 1.1416,
+      "step": 7638
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7806563984171117e-05,
+      "loss": 1.1076,
+      "step": 7639
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7805446531646518e-05,
+      "loss": 1.1256,
+      "step": 7640
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.780432896976562e-05,
+      "loss": 1.1767,
+      "step": 7641
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7803211298549484e-05,
+      "loss": 1.0412,
+      "step": 7642
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7802093518019158e-05,
+      "loss": 1.1196,
+      "step": 7643
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7800975628195693e-05,
+      "loss": 1.1248,
+      "step": 7644
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7799857629100153e-05,
+      "loss": 1.1141,
+      "step": 7645
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7798739520753592e-05,
+      "loss": 1.0226,
+      "step": 7646
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7797621303177078e-05,
+      "loss": 1.1271,
+      "step": 7647
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.779650297639167e-05,
+      "loss": 1.1067,
+      "step": 7648
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7795384540418434e-05,
+      "loss": 1.0888,
+      "step": 7649
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.779426599527844e-05,
+      "loss": 1.0542,
+      "step": 7650
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7793147340992757e-05,
+      "loss": 1.1511,
+      "step": 7651
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7792028577582454e-05,
+      "loss": 1.0793,
+      "step": 7652
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7790909705068613e-05,
+      "loss": 1.1353,
+      "step": 7653
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.77897907234723e-05,
+      "loss": 1.1139,
+      "step": 7654
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.77886716328146e-05,
+      "loss": 1.1238,
+      "step": 7655
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7787552433116587e-05,
+      "loss": 1.2326,
+      "step": 7656
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.778643312439935e-05,
+      "loss": 1.1124,
+      "step": 7657
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.778531370668397e-05,
+      "loss": 1.1463,
+      "step": 7658
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7784194179991535e-05,
+      "loss": 1.0708,
+      "step": 7659
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7783074544343133e-05,
+      "loss": 1.1897,
+      "step": 7660
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.778195479975985e-05,
+      "loss": 1.1567,
+      "step": 7661
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7780834946262783e-05,
+      "loss": 1.1457,
+      "step": 7662
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7779714983873024e-05,
+      "loss": 1.0899,
+      "step": 7663
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7778594912611678e-05,
+      "loss": 1.1403,
+      "step": 7664
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.777747473249983e-05,
+      "loss": 1.1618,
+      "step": 7665
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.777635444355859e-05,
+      "loss": 1.071,
+      "step": 7666
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.777523404580905e-05,
+      "loss": 1.1997,
+      "step": 7667
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7774113539272334e-05,
+      "loss": 1.1531,
+      "step": 7668
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.777299292396953e-05,
+      "loss": 1.0563,
+      "step": 7669
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7771872199921756e-05,
+      "loss": 1.0952,
+      "step": 7670
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7770751367150123e-05,
+      "loss": 1.1296,
+      "step": 7671
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7769630425675742e-05,
+      "loss": 1.0709,
+      "step": 7672
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7768509375519727e-05,
+      "loss": 1.1434,
+      "step": 7673
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7767388216703197e-05,
+      "loss": 1.1822,
+      "step": 7674
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.776626694924727e-05,
+      "loss": 1.1171,
+      "step": 7675
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7765145573173067e-05,
+      "loss": 1.1689,
+      "step": 7676
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7764024088501713e-05,
+      "loss": 1.1399,
+      "step": 7677
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.776290249525433e-05,
+      "loss": 1.0406,
+      "step": 7678
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.776178079345205e-05,
+      "loss": 1.2012,
+      "step": 7679
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7760658983115993e-05,
+      "loss": 1.1376,
+      "step": 7680
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7759537064267303e-05,
+      "loss": 1.0979,
+      "step": 7681
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7758415036927106e-05,
+      "loss": 1.1565,
+      "step": 7682
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.775729290111654e-05,
+      "loss": 1.056,
+      "step": 7683
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.775617065685674e-05,
+      "loss": 1.0949,
+      "step": 7684
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7755048304168843e-05,
+      "loss": 1.0985,
+      "step": 7685
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7753925843073998e-05,
+      "loss": 1.1884,
+      "step": 7686
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7752803273593345e-05,
+      "loss": 1.0476,
+      "step": 7687
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.775168059574803e-05,
+      "loss": 1.1027,
+      "step": 7688
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7750557809559198e-05,
+      "loss": 1.1349,
+      "step": 7689
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7749434915048005e-05,
+      "loss": 1.0277,
+      "step": 7690
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7748311912235596e-05,
+      "loss": 1.154,
+      "step": 7691
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7747188801143127e-05,
+      "loss": 1.1249,
+      "step": 7692
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7746065581791756e-05,
+      "loss": 1.1333,
+      "step": 7693
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.774494225420264e-05,
+      "loss": 1.2034,
+      "step": 7694
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.774381881839694e-05,
+      "loss": 1.1989,
+      "step": 7695
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.774269527439581e-05,
+      "loss": 1.1531,
+      "step": 7696
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7741571622220425e-05,
+      "loss": 1.1291,
+      "step": 7697
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7740447861891947e-05,
+      "loss": 1.101,
+      "step": 7698
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.773932399343154e-05,
+      "loss": 1.1248,
+      "step": 7699
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7738200016860383e-05,
+      "loss": 1.1589,
+      "step": 7700
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.773707593219964e-05,
+      "loss": 1.1179,
+      "step": 7701
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7735951739470487e-05,
+      "loss": 1.1915,
+      "step": 7702
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.77348274386941e-05,
+      "loss": 1.171,
+      "step": 7703
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.773370302989166e-05,
+      "loss": 1.163,
+      "step": 7704
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.773257851308435e-05,
+      "loss": 1.1629,
+      "step": 7705
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7731453888293344e-05,
+      "loss": 1.0938,
+      "step": 7706
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7730329155539835e-05,
+      "loss": 1.0591,
+      "step": 7707
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7729204314845003e-05,
+      "loss": 1.0924,
+      "step": 7708
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7728079366230036e-05,
+      "loss": 1.1789,
+      "step": 7709
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.772695430971613e-05,
+      "loss": 1.0937,
+      "step": 7710
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7725829145324472e-05,
+      "loss": 1.1067,
+      "step": 7711
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.772470387307626e-05,
+      "loss": 1.1291,
+      "step": 7712
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7723578492992688e-05,
+      "loss": 1.1857,
+      "step": 7713
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.772245300509496e-05,
+      "loss": 1.0759,
+      "step": 7714
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.772132740940427e-05,
+      "loss": 1.1021,
+      "step": 7715
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.772020170594183e-05,
+      "loss": 1.1129,
+      "step": 7716
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.771907589472883e-05,
+      "loss": 1.1021,
+      "step": 7717
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.771794997578649e-05,
+      "loss": 1.1658,
+      "step": 7718
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7716823949136012e-05,
+      "loss": 1.0934,
+      "step": 7719
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.771569781479861e-05,
+      "loss": 1.1216,
+      "step": 7720
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7714571572795492e-05,
+      "loss": 1.0995,
+      "step": 7721
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7713445223147877e-05,
+      "loss": 1.145,
+      "step": 7722
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.771231876587698e-05,
+      "loss": 1.0861,
+      "step": 7723
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7711192201004024e-05,
+      "loss": 1.0447,
+      "step": 7724
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.771006552855023e-05,
+      "loss": 1.1227,
+      "step": 7725
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7708938748536812e-05,
+      "loss": 1.0869,
+      "step": 7726
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7707811860985003e-05,
+      "loss": 1.0617,
+      "step": 7727
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7706684865916026e-05,
+      "loss": 1.1301,
+      "step": 7728
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7705557763351118e-05,
+      "loss": 1.1348,
+      "step": 7729
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7704430553311498e-05,
+      "loss": 1.0645,
+      "step": 7730
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7703303235818406e-05,
+      "loss": 1.0968,
+      "step": 7731
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7702175810893077e-05,
+      "loss": 1.1239,
+      "step": 7732
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7701048278556746e-05,
+      "loss": 1.0614,
+      "step": 7733
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7699920638830658e-05,
+      "loss": 1.0514,
+      "step": 7734
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7698792891736045e-05,
+      "loss": 1.1157,
+      "step": 7735
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.769766503729416e-05,
+      "loss": 1.0955,
+      "step": 7736
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7696537075526235e-05,
+      "loss": 1.0716,
+      "step": 7737
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.769540900645353e-05,
+      "loss": 1.0481,
+      "step": 7738
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.769428083009729e-05,
+      "loss": 1.1739,
+      "step": 7739
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7693152546478763e-05,
+      "loss": 1.0785,
+      "step": 7740
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.769202415561921e-05,
+      "loss": 1.0791,
+      "step": 7741
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7690895657539873e-05,
+      "loss": 1.0536,
+      "step": 7742
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.768976705226203e-05,
+      "loss": 0.9818,
+      "step": 7743
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7688638339806917e-05,
+      "loss": 1.2227,
+      "step": 7744
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.768750952019581e-05,
+      "loss": 1.2006,
+      "step": 7745
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7686380593449968e-05,
+      "loss": 1.0921,
+      "step": 7746
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.768525155959066e-05,
+      "loss": 1.0652,
+      "step": 7747
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7684122418639152e-05,
+      "loss": 1.1454,
+      "step": 7748
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.768299317061671e-05,
+      "loss": 1.0101,
+      "step": 7749
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7681863815544606e-05,
+      "loss": 1.1481,
+      "step": 7750
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7680734353444116e-05,
+      "loss": 1.0386,
+      "step": 7751
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7679604784336517e-05,
+      "loss": 0.9968,
+      "step": 7752
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.767847510824308e-05,
+      "loss": 1.047,
+      "step": 7753
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7677345325185094e-05,
+      "loss": 1.1386,
+      "step": 7754
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.767621543518383e-05,
+      "loss": 1.1609,
+      "step": 7755
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7675085438260585e-05,
+      "loss": 1.0953,
+      "step": 7756
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7673955334436628e-05,
+      "loss": 1.179,
+      "step": 7757
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7672825123733258e-05,
+      "loss": 1.1167,
+      "step": 7758
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7671694806171762e-05,
+      "loss": 1.0784,
+      "step": 7759
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.767056438177343e-05,
+      "loss": 1.0933,
+      "step": 7760
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.766943385055956e-05,
+      "loss": 1.0554,
+      "step": 7761
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.766830321255144e-05,
+      "loss": 1.1273,
+      "step": 7762
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7667172467770373e-05,
+      "loss": 1.1464,
+      "step": 7763
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.766604161623766e-05,
+      "loss": 1.1664,
+      "step": 7764
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7664910657974602e-05,
+      "loss": 1.1825,
+      "step": 7765
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.76637795930025e-05,
+      "loss": 1.1548,
+      "step": 7766
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7662648421342656e-05,
+      "loss": 1.0872,
+      "step": 7767
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7661517143016387e-05,
+      "loss": 1.075,
+      "step": 7768
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7660385758044998e-05,
+      "loss": 1.1728,
+      "step": 7769
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7659254266449802e-05,
+      "loss": 1.0599,
+      "step": 7770
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.765812266825211e-05,
+      "loss": 1.1144,
+      "step": 7771
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.765699096347324e-05,
+      "loss": 1.1836,
+      "step": 7772
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.765585915213451e-05,
+      "loss": 1.0649,
+      "step": 7773
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7654727234257237e-05,
+      "loss": 1.1356,
+      "step": 7774
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7653595209862746e-05,
+      "loss": 1.081,
+      "step": 7775
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.765246307897236e-05,
+      "loss": 1.1334,
+      "step": 7776
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7651330841607403e-05,
+      "loss": 1.1597,
+      "step": 7777
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7650198497789202e-05,
+      "loss": 1.1547,
+      "step": 7778
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7649066047539094e-05,
+      "loss": 1.0938,
+      "step": 7779
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7647933490878403e-05,
+      "loss": 1.0336,
+      "step": 7780
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7646800827828466e-05,
+      "loss": 1.169,
+      "step": 7781
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7645668058410618e-05,
+      "loss": 1.0633,
+      "step": 7782
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7644535182646198e-05,
+      "loss": 1.1403,
+      "step": 7783
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7643402200556545e-05,
+      "loss": 1.1249,
+      "step": 7784
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7642269112163e-05,
+      "loss": 1.1067,
+      "step": 7785
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7641135917486907e-05,
+      "loss": 1.0975,
+      "step": 7786
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.764000261654961e-05,
+      "loss": 1.0561,
+      "step": 7787
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.763886920937246e-05,
+      "loss": 1.1077,
+      "step": 7788
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7637735695976806e-05,
+      "loss": 1.0825,
+      "step": 7789
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7636602076384007e-05,
+      "loss": 1.2552,
+      "step": 7790
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7635468350615397e-05,
+      "loss": 1.1173,
+      "step": 7791
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7634334518692356e-05,
+      "loss": 1.1179,
+      "step": 7792
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7633200580636223e-05,
+      "loss": 1.2068,
+      "step": 7793
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.763206653646837e-05,
+      "loss": 1.1205,
+      "step": 7794
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7630932386210146e-05,
+      "loss": 1.1582,
+      "step": 7795
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7629798129882927e-05,
+      "loss": 1.0737,
+      "step": 7796
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7628663767508078e-05,
+      "loss": 1.082,
+      "step": 7797
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.762752929910696e-05,
+      "loss": 1.1544,
+      "step": 7798
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7626394724700944e-05,
+      "loss": 1.0427,
+      "step": 7799
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7625260044311407e-05,
+      "loss": 1.1168,
+      "step": 7800
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.762412525795972e-05,
+      "loss": 1.1197,
+      "step": 7801
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7622990365667257e-05,
+      "loss": 1.1396,
+      "step": 7802
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7621855367455396e-05,
+      "loss": 1.0881,
+      "step": 7803
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.762072026334552e-05,
+      "loss": 1.0504,
+      "step": 7804
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.761958505335901e-05,
+      "loss": 1.2452,
+      "step": 7805
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7618449737517243e-05,
+      "loss": 1.1692,
+      "step": 7806
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7617314315841615e-05,
+      "loss": 1.1194,
+      "step": 7807
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.761617878835351e-05,
+      "loss": 1.186,
+      "step": 7808
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7615043155074312e-05,
+      "loss": 1.1858,
+      "step": 7809
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7613907416025423e-05,
+      "loss": 1.249,
+      "step": 7810
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7612771571228224e-05,
+      "loss": 1.1026,
+      "step": 7811
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.761163562070413e-05,
+      "loss": 1.0793,
+      "step": 7812
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7610499564474514e-05,
+      "loss": 1.1936,
+      "step": 7813
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7609363402560797e-05,
+      "loss": 1.1496,
+      "step": 7814
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.760822713498437e-05,
+      "loss": 1.1521,
+      "step": 7815
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7607090761766635e-05,
+      "loss": 1.1679,
+      "step": 7816
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7605954282929006e-05,
+      "loss": 1.1534,
+      "step": 7817
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7604817698492887e-05,
+      "loss": 1.2027,
+      "step": 7818
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7603681008479684e-05,
+      "loss": 1.1008,
+      "step": 7819
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7602544212910815e-05,
+      "loss": 1.1548,
+      "step": 7820
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7601407311807688e-05,
+      "loss": 1.0912,
+      "step": 7821
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7600270305191717e-05,
+      "loss": 1.1542,
+      "step": 7822
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.759913319308433e-05,
+      "loss": 1.0459,
+      "step": 7823
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7597995975506937e-05,
+      "loss": 1.1613,
+      "step": 7824
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7596858652480964e-05,
+      "loss": 1.1603,
+      "step": 7825
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.759572122402783e-05,
+      "loss": 1.1315,
+      "step": 7826
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7594583690168967e-05,
+      "loss": 1.0545,
+      "step": 7827
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.75934460509258e-05,
+      "loss": 1.1775,
+      "step": 7828
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7592308306319754e-05,
+      "loss": 1.0859,
+      "step": 7829
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7591170456372265e-05,
+      "loss": 1.0976,
+      "step": 7830
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.759003250110477e-05,
+      "loss": 1.146,
+      "step": 7831
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7588894440538696e-05,
+      "loss": 1.0375,
+      "step": 7832
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.758775627469549e-05,
+      "loss": 1.2536,
+      "step": 7833
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.758661800359658e-05,
+      "loss": 1.1406,
+      "step": 7834
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7585479627263417e-05,
+      "loss": 1.1196,
+      "step": 7835
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7584341145717442e-05,
+      "loss": 1.0927,
+      "step": 7836
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7583202558980098e-05,
+      "loss": 1.1551,
+      "step": 7837
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7582063867072836e-05,
+      "loss": 1.1047,
+      "step": 7838
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.75809250700171e-05,
+      "loss": 1.087,
+      "step": 7839
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7579786167834346e-05,
+      "loss": 1.1892,
+      "step": 7840
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.757864716054603e-05,
+      "loss": 1.1,
+      "step": 7841
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7577508048173597e-05,
+      "loss": 1.1158,
+      "step": 7842
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7576368830738516e-05,
+      "loss": 1.0754,
+      "step": 7843
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.757522950826224e-05,
+      "loss": 1.0945,
+      "step": 7844
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.757409008076623e-05,
+      "loss": 1.0524,
+      "step": 7845
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7572950548271947e-05,
+      "loss": 1.1116,
+      "step": 7846
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7571810910800865e-05,
+      "loss": 1.0769,
+      "step": 7847
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.757067116837444e-05,
+      "loss": 1.088,
+      "step": 7848
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.756953132101415e-05,
+      "loss": 1.1676,
+      "step": 7849
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.756839136874146e-05,
+      "loss": 1.1058,
+      "step": 7850
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.756725131157785e-05,
+      "loss": 1.1034,
+      "step": 7851
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7566111149544793e-05,
+      "loss": 1.0783,
+      "step": 7852
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.756497088266376e-05,
+      "loss": 1.1169,
+      "step": 7853
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7563830510956235e-05,
+      "loss": 1.1287,
+      "step": 7854
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7562690034443697e-05,
+      "loss": 1.1637,
+      "step": 7855
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7561549453147633e-05,
+      "loss": 1.0891,
+      "step": 7856
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.756040876708952e-05,
+      "loss": 1.1485,
+      "step": 7857
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7559267976290855e-05,
+      "loss": 1.1464,
+      "step": 7858
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.755812708077312e-05,
+      "loss": 1.1115,
+      "step": 7859
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.755698608055781e-05,
+      "loss": 1.1577,
+      "step": 7860
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7555844975666416e-05,
+      "loss": 1.1127,
+      "step": 7861
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.755470376612043e-05,
+      "loss": 1.0926,
+      "step": 7862
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7553562451941355e-05,
+      "loss": 1.1328,
+      "step": 7863
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.755242103315068e-05,
+      "loss": 1.1877,
+      "step": 7864
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.755127950976992e-05,
+      "loss": 1.1691,
+      "step": 7865
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.755013788182056e-05,
+      "loss": 1.1335,
+      "step": 7866
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7548996149324123e-05,
+      "loss": 1.1326,
+      "step": 7867
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7547854312302097e-05,
+      "loss": 1.0893,
+      "step": 7868
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7546712370776007e-05,
+      "loss": 1.1135,
+      "step": 7869
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7545570324767358e-05,
+      "loss": 1.1655,
+      "step": 7870
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7544428174297657e-05,
+      "loss": 1.0948,
+      "step": 7871
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7543285919388427e-05,
+      "loss": 0.9936,
+      "step": 7872
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.754214356006118e-05,
+      "loss": 1.0848,
+      "step": 7873
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7541001096337434e-05,
+      "loss": 1.0505,
+      "step": 7874
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7539858528238707e-05,
+      "loss": 1.1206,
+      "step": 7875
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7538715855786528e-05,
+      "loss": 1.0783,
+      "step": 7876
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.753757307900242e-05,
+      "loss": 1.1447,
+      "step": 7877
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7536430197907905e-05,
+      "loss": 1.1544,
+      "step": 7878
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7535287212524513e-05,
+      "loss": 1.0745,
+      "step": 7879
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.753414412287378e-05,
+      "loss": 1.1393,
+      "step": 7880
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7533000928977225e-05,
+      "loss": 1.1044,
+      "step": 7881
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7531857630856397e-05,
+      "loss": 1.12,
+      "step": 7882
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.753071422853282e-05,
+      "loss": 1.1364,
+      "step": 7883
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7529570722028042e-05,
+      "loss": 1.1096,
+      "step": 7884
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.75284271113636e-05,
+      "loss": 1.0778,
+      "step": 7885
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7527283396561027e-05,
+      "loss": 1.0193,
+      "step": 7886
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7526139577641882e-05,
+      "loss": 1.1322,
+      "step": 7887
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7524995654627698e-05,
+      "loss": 1.069,
+      "step": 7888
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7523851627540032e-05,
+      "loss": 1.1071,
+      "step": 7889
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7522707496400426e-05,
+      "loss": 1.1338,
+      "step": 7890
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7521563261230443e-05,
+      "loss": 1.0656,
+      "step": 7891
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7520418922051625e-05,
+      "loss": 1.1228,
+      "step": 7892
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7519274478885536e-05,
+      "loss": 1.1837,
+      "step": 7893
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7518129931753725e-05,
+      "loss": 1.1012,
+      "step": 7894
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.751698528067776e-05,
+      "loss": 1.1519,
+      "step": 7895
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.75158405256792e-05,
+      "loss": 1.1717,
+      "step": 7896
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.751469566677961e-05,
+      "loss": 1.1156,
+      "step": 7897
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.751355070400055e-05,
+      "loss": 1.1107,
+      "step": 7898
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7512405637363587e-05,
+      "loss": 1.0339,
+      "step": 7899
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7511260466890304e-05,
+      "loss": 1.1473,
+      "step": 7900
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7510115192602256e-05,
+      "loss": 1.1306,
+      "step": 7901
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7508969814521027e-05,
+      "loss": 1.0808,
+      "step": 7902
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7507824332668186e-05,
+      "loss": 1.0634,
+      "step": 7903
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7506678747065315e-05,
+      "loss": 1.1197,
+      "step": 7904
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7505533057733993e-05,
+      "loss": 1.1377,
+      "step": 7905
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7504387264695796e-05,
+      "loss": 1.1531,
+      "step": 7906
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7503241367972313e-05,
+      "loss": 1.1271,
+      "step": 7907
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7502095367585126e-05,
+      "loss": 1.1556,
+      "step": 7908
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.750094926355582e-05,
+      "loss": 1.1287,
+      "step": 7909
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.749980305590599e-05,
+      "loss": 1.0836,
+      "step": 7910
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7498656744657224e-05,
+      "loss": 1.0268,
+      "step": 7911
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7497510329831106e-05,
+      "loss": 1.0554,
+      "step": 7912
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.749636381144925e-05,
+      "loss": 1.007,
+      "step": 7913
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7495217189533232e-05,
+      "loss": 1.077,
+      "step": 7914
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7494070464104667e-05,
+      "loss": 1.1215,
+      "step": 7915
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.749292363518515e-05,
+      "loss": 1.0756,
+      "step": 7916
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.749177670279628e-05,
+      "loss": 1.1032,
+      "step": 7917
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.749062966695967e-05,
+      "loss": 1.1123,
+      "step": 7918
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7489482527696915e-05,
+      "loss": 1.0767,
+      "step": 7919
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.748833528502963e-05,
+      "loss": 1.1453,
+      "step": 7920
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7487187938979426e-05,
+      "loss": 1.0686,
+      "step": 7921
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7486040489567918e-05,
+      "loss": 1.0831,
+      "step": 7922
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7484892936816712e-05,
+      "loss": 1.1062,
+      "step": 7923
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7483745280747434e-05,
+      "loss": 1.0739,
+      "step": 7924
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.748259752138169e-05,
+      "loss": 1.1577,
+      "step": 7925
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7481449658741114e-05,
+      "loss": 1.0945,
+      "step": 7926
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7480301692847322e-05,
+      "loss": 1.0601,
+      "step": 7927
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.747915362372193e-05,
+      "loss": 1.1021,
+      "step": 7928
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.747800545138658e-05,
+      "loss": 1.0114,
+      "step": 7929
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7476857175862882e-05,
+      "loss": 1.0956,
+      "step": 7930
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7475708797172486e-05,
+      "loss": 1.1215,
+      "step": 7931
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7474560315337008e-05,
+      "loss": 1.1245,
+      "step": 7932
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7473411730378087e-05,
+      "loss": 1.0354,
+      "step": 7933
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7472263042317357e-05,
+      "loss": 1.1065,
+      "step": 7934
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7471114251176465e-05,
+      "loss": 1.0878,
+      "step": 7935
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7469965356977036e-05,
+      "loss": 1.0841,
+      "step": 7936
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7468816359740722e-05,
+      "loss": 1.026,
+      "step": 7937
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.746766725948916e-05,
+      "loss": 1.1094,
+      "step": 7938
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7466518056244e-05,
+      "loss": 1.1575,
+      "step": 7939
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7465368750026888e-05,
+      "loss": 1.1038,
+      "step": 7940
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7464219340859475e-05,
+      "loss": 1.0507,
+      "step": 7941
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7463069828763406e-05,
+      "loss": 1.1134,
+      "step": 7942
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7461920213760343e-05,
+      "loss": 1.1221,
+      "step": 7943
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.746077049587193e-05,
+      "loss": 1.0932,
+      "step": 7944
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7459620675119838e-05,
+      "loss": 1.091,
+      "step": 7945
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7458470751525717e-05,
+      "loss": 1.1226,
+      "step": 7946
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.745732072511123e-05,
+      "loss": 1.0775,
+      "step": 7947
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7456170595898034e-05,
+      "loss": 1.0951,
+      "step": 7948
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7455020363907807e-05,
+      "loss": 1.152,
+      "step": 7949
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7453870029162203e-05,
+      "loss": 1.1166,
+      "step": 7950
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7452719591682897e-05,
+      "loss": 1.153,
+      "step": 7951
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.745156905149156e-05,
+      "loss": 1.234,
+      "step": 7952
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.745041840860986e-05,
+      "loss": 1.1091,
+      "step": 7953
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7449267663059477e-05,
+      "loss": 1.1009,
+      "step": 7954
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7448116814862084e-05,
+      "loss": 1.0504,
+      "step": 7955
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7446965864039358e-05,
+      "loss": 1.1619,
+      "step": 7956
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7445814810612987e-05,
+      "loss": 1.0683,
+      "step": 7957
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7444663654604644e-05,
+      "loss": 1.0838,
+      "step": 7958
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7443512396036014e-05,
+      "loss": 1.15,
+      "step": 7959
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.744236103492879e-05,
+      "loss": 1.0268,
+      "step": 7960
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7441209571304654e-05,
+      "loss": 1.0756,
+      "step": 7961
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7440058005185296e-05,
+      "loss": 1.0345,
+      "step": 7962
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7438906336592414e-05,
+      "loss": 1.0343,
+      "step": 7963
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7437754565547695e-05,
+      "loss": 1.0695,
+      "step": 7964
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7436602692072837e-05,
+      "loss": 1.0844,
+      "step": 7965
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.743545071618954e-05,
+      "loss": 1.1394,
+      "step": 7966
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.74342986379195e-05,
+      "loss": 1.0546,
+      "step": 7967
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.743314645728442e-05,
+      "loss": 1.0579,
+      "step": 7968
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7431994174306006e-05,
+      "loss": 1.0287,
+      "step": 7969
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7430841789005957e-05,
+      "loss": 1.0228,
+      "step": 7970
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7429689301405988e-05,
+      "loss": 1.1143,
+      "step": 7971
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.74285367115278e-05,
+      "loss": 0.9606,
+      "step": 7972
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7427384019393112e-05,
+      "loss": 1.1424,
+      "step": 7973
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.742623122502363e-05,
+      "loss": 1.0938,
+      "step": 7974
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7425078328441078e-05,
+      "loss": 1.0926,
+      "step": 7975
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7423925329667164e-05,
+      "loss": 1.1288,
+      "step": 7976
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.742277222872361e-05,
+      "loss": 1.0373,
+      "step": 7977
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.742161902563214e-05,
+      "loss": 1.142,
+      "step": 7978
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.742046572041447e-05,
+      "loss": 1.0966,
+      "step": 7979
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.741931231309233e-05,
+      "loss": 1.0858,
+      "step": 7980
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7418158803687446e-05,
+      "loss": 1.0614,
+      "step": 7981
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7417005192221544e-05,
+      "loss": 1.1271,
+      "step": 7982
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7415851478716357e-05,
+      "loss": 1.0709,
+      "step": 7983
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7414697663193617e-05,
+      "loss": 1.0875,
+      "step": 7984
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7413543745675056e-05,
+      "loss": 1.0399,
+      "step": 7985
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7412389726182413e-05,
+      "loss": 1.1443,
+      "step": 7986
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7411235604737427e-05,
+      "loss": 1.0037,
+      "step": 7987
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7410081381361832e-05,
+      "loss": 0.9542,
+      "step": 7988
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7408927056077375e-05,
+      "loss": 1.1189,
+      "step": 7989
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7407772628905795e-05,
+      "loss": 1.1651,
+      "step": 7990
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.740661809986885e-05,
+      "loss": 1.0857,
+      "step": 7991
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7405463468988272e-05,
+      "loss": 1.0915,
+      "step": 7992
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.740430873628582e-05,
+      "loss": 1.1363,
+      "step": 7993
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7403153901783244e-05,
+      "loss": 1.0534,
+      "step": 7994
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7401998965502294e-05,
+      "loss": 1.082,
+      "step": 7995
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7400843927464728e-05,
+      "loss": 1.0673,
+      "step": 7996
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7399688787692302e-05,
+      "loss": 1.165,
+      "step": 7997
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.739853354620678e-05,
+      "loss": 1.1537,
+      "step": 7998
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.739737820302992e-05,
+      "loss": 1.106,
+      "step": 7999
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.739622275818348e-05,
+      "loss": 1.0485,
+      "step": 8000
+    },
+    {
+      "epoch": 0.34,
+      "eval_text_loss": 0.8169276118278503,
+      "eval_text_runtime": 1.3955,
+      "eval_text_samples_per_second": 53.027,
+      "eval_text_steps_per_second": 1.433,
+      "step": 8000
+    },
+    {
+      "epoch": 0.34,
+      "eval_svit_loss": 1.176889419555664,
+      "eval_svit_runtime": 5.0631,
+      "eval_svit_samples_per_second": 134.503,
+      "eval_svit_steps_per_second": 2.173,
+      "step": 8000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7395067211689233e-05,
+      "loss": 1.1725,
+      "step": 8001
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.739391156356894e-05,
+      "loss": 1.0697,
+      "step": 8002
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.739275581384437e-05,
+      "loss": 1.0305,
+      "step": 8003
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7391599962537303e-05,
+      "loss": 1.0723,
+      "step": 8004
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.73904440096695e-05,
+      "loss": 1.0977,
+      "step": 8005
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7389287955262742e-05,
+      "loss": 1.126,
+      "step": 8006
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7388131799338797e-05,
+      "loss": 1.1297,
+      "step": 8007
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.738697554191946e-05,
+      "loss": 1.1199,
+      "step": 8008
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.738581918302649e-05,
+      "loss": 1.0924,
+      "step": 8009
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.738466272268169e-05,
+      "loss": 1.1646,
+      "step": 8010
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7383506160906828e-05,
+      "loss": 1.1026,
+      "step": 8011
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7382349497723694e-05,
+      "loss": 1.0671,
+      "step": 8012
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7381192733154084e-05,
+      "loss": 1.0924,
+      "step": 8013
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7380035867219777e-05,
+      "loss": 1.0502,
+      "step": 8014
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7378878899942573e-05,
+      "loss": 1.1476,
+      "step": 8015
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.737772183134426e-05,
+      "loss": 1.0979,
+      "step": 8016
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7376564661446633e-05,
+      "loss": 1.0602,
+      "step": 8017
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7375407390271493e-05,
+      "loss": 0.9534,
+      "step": 8018
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7374250017840642e-05,
+      "loss": 1.1054,
+      "step": 8019
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7373092544175876e-05,
+      "loss": 1.228,
+      "step": 8020
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7371934969298997e-05,
+      "loss": 1.0976,
+      "step": 8021
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7370777293231815e-05,
+      "loss": 1.0675,
+      "step": 8022
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7369619515996133e-05,
+      "loss": 1.0712,
+      "step": 8023
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7368461637613763e-05,
+      "loss": 1.1647,
+      "step": 8024
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7367303658106518e-05,
+      "loss": 1.0891,
+      "step": 8025
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.73661455774962e-05,
+      "loss": 1.0794,
+      "step": 8026
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7364987395804637e-05,
+      "loss": 1.1067,
+      "step": 8027
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7363829113053634e-05,
+      "loss": 1.091,
+      "step": 8028
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7362670729265017e-05,
+      "loss": 1.1714,
+      "step": 8029
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7361512244460606e-05,
+      "loss": 1.078,
+      "step": 8030
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7360353658662217e-05,
+      "loss": 1.1096,
+      "step": 8031
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.735919497189168e-05,
+      "loss": 1.1327,
+      "step": 8032
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7358036184170818e-05,
+      "loss": 1.0397,
+      "step": 8033
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.735687729552146e-05,
+      "loss": 1.1299,
+      "step": 8034
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7355718305965437e-05,
+      "loss": 1.2041,
+      "step": 8035
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.735455921552458e-05,
+      "loss": 1.0443,
+      "step": 8036
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.735340002422072e-05,
+      "loss": 1.1254,
+      "step": 8037
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7352240732075695e-05,
+      "loss": 1.2008,
+      "step": 8038
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7351081339111347e-05,
+      "loss": 1.1469,
+      "step": 8039
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7349921845349505e-05,
+      "loss": 1.0544,
+      "step": 8040
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7348762250812017e-05,
+      "loss": 1.0914,
+      "step": 8041
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7347602555520726e-05,
+      "loss": 1.0634,
+      "step": 8042
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.734644275949747e-05,
+      "loss": 1.0907,
+      "step": 8043
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.734528286276411e-05,
+      "loss": 1.1198,
+      "step": 8044
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7344122865342478e-05,
+      "loss": 0.9819,
+      "step": 8045
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.734296276725444e-05,
+      "loss": 1.095,
+      "step": 8046
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7341802568521836e-05,
+      "loss": 1.0626,
+      "step": 8047
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.734064226916653e-05,
+      "loss": 1.1325,
+      "step": 8048
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.733948186921037e-05,
+      "loss": 1.1447,
+      "step": 8049
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7338321368675228e-05,
+      "loss": 1.0201,
+      "step": 8050
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.733716076758295e-05,
+      "loss": 1.0757,
+      "step": 8051
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.73360000659554e-05,
+      "loss": 1.0976,
+      "step": 8052
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7334839263814448e-05,
+      "loss": 1.1548,
+      "step": 8053
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7333678361181957e-05,
+      "loss": 1.1045,
+      "step": 8054
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.733251735807979e-05,
+      "loss": 1.0475,
+      "step": 8055
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7331356254529827e-05,
+      "loss": 1.0975,
+      "step": 8056
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.733019505055393e-05,
+      "loss": 1.0648,
+      "step": 8057
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7329033746173976e-05,
+      "loss": 1.0531,
+      "step": 8058
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7327872341411843e-05,
+      "loss": 1.1554,
+      "step": 8059
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7326710836289407e-05,
+      "loss": 1.1061,
+      "step": 8060
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7325549230828543e-05,
+      "loss": 1.0823,
+      "step": 8061
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.732438752505113e-05,
+      "loss": 1.0905,
+      "step": 8062
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7323225718979064e-05,
+      "loss": 1.0549,
+      "step": 8063
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7322063812634214e-05,
+      "loss": 1.0545,
+      "step": 8064
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7320901806038482e-05,
+      "loss": 1.1117,
+      "step": 8065
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7319739699213744e-05,
+      "loss": 1.0646,
+      "step": 8066
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7318577492181895e-05,
+      "loss": 1.0845,
+      "step": 8067
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7317415184964825e-05,
+      "loss": 1.0924,
+      "step": 8068
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7316252777584436e-05,
+      "loss": 1.0367,
+      "step": 8069
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7315090270062614e-05,
+      "loss": 1.1126,
+      "step": 8070
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7313927662421263e-05,
+      "loss": 1.0972,
+      "step": 8071
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7312764954682283e-05,
+      "loss": 1.127,
+      "step": 8072
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7311602146867572e-05,
+      "loss": 1.0254,
+      "step": 8073
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7310439238999033e-05,
+      "loss": 1.103,
+      "step": 8074
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7309276231098582e-05,
+      "loss": 1.114,
+      "step": 8075
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.730811312318811e-05,
+      "loss": 1.0068,
+      "step": 8076
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.730694991528954e-05,
+      "loss": 0.9962,
+      "step": 8077
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.730578660742478e-05,
+      "loss": 1.1239,
+      "step": 8078
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7304623199615738e-05,
+      "loss": 1.1034,
+      "step": 8079
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7303459691884334e-05,
+      "loss": 1.1322,
+      "step": 8080
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.730229608425248e-05,
+      "loss": 1.0772,
+      "step": 8081
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7301132376742098e-05,
+      "loss": 1.1581,
+      "step": 8082
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.729996856937511e-05,
+      "loss": 1.1295,
+      "step": 8083
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.729880466217344e-05,
+      "loss": 1.1234,
+      "step": 8084
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7297640655159003e-05,
+      "loss": 1.1167,
+      "step": 8085
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7296476548353735e-05,
+      "loss": 1.108,
+      "step": 8086
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7295312341779556e-05,
+      "loss": 1.0244,
+      "step": 8087
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7294148035458407e-05,
+      "loss": 1.1721,
+      "step": 8088
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7292983629412208e-05,
+      "loss": 0.96,
+      "step": 8089
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.72918191236629e-05,
+      "loss": 1.0645,
+      "step": 8090
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7290654518232414e-05,
+      "loss": 1.0991,
+      "step": 8091
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7289489813142692e-05,
+      "loss": 1.0547,
+      "step": 8092
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.728832500841567e-05,
+      "loss": 1.0874,
+      "step": 8093
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.728716010407329e-05,
+      "loss": 1.2155,
+      "step": 8094
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.72859951001375e-05,
+      "loss": 1.0533,
+      "step": 8095
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7284829996630235e-05,
+      "loss": 1.0967,
+      "step": 8096
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7283664793573452e-05,
+      "loss": 1.0361,
+      "step": 8097
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7282499490989095e-05,
+      "loss": 1.0643,
+      "step": 8098
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7281334088899116e-05,
+      "loss": 1.0311,
+      "step": 8099
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7280168587325463e-05,
+      "loss": 1.0734,
+      "step": 8100
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7279002986290097e-05,
+      "loss": 1.084,
+      "step": 8101
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7277837285814965e-05,
+      "loss": 1.0332,
+      "step": 8102
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7276671485922037e-05,
+      "loss": 1.1063,
+      "step": 8103
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7275505586633265e-05,
+      "loss": 1.0739,
+      "step": 8104
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7274339587970616e-05,
+      "loss": 1.1758,
+      "step": 8105
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7273173489956046e-05,
+      "loss": 1.096,
+      "step": 8106
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7272007292611532e-05,
+      "loss": 1.0882,
+      "step": 8107
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.727084099595903e-05,
+      "loss": 1.0449,
+      "step": 8108
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7269674600020512e-05,
+      "loss": 1.1896,
+      "step": 8109
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7268508104817957e-05,
+      "loss": 1.034,
+      "step": 8110
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7267341510373328e-05,
+      "loss": 1.098,
+      "step": 8111
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.726617481670861e-05,
+      "loss": 1.1309,
+      "step": 8112
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.726500802384577e-05,
+      "loss": 1.0743,
+      "step": 8113
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.726384113180679e-05,
+      "loss": 1.0695,
+      "step": 8114
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7262674140613658e-05,
+      "loss": 1.0398,
+      "step": 8115
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7261507050288342e-05,
+      "loss": 1.0478,
+      "step": 8116
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7260339860852843e-05,
+      "loss": 1.0945,
+      "step": 8117
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7259172572329133e-05,
+      "loss": 1.0036,
+      "step": 8118
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7258005184739204e-05,
+      "loss": 1.0491,
+      "step": 8119
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7256837698105048e-05,
+      "loss": 1.0183,
+      "step": 8120
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7255670112448658e-05,
+      "loss": 1.1614,
+      "step": 8121
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7254502427792026e-05,
+      "loss": 1.0452,
+      "step": 8122
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7253334644157148e-05,
+      "loss": 1.0064,
+      "step": 8123
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.725216676156602e-05,
+      "loss": 1.0585,
+      "step": 8124
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.725099878004064e-05,
+      "loss": 1.0432,
+      "step": 8125
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7249830699603015e-05,
+      "loss": 1.0438,
+      "step": 8126
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7248662520275137e-05,
+      "loss": 1.0304,
+      "step": 8127
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7247494242079025e-05,
+      "loss": 1.0913,
+      "step": 8128
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7246325865036673e-05,
+      "loss": 1.0904,
+      "step": 8129
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.72451573891701e-05,
+      "loss": 1.0861,
+      "step": 8130
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7243988814501305e-05,
+      "loss": 1.0982,
+      "step": 8131
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7242820141052312e-05,
+      "loss": 1.0788,
+      "step": 8132
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.724165136884513e-05,
+      "loss": 1.1243,
+      "step": 8133
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.724048249790177e-05,
+      "loss": 1.0929,
+      "step": 8134
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.723931352824426e-05,
+      "loss": 1.1712,
+      "step": 8135
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.723814445989461e-05,
+      "loss": 1.113,
+      "step": 8136
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.723697529287485e-05,
+      "loss": 1.1557,
+      "step": 8137
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7235806027206998e-05,
+      "loss": 1.1294,
+      "step": 8138
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7234636662913075e-05,
+      "loss": 1.2237,
+      "step": 8139
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.723346720001512e-05,
+      "loss": 1.05,
+      "step": 8140
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7232297638535153e-05,
+      "loss": 1.1159,
+      "step": 8141
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7231127978495207e-05,
+      "loss": 1.0689,
+      "step": 8142
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7229958219917317e-05,
+      "loss": 1.0326,
+      "step": 8143
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7228788362823514e-05,
+      "loss": 1.0936,
+      "step": 8144
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.722761840723584e-05,
+      "loss": 1.1672,
+      "step": 8145
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7226448353176326e-05,
+      "loss": 1.1363,
+      "step": 8146
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7225278200667017e-05,
+      "loss": 1.0728,
+      "step": 8147
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7224107949729952e-05,
+      "loss": 1.0649,
+      "step": 8148
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7222937600387177e-05,
+      "loss": 1.133,
+      "step": 8149
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7221767152660736e-05,
+      "loss": 1.0648,
+      "step": 8150
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.722059660657268e-05,
+      "loss": 1.1106,
+      "step": 8151
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7219425962145055e-05,
+      "loss": 1.0728,
+      "step": 8152
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.721825521939991e-05,
+      "loss": 1.0621,
+      "step": 8153
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7217084378359305e-05,
+      "loss": 1.1382,
+      "step": 8154
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7215913439045292e-05,
+      "loss": 1.0911,
+      "step": 8155
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7214742401479927e-05,
+      "loss": 1.1089,
+      "step": 8156
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.721357126568527e-05,
+      "loss": 1.0931,
+      "step": 8157
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7212400031683374e-05,
+      "loss": 0.9724,
+      "step": 8158
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7211228699496314e-05,
+      "loss": 1.0981,
+      "step": 8159
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.721005726914614e-05,
+      "loss": 1.0758,
+      "step": 8160
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7208885740654936e-05,
+      "loss": 1.1009,
+      "step": 8161
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7207714114044753e-05,
+      "loss": 1.0922,
+      "step": 8162
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.720654238933767e-05,
+      "loss": 1.0401,
+      "step": 8163
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7205370566555755e-05,
+      "loss": 1.0904,
+      "step": 8164
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.720419864572108e-05,
+      "loss": 1.1646,
+      "step": 8165
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7203026626855727e-05,
+      "loss": 1.1555,
+      "step": 8166
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.720185450998177e-05,
+      "loss": 1.1182,
+      "step": 8167
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7200682295121285e-05,
+      "loss": 1.0852,
+      "step": 8168
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.719950998229635e-05,
+      "loss": 1.0481,
+      "step": 8169
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.719833757152906e-05,
+      "loss": 1.135,
+      "step": 8170
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.719716506284149e-05,
+      "loss": 1.0977,
+      "step": 8171
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7195992456255728e-05,
+      "loss": 1.1119,
+      "step": 8172
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7194819751793862e-05,
+      "loss": 1.1597,
+      "step": 8173
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7193646949477984e-05,
+      "loss": 1.1234,
+      "step": 8174
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.719247404933018e-05,
+      "loss": 1.1649,
+      "step": 8175
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7191301051372554e-05,
+      "loss": 1.078,
+      "step": 8176
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.719012795562719e-05,
+      "loss": 1.0576,
+      "step": 8177
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.71889547621162e-05,
+      "loss": 1.1271,
+      "step": 8178
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7187781470861668e-05,
+      "loss": 1.0744,
+      "step": 8179
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.71866080818857e-05,
+      "loss": 1.0811,
+      "step": 8180
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7185434595210404e-05,
+      "loss": 1.2134,
+      "step": 8181
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7184261010857882e-05,
+      "loss": 1.0561,
+      "step": 8182
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7183087328850238e-05,
+      "loss": 1.0687,
+      "step": 8183
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.718191354920958e-05,
+      "loss": 0.9876,
+      "step": 8184
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7180739671958027e-05,
+      "loss": 1.0447,
+      "step": 8185
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.717956569711768e-05,
+      "loss": 1.1054,
+      "step": 8186
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.717839162471066e-05,
+      "loss": 1.1119,
+      "step": 8187
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7177217454759083e-05,
+      "loss": 1.0897,
+      "step": 8188
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.717604318728506e-05,
+      "loss": 1.1144,
+      "step": 8189
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7174868822310715e-05,
+      "loss": 1.0303,
+      "step": 8190
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.717369435985817e-05,
+      "loss": 1.1086,
+      "step": 8191
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.717251979994955e-05,
+      "loss": 1.182,
+      "step": 8192
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7171345142606974e-05,
+      "loss": 1.0231,
+      "step": 8193
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7170170387852573e-05,
+      "loss": 1.0224,
+      "step": 8194
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7168995535708475e-05,
+      "loss": 1.0527,
+      "step": 8195
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.716782058619681e-05,
+      "loss": 1.1112,
+      "step": 8196
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7166645539339714e-05,
+      "loss": 1.1142,
+      "step": 8197
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7165470395159313e-05,
+      "loss": 1.0518,
+      "step": 8198
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7164295153677754e-05,
+      "loss": 1.0688,
+      "step": 8199
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.716311981491716e-05,
+      "loss": 1.0285,
+      "step": 8200
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.716194437889969e-05,
+      "loss": 1.1643,
+      "step": 8201
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7160768845647465e-05,
+      "loss": 1.1446,
+      "step": 8202
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7159593215182647e-05,
+      "loss": 1.0777,
+      "step": 8203
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7158417487527365e-05,
+      "loss": 1.0564,
+      "step": 8204
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7157241662703776e-05,
+      "loss": 1.1079,
+      "step": 8205
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7156065740734027e-05,
+      "loss": 1.0736,
+      "step": 8206
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7154889721640267e-05,
+      "loss": 1.0788,
+      "step": 8207
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.715371360544465e-05,
+      "loss": 1.0862,
+      "step": 8208
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.715253739216933e-05,
+      "loss": 1.1101,
+      "step": 8209
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.715136108183646e-05,
+      "loss": 1.0295,
+      "step": 8210
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7150184674468206e-05,
+      "loss": 1.1024,
+      "step": 8211
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7149008170086723e-05,
+      "loss": 1.0952,
+      "step": 8212
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7147831568714164e-05,
+      "loss": 1.0965,
+      "step": 8213
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.714665487037271e-05,
+      "loss": 1.1248,
+      "step": 8214
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.714547807508451e-05,
+      "loss": 1.114,
+      "step": 8215
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7144301182871744e-05,
+      "loss": 1.0952,
+      "step": 8216
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.714312419375657e-05,
+      "loss": 1.1121,
+      "step": 8217
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7141947107761166e-05,
+      "loss": 1.0165,
+      "step": 8218
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7140769924907705e-05,
+      "loss": 1.043,
+      "step": 8219
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7139592645218355e-05,
+      "loss": 1.0595,
+      "step": 8220
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.71384152687153e-05,
+      "loss": 1.1306,
+      "step": 8221
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7137237795420706e-05,
+      "loss": 1.1333,
+      "step": 8222
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.713606022535677e-05,
+      "loss": 0.9504,
+      "step": 8223
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.713488255854566e-05,
+      "loss": 1.073,
+      "step": 8224
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7133704795009565e-05,
+      "loss": 1.0108,
+      "step": 8225
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.713252693477067e-05,
+      "loss": 0.9925,
+      "step": 8226
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7131348977851163e-05,
+      "loss": 1.0531,
+      "step": 8227
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.713017092427323e-05,
+      "loss": 0.9895,
+      "step": 8228
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7128992774059063e-05,
+      "loss": 1.0342,
+      "step": 8229
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7127814527230858e-05,
+      "loss": 1.085,
+      "step": 8230
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7126636183810803e-05,
+      "loss": 1.0178,
+      "step": 8231
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7125457743821098e-05,
+      "loss": 1.0798,
+      "step": 8232
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7124279207283942e-05,
+      "loss": 1.0768,
+      "step": 8233
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7123100574221537e-05,
+      "loss": 1.1177,
+      "step": 8234
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.712192184465608e-05,
+      "loss": 0.9662,
+      "step": 8235
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7120743018609772e-05,
+      "loss": 1.1894,
+      "step": 8236
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.711956409610483e-05,
+      "loss": 1.0482,
+      "step": 8237
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7118385077163445e-05,
+      "loss": 1.0194,
+      "step": 8238
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7117205961807843e-05,
+      "loss": 1.0237,
+      "step": 8239
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7116026750060222e-05,
+      "loss": 1.0291,
+      "step": 8240
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.71148474419428e-05,
+      "loss": 1.0973,
+      "step": 8241
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.711366803747779e-05,
+      "loss": 1.1884,
+      "step": 8242
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7112488536687414e-05,
+      "loss": 1.068,
+      "step": 8243
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7111308939593877e-05,
+      "loss": 1.1439,
+      "step": 8244
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7110129246219413e-05,
+      "loss": 1.1201,
+      "step": 8245
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.710894945658623e-05,
+      "loss": 1.1031,
+      "step": 8246
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7107769570716567e-05,
+      "loss": 1.0899,
+      "step": 8247
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7106589588632638e-05,
+      "loss": 1.014,
+      "step": 8248
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7105409510356675e-05,
+      "loss": 1.095,
+      "step": 8249
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7104229335910902e-05,
+      "loss": 1.1045,
+      "step": 8250
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7103049065317556e-05,
+      "loss": 1.0785,
+      "step": 8251
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.710186869859887e-05,
+      "loss": 1.1386,
+      "step": 8252
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7100688235777067e-05,
+      "loss": 1.0965,
+      "step": 8253
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7099507676874396e-05,
+      "loss": 1.0842,
+      "step": 8254
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.709832702191309e-05,
+      "loss": 1.0242,
+      "step": 8255
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7097146270915392e-05,
+      "loss": 1.0351,
+      "step": 8256
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7095965423903535e-05,
+      "loss": 1.0704,
+      "step": 8257
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7094784480899775e-05,
+      "loss": 1.0373,
+      "step": 8258
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7093603441926345e-05,
+      "loss": 1.0957,
+      "step": 8259
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.70924223070055e-05,
+      "loss": 1.0329,
+      "step": 8260
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7091241076159482e-05,
+      "loss": 1.151,
+      "step": 8261
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7090059749410552e-05,
+      "loss": 1.0832,
+      "step": 8262
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7088878326780955e-05,
+      "loss": 1.1432,
+      "step": 8263
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7087696808292948e-05,
+      "loss": 1.0833,
+      "step": 8264
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7086515193968784e-05,
+      "loss": 1.1623,
+      "step": 8265
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7085333483830723e-05,
+      "loss": 1.0804,
+      "step": 8266
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.708415167790102e-05,
+      "loss": 1.0742,
+      "step": 8267
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.708296977620195e-05,
+      "loss": 0.9956,
+      "step": 8268
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.708178777875576e-05,
+      "loss": 1.1426,
+      "step": 8269
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7080605685584724e-05,
+      "loss": 1.0863,
+      "step": 8270
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.707942349671111e-05,
+      "loss": 1.089,
+      "step": 8271
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.707824121215718e-05,
+      "loss": 1.0707,
+      "step": 8272
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.707705883194521e-05,
+      "loss": 1.0386,
+      "step": 8273
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7075876356097474e-05,
+      "loss": 1.0671,
+      "step": 8274
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.707469378463624e-05,
+      "loss": 1.0647,
+      "step": 8275
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7073511117583787e-05,
+      "loss": 1.1375,
+      "step": 8276
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7072328354962393e-05,
+      "loss": 1.0694,
+      "step": 8277
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7071145496794337e-05,
+      "loss": 1.1735,
+      "step": 8278
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.70699625431019e-05,
+      "loss": 1.0838,
+      "step": 8279
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7068779493907365e-05,
+      "loss": 1.098,
+      "step": 8280
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7067596349233017e-05,
+      "loss": 1.0744,
+      "step": 8281
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7066413109101142e-05,
+      "loss": 1.0767,
+      "step": 8282
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7065229773534035e-05,
+      "loss": 1.1333,
+      "step": 8283
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7064046342553975e-05,
+      "loss": 1.1448,
+      "step": 8284
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7062862816183267e-05,
+      "loss": 1.0397,
+      "step": 8285
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.706167919444419e-05,
+      "loss": 1.1217,
+      "step": 8286
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.706049547735905e-05,
+      "loss": 1.1249,
+      "step": 8287
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7059311664950145e-05,
+      "loss": 1.0333,
+      "step": 8288
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7058127757239773e-05,
+      "loss": 0.9824,
+      "step": 8289
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7056943754250226e-05,
+      "loss": 1.1905,
+      "step": 8290
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.705575965600382e-05,
+      "loss": 1.1051,
+      "step": 8291
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7054575462522852e-05,
+      "loss": 1.0629,
+      "step": 8292
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7053391173829633e-05,
+      "loss": 1.0493,
+      "step": 8293
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7052206789946466e-05,
+      "loss": 1.1078,
+      "step": 8294
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7051022310895668e-05,
+      "loss": 1.0288,
+      "step": 8295
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7049837736699542e-05,
+      "loss": 1.0375,
+      "step": 8296
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7048653067380407e-05,
+      "loss": 1.1544,
+      "step": 8297
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7047468302960576e-05,
+      "loss": 1.0481,
+      "step": 8298
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.704628344346237e-05,
+      "loss": 1.0436,
+      "step": 8299
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7045098488908107e-05,
+      "loss": 1.093,
+      "step": 8300
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7043913439320106e-05,
+      "loss": 1.1347,
+      "step": 8301
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.704272829472069e-05,
+      "loss": 1.1188,
+      "step": 8302
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7041543055132186e-05,
+      "loss": 1.0455,
+      "step": 8303
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7040357720576918e-05,
+      "loss": 1.096,
+      "step": 8304
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7039172291077212e-05,
+      "loss": 1.0554,
+      "step": 8305
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.70379867666554e-05,
+      "loss": 1.1192,
+      "step": 8306
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7036801147333814e-05,
+      "loss": 1.1646,
+      "step": 8307
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7035615433134788e-05,
+      "loss": 1.0167,
+      "step": 8308
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7034429624080652e-05,
+      "loss": 1.1173,
+      "step": 8309
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7033243720193747e-05,
+      "loss": 1.0634,
+      "step": 8310
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.703205772149641e-05,
+      "loss": 1.0256,
+      "step": 8311
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7030871628010987e-05,
+      "loss": 1.0872,
+      "step": 8312
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7029685439759814e-05,
+      "loss": 1.1639,
+      "step": 8313
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.702849915676524e-05,
+      "loss": 1.0205,
+      "step": 8314
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7027312779049603e-05,
+      "loss": 1.0965,
+      "step": 8315
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7026126306635255e-05,
+      "loss": 0.9517,
+      "step": 8316
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.702493973954455e-05,
+      "loss": 1.0902,
+      "step": 8317
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7023753077799833e-05,
+      "loss": 1.1507,
+      "step": 8318
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7022566321423463e-05,
+      "loss": 1.0658,
+      "step": 8319
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7021379470437784e-05,
+      "loss": 1.1473,
+      "step": 8320
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7020192524865163e-05,
+      "loss": 1.0422,
+      "step": 8321
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7019005484727955e-05,
+      "loss": 1.109,
+      "step": 8322
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.701781835004852e-05,
+      "loss": 1.1221,
+      "step": 8323
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7016631120849217e-05,
+      "loss": 1.148,
+      "step": 8324
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7015443797152414e-05,
+      "loss": 1.0337,
+      "step": 8325
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.701425637898047e-05,
+      "loss": 1.1097,
+      "step": 8326
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7013068866355765e-05,
+      "loss": 1.0562,
+      "step": 8327
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7011881259300655e-05,
+      "loss": 1.0372,
+      "step": 8328
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7010693557837516e-05,
+      "loss": 1.1455,
+      "step": 8329
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.700950576198872e-05,
+      "loss": 1.0608,
+      "step": 8330
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7008317871776644e-05,
+      "loss": 1.018,
+      "step": 8331
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7007129887223656e-05,
+      "loss": 1.0913,
+      "step": 8332
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7005941808352146e-05,
+      "loss": 1.1148,
+      "step": 8333
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7004753635184483e-05,
+      "loss": 1.1041,
+      "step": 8334
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7003565367743052e-05,
+      "loss": 1.0475,
+      "step": 8335
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7002377006050243e-05,
+      "loss": 1.1197,
+      "step": 8336
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7001188550128427e-05,
+      "loss": 1.0553,
+      "step": 8337
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7000000000000003e-05,
+      "loss": 1.0365,
+      "step": 8338
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6998811355687353e-05,
+      "loss": 1.0273,
+      "step": 8339
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.699762261721287e-05,
+      "loss": 1.0473,
+      "step": 8340
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6996433784598944e-05,
+      "loss": 1.1192,
+      "step": 8341
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6995244857867972e-05,
+      "loss": 1.0366,
+      "step": 8342
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6994055837042347e-05,
+      "loss": 0.9871,
+      "step": 8343
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.699286672214447e-05,
+      "loss": 1.0314,
+      "step": 8344
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6991677513196736e-05,
+      "loss": 1.0862,
+      "step": 8345
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6990488210221546e-05,
+      "loss": 0.9691,
+      "step": 8346
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6989298813241308e-05,
+      "loss": 1.0803,
+      "step": 8347
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6988109322278417e-05,
+      "loss": 1.1406,
+      "step": 8348
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.698691973735529e-05,
+      "loss": 1.0713,
+      "step": 8349
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6985730058494324e-05,
+      "loss": 1.1209,
+      "step": 8350
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.698454028571794e-05,
+      "loss": 1.0218,
+      "step": 8351
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6983350419048545e-05,
+      "loss": 1.0625,
+      "step": 8352
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6982160458508545e-05,
+      "loss": 1.0792,
+      "step": 8353
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6980970404120368e-05,
+      "loss": 1.0304,
+      "step": 8354
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6979780255906422e-05,
+      "loss": 1.0197,
+      "step": 8355
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.697859001388913e-05,
+      "loss": 1.0782,
+      "step": 8356
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.697739967809091e-05,
+      "loss": 1.0915,
+      "step": 8357
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6976209248534185e-05,
+      "loss": 1.0963,
+      "step": 8358
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6975018725241375e-05,
+      "loss": 1.0682,
+      "step": 8359
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6973828108234913e-05,
+      "loss": 1.1168,
+      "step": 8360
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6972637397537223e-05,
+      "loss": 1.1087,
+      "step": 8361
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6971446593170733e-05,
+      "loss": 1.0998,
+      "step": 8362
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6970255695157876e-05,
+      "loss": 1.1002,
+      "step": 8363
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.696906470352108e-05,
+      "loss": 0.9954,
+      "step": 8364
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.696787361828279e-05,
+      "loss": 1.1226,
+      "step": 8365
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6966682439465432e-05,
+      "loss": 1.141,
+      "step": 8366
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.696549116709145e-05,
+      "loss": 1.1265,
+      "step": 8367
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6964299801183278e-05,
+      "loss": 1.0562,
+      "step": 8368
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6963108341763366e-05,
+      "loss": 1.0739,
+      "step": 8369
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.696191678885415e-05,
+      "loss": 1.0605,
+      "step": 8370
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6960725142478077e-05,
+      "loss": 1.0343,
+      "step": 8371
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6959533402657592e-05,
+      "loss": 1.0123,
+      "step": 8372
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.695834156941515e-05,
+      "loss": 1.0046,
+      "step": 8373
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6957149642773196e-05,
+      "loss": 1.1442,
+      "step": 8374
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6955957622754182e-05,
+      "loss": 1.0259,
+      "step": 8375
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6954765509380566e-05,
+      "loss": 1.0583,
+      "step": 8376
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.69535733026748e-05,
+      "loss": 1.1062,
+      "step": 8377
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6952381002659343e-05,
+      "loss": 1.0769,
+      "step": 8378
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6951188609356656e-05,
+      "loss": 1.0329,
+      "step": 8379
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6949996122789194e-05,
+      "loss": 1.0945,
+      "step": 8380
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6948803542979423e-05,
+      "loss": 1.068,
+      "step": 8381
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.694761086994981e-05,
+      "loss": 1.0313,
+      "step": 8382
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.694641810372282e-05,
+      "loss": 1.0622,
+      "step": 8383
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6945225244320918e-05,
+      "loss": 1.1035,
+      "step": 8384
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6944032291766574e-05,
+      "loss": 1.0879,
+      "step": 8385
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6942839246082265e-05,
+      "loss": 1.0842,
+      "step": 8386
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6941646107290455e-05,
+      "loss": 1.0582,
+      "step": 8387
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6940452875413627e-05,
+      "loss": 1.0752,
+      "step": 8388
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6939259550474258e-05,
+      "loss": 1.0866,
+      "step": 8389
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.693806613249482e-05,
+      "loss": 1.0474,
+      "step": 8390
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6936872621497798e-05,
+      "loss": 1.1371,
+      "step": 8391
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.693567901750567e-05,
+      "loss": 1.0254,
+      "step": 8392
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6934485320540927e-05,
+      "loss": 1.1182,
+      "step": 8393
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6933291530626048e-05,
+      "loss": 1.0681,
+      "step": 8394
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.693209764778352e-05,
+      "loss": 1.107,
+      "step": 8395
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6930903672035834e-05,
+      "loss": 1.1066,
+      "step": 8396
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.692970960340548e-05,
+      "loss": 1.117,
+      "step": 8397
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.692851544191496e-05,
+      "loss": 1.0997,
+      "step": 8398
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6927321187586753e-05,
+      "loss": 1.0634,
+      "step": 8399
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.692612684044336e-05,
+      "loss": 1.0819,
+      "step": 8400
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6924932400507283e-05,
+      "loss": 1.1358,
+      "step": 8401
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.692373786780102e-05,
+      "loss": 1.1228,
+      "step": 8402
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6922543242347067e-05,
+      "loss": 1.1497,
+      "step": 8403
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6921348524167933e-05,
+      "loss": 1.1321,
+      "step": 8404
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.692015371328612e-05,
+      "loss": 1.1569,
+      "step": 8405
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6918958809724135e-05,
+      "loss": 1.0815,
+      "step": 8406
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6917763813504488e-05,
+      "loss": 1.0958,
+      "step": 8407
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.691656872464969e-05,
+      "loss": 1.0416,
+      "step": 8408
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6915373543182246e-05,
+      "loss": 1.0678,
+      "step": 8409
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.691417826912467e-05,
+      "loss": 1.1832,
+      "step": 8410
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6912982902499487e-05,
+      "loss": 1.0935,
+      "step": 8411
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6911787443329205e-05,
+      "loss": 1.1198,
+      "step": 8412
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6910591891636346e-05,
+      "loss": 1.0842,
+      "step": 8413
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.690939624744343e-05,
+      "loss": 1.117,
+      "step": 8414
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6908200510772978e-05,
+      "loss": 0.9872,
+      "step": 8415
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6907004681647515e-05,
+      "loss": 1.0817,
+      "step": 8416
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6905808760089567e-05,
+      "loss": 0.9924,
+      "step": 8417
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6904612746121657e-05,
+      "loss": 1.089,
+      "step": 8418
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6903416639766325e-05,
+      "loss": 1.0762,
+      "step": 8419
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.690222044104609e-05,
+      "loss": 1.0668,
+      "step": 8420
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.690102414998349e-05,
+      "loss": 1.0481,
+      "step": 8421
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.689982776660106e-05,
+      "loss": 1.0836,
+      "step": 8422
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6898631290921332e-05,
+      "loss": 1.041,
+      "step": 8423
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.689743472296685e-05,
+      "loss": 1.0724,
+      "step": 8424
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.689623806276015e-05,
+      "loss": 1.0616,
+      "step": 8425
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6895041310323773e-05,
+      "loss": 1.1521,
+      "step": 8426
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.689384446568026e-05,
+      "loss": 1.1037,
+      "step": 8427
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.689264752885216e-05,
+      "loss": 1.1541,
+      "step": 8428
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.689145049986202e-05,
+      "loss": 1.082,
+      "step": 8429
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6890253378732386e-05,
+      "loss": 1.1466,
+      "step": 8430
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6889056165485806e-05,
+      "loss": 1.0836,
+      "step": 8431
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6887858860144834e-05,
+      "loss": 1.1156,
+      "step": 8432
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.688666146273202e-05,
+      "loss": 1.1056,
+      "step": 8433
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6885463973269928e-05,
+      "loss": 1.1062,
+      "step": 8434
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6884266391781108e-05,
+      "loss": 1.0926,
+      "step": 8435
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.688306871828812e-05,
+      "loss": 0.9471,
+      "step": 8436
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6881870952813525e-05,
+      "loss": 1.0609,
+      "step": 8437
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6880673095379883e-05,
+      "loss": 1.0792,
+      "step": 8438
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.687947514600976e-05,
+      "loss": 1.1075,
+      "step": 8439
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.687827710472572e-05,
+      "loss": 1.0303,
+      "step": 8440
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6877078971550336e-05,
+      "loss": 1.1585,
+      "step": 8441
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.687588074650617e-05,
+      "loss": 1.0578,
+      "step": 8442
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.687468242961579e-05,
+      "loss": 1.0398,
+      "step": 8443
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6873484020901778e-05,
+      "loss": 1.0207,
+      "step": 8444
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6872285520386705e-05,
+      "loss": 1.1399,
+      "step": 8445
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6871086928093143e-05,
+      "loss": 1.0332,
+      "step": 8446
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6869888244043673e-05,
+      "loss": 1.027,
+      "step": 8447
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6868689468260877e-05,
+      "loss": 1.0336,
+      "step": 8448
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.686749060076733e-05,
+      "loss": 1.0742,
+      "step": 8449
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.686629164158562e-05,
+      "loss": 1.0043,
+      "step": 8450
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.686509259073833e-05,
+      "loss": 1.1066,
+      "step": 8451
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6863893448248044e-05,
+      "loss": 1.0591,
+      "step": 8452
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6862694214137353e-05,
+      "loss": 1.0627,
+      "step": 8453
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6861494888428845e-05,
+      "loss": 1.0407,
+      "step": 8454
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6860295471145114e-05,
+      "loss": 1.0464,
+      "step": 8455
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.685909596230875e-05,
+      "loss": 1.097,
+      "step": 8456
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.685789636194235e-05,
+      "loss": 1.0591,
+      "step": 8457
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.685669667006851e-05,
+      "loss": 1.1509,
+      "step": 8458
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.685549688670983e-05,
+      "loss": 1.0603,
+      "step": 8459
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6854297011888904e-05,
+      "loss": 1.196,
+      "step": 8460
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.685309704562835e-05,
+      "loss": 1.0361,
+      "step": 8461
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.685189698795075e-05,
+      "loss": 1.1299,
+      "step": 8462
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.685069683887872e-05,
+      "loss": 1.0395,
+      "step": 8463
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.684949659843487e-05,
+      "loss": 1.0453,
+      "step": 8464
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6848296266641803e-05,
+      "loss": 1.0299,
+      "step": 8465
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6847095843522134e-05,
+      "loss": 1.0903,
+      "step": 8466
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.684589532909847e-05,
+      "loss": 1.0712,
+      "step": 8467
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.684469472339343e-05,
+      "loss": 1.0367,
+      "step": 8468
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6843494026429626e-05,
+      "loss": 1.0129,
+      "step": 8469
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6842293238229677e-05,
+      "loss": 1.0589,
+      "step": 8470
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6841092358816203e-05,
+      "loss": 0.9856,
+      "step": 8471
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6839891388211822e-05,
+      "loss": 1.1505,
+      "step": 8472
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.683869032643916e-05,
+      "loss": 1.1311,
+      "step": 8473
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6837489173520838e-05,
+      "loss": 1.0483,
+      "step": 8474
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6836287929479483e-05,
+      "loss": 0.977,
+      "step": 8475
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6835086594337726e-05,
+      "loss": 1.0,
+      "step": 8476
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.683388516811819e-05,
+      "loss": 1.0448,
+      "step": 8477
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.683268365084351e-05,
+      "loss": 1.0081,
+      "step": 8478
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6831482042536318e-05,
+      "loss": 1.1247,
+      "step": 8479
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6830280343219247e-05,
+      "loss": 1.0959,
+      "step": 8480
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6829078552914937e-05,
+      "loss": 1.0548,
+      "step": 8481
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6827876671646027e-05,
+      "loss": 1.0695,
+      "step": 8482
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6826674699435147e-05,
+      "loss": 0.9862,
+      "step": 8483
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6825472636304953e-05,
+      "loss": 1.1115,
+      "step": 8484
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6824270482278072e-05,
+      "loss": 1.0129,
+      "step": 8485
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.682306823737716e-05,
+      "loss": 1.0697,
+      "step": 8486
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6821865901624858e-05,
+      "loss": 1.0003,
+      "step": 8487
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6820663475043817e-05,
+      "loss": 1.103,
+      "step": 8488
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6819460957656688e-05,
+      "loss": 1.0275,
+      "step": 8489
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6818258349486122e-05,
+      "loss": 1.0769,
+      "step": 8490
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6817055650554767e-05,
+      "loss": 0.9181,
+      "step": 8491
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6815852860885288e-05,
+      "loss": 1.0942,
+      "step": 8492
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6814649980500333e-05,
+      "loss": 1.056,
+      "step": 8493
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6813447009422563e-05,
+      "loss": 1.0303,
+      "step": 8494
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.681224394767464e-05,
+      "loss": 1.003,
+      "step": 8495
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6811040795279222e-05,
+      "loss": 1.1228,
+      "step": 8496
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.680983755225898e-05,
+      "loss": 1.0982,
+      "step": 8497
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.680863421863657e-05,
+      "loss": 1.0576,
+      "step": 8498
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.680743079443467e-05,
+      "loss": 1.0136,
+      "step": 8499
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6806227279675933e-05,
+      "loss": 1.0578,
+      "step": 8500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6805023674383047e-05,
+      "loss": 1.043,
+      "step": 8501
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.680381997857867e-05,
+      "loss": 1.0739,
+      "step": 8502
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6802616192285486e-05,
+      "loss": 1.1264,
+      "step": 8503
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6801412315526164e-05,
+      "loss": 1.0547,
+      "step": 8504
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6800208348323388e-05,
+      "loss": 1.0821,
+      "step": 8505
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6799004290699824e-05,
+      "loss": 1.122,
+      "step": 8506
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6797800142678167e-05,
+      "loss": 1.1793,
+      "step": 8507
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.679659590428109e-05,
+      "loss": 0.9163,
+      "step": 8508
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6795391575531285e-05,
+      "loss": 1.0431,
+      "step": 8509
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6794187156451428e-05,
+      "loss": 1.1154,
+      "step": 8510
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6792982647064214e-05,
+      "loss": 1.1023,
+      "step": 8511
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.679177804739233e-05,
+      "loss": 1.0785,
+      "step": 8512
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6790573357458466e-05,
+      "loss": 1.0921,
+      "step": 8513
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6789368577285315e-05,
+      "loss": 1.1034,
+      "step": 8514
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.678816370689557e-05,
+      "loss": 1.1005,
+      "step": 8515
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6786958746311933e-05,
+      "loss": 1.0731,
+      "step": 8516
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6785753695557092e-05,
+      "loss": 1.0062,
+      "step": 8517
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6784548554653755e-05,
+      "loss": 1.0752,
+      "step": 8518
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.678334332362462e-05,
+      "loss": 1.1512,
+      "step": 8519
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6782138002492385e-05,
+      "loss": 0.9522,
+      "step": 8520
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6780932591279763e-05,
+      "loss": 1.0583,
+      "step": 8521
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6779727090009453e-05,
+      "loss": 1.0979,
+      "step": 8522
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6778521498704166e-05,
+      "loss": 1.0189,
+      "step": 8523
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6777315817386615e-05,
+      "loss": 1.0347,
+      "step": 8524
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.67761100460795e-05,
+      "loss": 1.1636,
+      "step": 8525
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.677490418480555e-05,
+      "loss": 1.0663,
+      "step": 8526
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6773698233587465e-05,
+      "loss": 1.1308,
+      "step": 8527
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.677249219244797e-05,
+      "loss": 1.1142,
+      "step": 8528
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.677128606140978e-05,
+      "loss": 1.0408,
+      "step": 8529
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6770079840495615e-05,
+      "loss": 1.1597,
+      "step": 8530
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6768873529728192e-05,
+      "loss": 1.0103,
+      "step": 8531
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6767667129130242e-05,
+      "loss": 1.0848,
+      "step": 8532
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6766460638724486e-05,
+      "loss": 1.0761,
+      "step": 8533
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.676525405853365e-05,
+      "loss": 0.9992,
+      "step": 8534
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.676404738858047e-05,
+      "loss": 1.0619,
+      "step": 8535
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.676284062888766e-05,
+      "loss": 0.9866,
+      "step": 8536
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6761633779477966e-05,
+      "loss": 1.0385,
+      "step": 8537
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.676042684037411e-05,
+      "loss": 1.0416,
+      "step": 8538
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6759219811598834e-05,
+      "loss": 1.1138,
+      "step": 8539
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6758012693174875e-05,
+      "loss": 1.0927,
+      "step": 8540
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.675680548512497e-05,
+      "loss": 1.018,
+      "step": 8541
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6755598187471855e-05,
+      "loss": 1.0151,
+      "step": 8542
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.675439080023828e-05,
+      "loss": 1.0438,
+      "step": 8543
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.675318332344698e-05,
+      "loss": 1.0233,
+      "step": 8544
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.67519757571207e-05,
+      "loss": 1.0633,
+      "step": 8545
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6750768101282197e-05,
+      "loss": 1.0951,
+      "step": 8546
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6749560355954212e-05,
+      "loss": 1.0425,
+      "step": 8547
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6748352521159493e-05,
+      "loss": 1.069,
+      "step": 8548
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6747144596920792e-05,
+      "loss": 1.1128,
+      "step": 8549
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.674593658326087e-05,
+      "loss": 1.024,
+      "step": 8550
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6744728480202476e-05,
+      "loss": 1.1558,
+      "step": 8551
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6743520287768366e-05,
+      "loss": 1.0796,
+      "step": 8552
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6742312005981303e-05,
+      "loss": 1.0225,
+      "step": 8553
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.674110363486404e-05,
+      "loss": 1.0605,
+      "step": 8554
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.673989517443935e-05,
+      "loss": 1.0809,
+      "step": 8555
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.673868662472999e-05,
+      "loss": 1.0514,
+      "step": 8556
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6737477985758723e-05,
+      "loss": 1.0926,
+      "step": 8557
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6736269257548316e-05,
+      "loss": 1.1024,
+      "step": 8558
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6735060440121544e-05,
+      "loss": 1.0695,
+      "step": 8559
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6733851533501173e-05,
+      "loss": 1.0301,
+      "step": 8560
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6732642537709974e-05,
+      "loss": 1.088,
+      "step": 8561
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6731433452770722e-05,
+      "loss": 0.9899,
+      "step": 8562
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6730224278706193e-05,
+      "loss": 1.0892,
+      "step": 8563
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6729015015539165e-05,
+      "loss": 1.0975,
+      "step": 8564
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6727805663292415e-05,
+      "loss": 1.061,
+      "step": 8565
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6726596221988724e-05,
+      "loss": 1.1332,
+      "step": 8566
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6725386691650868e-05,
+      "loss": 1.0765,
+      "step": 8567
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6724177072301643e-05,
+      "loss": 1.0911,
+      "step": 8568
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6722967363963822e-05,
+      "loss": 1.1184,
+      "step": 8569
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6721757566660202e-05,
+      "loss": 1.1721,
+      "step": 8570
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6720547680413563e-05,
+      "loss": 1.0518,
+      "step": 8571
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.671933770524671e-05,
+      "loss": 1.113,
+      "step": 8572
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6718127641182415e-05,
+      "loss": 1.0123,
+      "step": 8573
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6716917488243485e-05,
+      "loss": 1.0634,
+      "step": 8574
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6715707246452712e-05,
+      "loss": 1.0571,
+      "step": 8575
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6714496915832893e-05,
+      "loss": 1.0817,
+      "step": 8576
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6713286496406833e-05,
+      "loss": 0.99,
+      "step": 8577
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6712075988197317e-05,
+      "loss": 1.0828,
+      "step": 8578
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6710865391227165e-05,
+      "loss": 1.0612,
+      "step": 8579
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6709654705519167e-05,
+      "loss": 1.0779,
+      "step": 8580
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.670844393109614e-05,
+      "loss": 1.0521,
+      "step": 8581
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.670723306798088e-05,
+      "loss": 1.0552,
+      "step": 8582
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.670602211619621e-05,
+      "loss": 1.049,
+      "step": 8583
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6704811075764922e-05,
+      "loss": 1.1845,
+      "step": 8584
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6703599946709844e-05,
+      "loss": 1.0417,
+      "step": 8585
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6702388729053782e-05,
+      "loss": 1.0932,
+      "step": 8586
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6701177422819554e-05,
+      "loss": 1.0049,
+      "step": 8587
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6699966028029977e-05,
+      "loss": 1.0506,
+      "step": 8588
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6698754544707865e-05,
+      "loss": 1.1194,
+      "step": 8589
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.669754297287605e-05,
+      "loss": 1.0534,
+      "step": 8590
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.669633131255734e-05,
+      "loss": 1.0878,
+      "step": 8591
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.669511956377457e-05,
+      "loss": 1.1477,
+      "step": 8592
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.669390772655056e-05,
+      "loss": 1.1056,
+      "step": 8593
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.669269580090814e-05,
+      "loss": 1.0505,
+      "step": 8594
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6691483786870136e-05,
+      "loss": 1.0773,
+      "step": 8595
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6690271684459382e-05,
+      "loss": 1.0776,
+      "step": 8596
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6689059493698707e-05,
+      "loss": 1.1116,
+      "step": 8597
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6687847214610944e-05,
+      "loss": 1.0433,
+      "step": 8598
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6686634847218935e-05,
+      "loss": 1.0571,
+      "step": 8599
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.668542239154551e-05,
+      "loss": 1.1186,
+      "step": 8600
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.668420984761351e-05,
+      "loss": 1.1708,
+      "step": 8601
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.668299721544578e-05,
+      "loss": 1.0367,
+      "step": 8602
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6681784495065153e-05,
+      "loss": 1.0931,
+      "step": 8603
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6680571686494483e-05,
+      "loss": 1.0342,
+      "step": 8604
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.667935878975661e-05,
+      "loss": 1.0957,
+      "step": 8605
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6678145804874378e-05,
+      "loss": 1.0341,
+      "step": 8606
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6676932731870643e-05,
+      "loss": 1.0371,
+      "step": 8607
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.667571957076825e-05,
+      "loss": 1.1134,
+      "step": 8608
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6674506321590056e-05,
+      "loss": 1.0443,
+      "step": 8609
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6673292984358912e-05,
+      "loss": 1.0758,
+      "step": 8610
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6672079559097668e-05,
+      "loss": 1.1178,
+      "step": 8611
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6670866045829192e-05,
+      "loss": 1.0423,
+      "step": 8612
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.666965244457633e-05,
+      "loss": 1.0582,
+      "step": 8613
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.666843875536196e-05,
+      "loss": 1.1566,
+      "step": 8614
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6667224978208926e-05,
+      "loss": 1.1479,
+      "step": 8615
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6666011113140104e-05,
+      "loss": 0.984,
+      "step": 8616
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.666479716017835e-05,
+      "loss": 0.9882,
+      "step": 8617
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.666358311934654e-05,
+      "loss": 1.1093,
+      "step": 8618
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.666236899066754e-05,
+      "loss": 1.1033,
+      "step": 8619
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6661154774164214e-05,
+      "loss": 1.0743,
+      "step": 8620
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.665994046985944e-05,
+      "loss": 1.0857,
+      "step": 8621
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6658726077776095e-05,
+      "loss": 1.1305,
+      "step": 8622
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6657511597937045e-05,
+      "loss": 1.0478,
+      "step": 8623
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6656297030365176e-05,
+      "loss": 1.0871,
+      "step": 8624
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.665508237508336e-05,
+      "loss": 1.1021,
+      "step": 8625
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6653867632114478e-05,
+      "loss": 1.0928,
+      "step": 8626
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.665265280148142e-05,
+      "loss": 1.1023,
+      "step": 8627
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6651437883207057e-05,
+      "loss": 1.059,
+      "step": 8628
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6650222877314282e-05,
+      "loss": 1.1279,
+      "step": 8629
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6649007783825982e-05,
+      "loss": 1.0527,
+      "step": 8630
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6647792602765042e-05,
+      "loss": 1.0273,
+      "step": 8631
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6646577334154357e-05,
+      "loss": 1.0746,
+      "step": 8632
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6645361978016813e-05,
+      "loss": 1.0124,
+      "step": 8633
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6644146534375307e-05,
+      "loss": 1.0099,
+      "step": 8634
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6642931003252736e-05,
+      "loss": 1.0728,
+      "step": 8635
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.664171538467199e-05,
+      "loss": 1.0352,
+      "step": 8636
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6640499678655975e-05,
+      "loss": 1.1058,
+      "step": 8637
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6639283885227586e-05,
+      "loss": 0.99,
+      "step": 8638
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.663806800440973e-05,
+      "loss": 1.0363,
+      "step": 8639
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6636852036225304e-05,
+      "loss": 1.1214,
+      "step": 8640
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6635635980697216e-05,
+      "loss": 1.0497,
+      "step": 8641
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6634419837848372e-05,
+      "loss": 1.1016,
+      "step": 8642
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6633203607701684e-05,
+      "loss": 1.0311,
+      "step": 8643
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6631987290280055e-05,
+      "loss": 1.0514,
+      "step": 8644
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6630770885606402e-05,
+      "loss": 1.0929,
+      "step": 8645
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6629554393703636e-05,
+      "loss": 1.0706,
+      "step": 8646
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6628337814594675e-05,
+      "loss": 1.0545,
+      "step": 8647
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6627121148302428e-05,
+      "loss": 1.078,
+      "step": 8648
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6625904394849823e-05,
+      "loss": 1.1024,
+      "step": 8649
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6624687554259775e-05,
+      "loss": 1.0855,
+      "step": 8650
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.66234706265552e-05,
+      "loss": 1.0899,
+      "step": 8651
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.662225361175903e-05,
+      "loss": 1.1019,
+      "step": 8652
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6621036509894184e-05,
+      "loss": 1.1126,
+      "step": 8653
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6619819320983594e-05,
+      "loss": 1.1119,
+      "step": 8654
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6618602045050182e-05,
+      "loss": 1.0873,
+      "step": 8655
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6617384682116883e-05,
+      "loss": 1.0808,
+      "step": 8656
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.661616723220662e-05,
+      "loss": 1.0939,
+      "step": 8657
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6614949695342336e-05,
+      "loss": 1.0878,
+      "step": 8658
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6613732071546956e-05,
+      "loss": 1.1383,
+      "step": 8659
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6612514360843427e-05,
+      "loss": 1.1158,
+      "step": 8660
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6611296563254677e-05,
+      "loss": 1.0676,
+      "step": 8661
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.661007867880365e-05,
+      "loss": 1.0338,
+      "step": 8662
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.660886070751329e-05,
+      "loss": 1.0524,
+      "step": 8663
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.660764264940653e-05,
+      "loss": 1.053,
+      "step": 8664
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6606424504506325e-05,
+      "loss": 1.0594,
+      "step": 8665
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6605206272835616e-05,
+      "loss": 1.1456,
+      "step": 8666
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6603987954417348e-05,
+      "loss": 1.0685,
+      "step": 8667
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6602769549274478e-05,
+      "loss": 1.0868,
+      "step": 8668
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.660155105742995e-05,
+      "loss": 1.1249,
+      "step": 8669
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6600332478906718e-05,
+      "loss": 1.0077,
+      "step": 8670
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6599113813727738e-05,
+      "loss": 1.0823,
+      "step": 8671
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.659789506191597e-05,
+      "loss": 1.1159,
+      "step": 8672
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.659667622349436e-05,
+      "loss": 1.0437,
+      "step": 8673
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6595457298485872e-05,
+      "loss": 1.0114,
+      "step": 8674
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6594238286913473e-05,
+      "loss": 1.0938,
+      "step": 8675
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.659301918880012e-05,
+      "loss": 1.1555,
+      "step": 8676
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6591800004168774e-05,
+      "loss": 1.0558,
+      "step": 8677
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.659058073304241e-05,
+      "loss": 1.0981,
+      "step": 8678
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6589361375443983e-05,
+      "loss": 0.953,
+      "step": 8679
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.658814193139647e-05,
+      "loss": 1.0516,
+      "step": 8680
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.658692240092284e-05,
+      "loss": 0.9968,
+      "step": 8681
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6585702784046064e-05,
+      "loss": 1.0999,
+      "step": 8682
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.658448308078912e-05,
+      "loss": 1.1477,
+      "step": 8683
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6583263291174975e-05,
+      "loss": 1.11,
+      "step": 8684
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6582043415226612e-05,
+      "loss": 0.9798,
+      "step": 8685
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.658082345296701e-05,
+      "loss": 1.0939,
+      "step": 8686
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6579603404419148e-05,
+      "loss": 1.1067,
+      "step": 8687
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6578383269606004e-05,
+      "loss": 1.1455,
+      "step": 8688
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6577163048550566e-05,
+      "loss": 1.1458,
+      "step": 8689
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.657594274127582e-05,
+      "loss": 1.1279,
+      "step": 8690
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.657472234780475e-05,
+      "loss": 1.0786,
+      "step": 8691
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6573501868160347e-05,
+      "loss": 1.1045,
+      "step": 8692
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6572281302365597e-05,
+      "loss": 1.1197,
+      "step": 8693
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6571060650443494e-05,
+      "loss": 1.102,
+      "step": 8694
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.656983991241703e-05,
+      "loss": 1.1205,
+      "step": 8695
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6568619088309203e-05,
+      "loss": 1.0244,
+      "step": 8696
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6567398178143004e-05,
+      "loss": 1.1146,
+      "step": 8697
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6566177181941438e-05,
+      "loss": 1.116,
+      "step": 8698
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.65649560997275e-05,
+      "loss": 1.1328,
+      "step": 8699
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.656373493152419e-05,
+      "loss": 1.0637,
+      "step": 8700
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6562513677354516e-05,
+      "loss": 1.0647,
+      "step": 8701
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.656129233724148e-05,
+      "loss": 1.1112,
+      "step": 8702
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.656007091120809e-05,
+      "loss": 1.0783,
+      "step": 8703
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6558849399277347e-05,
+      "loss": 1.0796,
+      "step": 8704
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.655762780147227e-05,
+      "loss": 1.11,
+      "step": 8705
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6556406117815865e-05,
+      "loss": 1.0866,
+      "step": 8706
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6555184348331142e-05,
+      "loss": 1.1326,
+      "step": 8707
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.655396249304112e-05,
+      "loss": 1.082,
+      "step": 8708
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6552740551968818e-05,
+      "loss": 1.0904,
+      "step": 8709
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.655151852513724e-05,
+      "loss": 1.0619,
+      "step": 8710
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6550296412569422e-05,
+      "loss": 1.0905,
+      "step": 8711
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6549074214288368e-05,
+      "loss": 1.0935,
+      "step": 8712
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.654785193031712e-05,
+      "loss": 1.0997,
+      "step": 8713
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.654662956067868e-05,
+      "loss": 1.0772,
+      "step": 8714
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.654540710539609e-05,
+      "loss": 1.1011,
+      "step": 8715
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6544184564492375e-05,
+      "loss": 1.21,
+      "step": 8716
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6542961937990558e-05,
+      "loss": 1.1399,
+      "step": 8717
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.654173922591367e-05,
+      "loss": 1.091,
+      "step": 8718
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.654051642828475e-05,
+      "loss": 1.0739,
+      "step": 8719
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.653929354512682e-05,
+      "loss": 1.0917,
+      "step": 8720
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6538070576462927e-05,
+      "loss": 1.104,
+      "step": 8721
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.65368475223161e-05,
+      "loss": 1.094,
+      "step": 8722
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.653562438270938e-05,
+      "loss": 1.1439,
+      "step": 8723
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6534401157665813e-05,
+      "loss": 1.1134,
+      "step": 8724
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.653317784720843e-05,
+      "loss": 1.0409,
+      "step": 8725
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.653195445136028e-05,
+      "loss": 1.0715,
+      "step": 8726
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.653073097014441e-05,
+      "loss": 1.0093,
+      "step": 8727
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.652950740358386e-05,
+      "loss": 1.0711,
+      "step": 8728
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6528283751701685e-05,
+      "loss": 1.0277,
+      "step": 8729
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6527060014520933e-05,
+      "loss": 1.1196,
+      "step": 8730
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6525836192064652e-05,
+      "loss": 1.0608,
+      "step": 8731
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6524612284355896e-05,
+      "loss": 1.0632,
+      "step": 8732
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.652338829141772e-05,
+      "loss": 1.078,
+      "step": 8733
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6522164213273186e-05,
+      "loss": 1.0981,
+      "step": 8734
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6520940049945344e-05,
+      "loss": 1.1239,
+      "step": 8735
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6519715801457257e-05,
+      "loss": 1.0961,
+      "step": 8736
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.651849146783198e-05,
+      "loss": 1.0782,
+      "step": 8737
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6517267049092585e-05,
+      "loss": 1.1848,
+      "step": 8738
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.651604254526213e-05,
+      "loss": 1.1519,
+      "step": 8739
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6514817956363684e-05,
+      "loss": 1.1229,
+      "step": 8740
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6513593282420312e-05,
+      "loss": 1.0845,
+      "step": 8741
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6512368523455086e-05,
+      "loss": 1.0536,
+      "step": 8742
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.651114367949107e-05,
+      "loss": 1.04,
+      "step": 8743
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6509918750551343e-05,
+      "loss": 1.0952,
+      "step": 8744
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6508693736658978e-05,
+      "loss": 1.0294,
+      "step": 8745
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.650746863783705e-05,
+      "loss": 1.0482,
+      "step": 8746
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.650624345410863e-05,
+      "loss": 1.128,
+      "step": 8747
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.65050181854968e-05,
+      "loss": 1.0696,
+      "step": 8748
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.650379283202465e-05,
+      "loss": 1.0921,
+      "step": 8749
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6502567393715246e-05,
+      "loss": 1.0728,
+      "step": 8750
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6501341870591684e-05,
+      "loss": 1.1384,
+      "step": 8751
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6500116262677038e-05,
+      "loss": 1.0984,
+      "step": 8752
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6498890569994408e-05,
+      "loss": 1.1088,
+      "step": 8753
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6497664792566866e-05,
+      "loss": 1.1212,
+      "step": 8754
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6496438930417517e-05,
+      "loss": 1.0667,
+      "step": 8755
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6495212983569444e-05,
+      "loss": 1.1295,
+      "step": 8756
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.649398695204574e-05,
+      "loss": 1.1239,
+      "step": 8757
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6492760835869503e-05,
+      "loss": 1.0532,
+      "step": 8758
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.649153463506383e-05,
+      "loss": 1.0665,
+      "step": 8759
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.649030834965181e-05,
+      "loss": 1.1442,
+      "step": 8760
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6489081979656554e-05,
+      "loss": 1.0627,
+      "step": 8761
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6487855525101157e-05,
+      "loss": 1.0337,
+      "step": 8762
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6486628986008718e-05,
+      "loss": 1.0492,
+      "step": 8763
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6485402362402347e-05,
+      "loss": 1.1952,
+      "step": 8764
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6484175654305147e-05,
+      "loss": 1.0757,
+      "step": 8765
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.648294886174023e-05,
+      "loss": 1.0658,
+      "step": 8766
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6481721984730698e-05,
+      "loss": 1.0839,
+      "step": 8767
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6480495023299666e-05,
+      "loss": 1.0966,
+      "step": 8768
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6479267977470244e-05,
+      "loss": 1.1173,
+      "step": 8769
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6478040847265546e-05,
+      "loss": 1.0777,
+      "step": 8770
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.647681363270869e-05,
+      "loss": 1.1523,
+      "step": 8771
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6475586333822786e-05,
+      "loss": 0.9908,
+      "step": 8772
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.647435895063096e-05,
+      "loss": 1.0485,
+      "step": 8773
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6473131483156326e-05,
+      "loss": 1.0616,
+      "step": 8774
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6471903931422012e-05,
+      "loss": 1.0669,
+      "step": 8775
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.647067629545114e-05,
+      "loss": 1.023,
+      "step": 8776
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6469448575266824e-05,
+      "loss": 1.0558,
+      "step": 8777
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6468220770892208e-05,
+      "loss": 1.1459,
+      "step": 8778
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6466992882350404e-05,
+      "loss": 1.0966,
+      "step": 8779
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.646576490966455e-05,
+      "loss": 1.0797,
+      "step": 8780
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6464536852857776e-05,
+      "loss": 1.0414,
+      "step": 8781
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.646330871195322e-05,
+      "loss": 1.0333,
+      "step": 8782
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6462080486974e-05,
+      "loss": 1.0705,
+      "step": 8783
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.646085217794327e-05,
+      "loss": 1.092,
+      "step": 8784
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6459623784884155e-05,
+      "loss": 1.0521,
+      "step": 8785
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6458395307819807e-05,
+      "loss": 1.0586,
+      "step": 8786
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.645716674677335e-05,
+      "loss": 1.0772,
+      "step": 8787
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.645593810176794e-05,
+      "loss": 1.0205,
+      "step": 8788
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6454709372826713e-05,
+      "loss": 1.0758,
+      "step": 8789
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6453480559972817e-05,
+      "loss": 1.1094,
+      "step": 8790
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.64522516632294e-05,
+      "loss": 1.1071,
+      "step": 8791
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6451022682619606e-05,
+      "loss": 1.0515,
+      "step": 8792
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6449793618166594e-05,
+      "loss": 1.0785,
+      "step": 8793
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6448564469893508e-05,
+      "loss": 1.1283,
+      "step": 8794
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6447335237823504e-05,
+      "loss": 1.135,
+      "step": 8795
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6446105921979732e-05,
+      "loss": 1.0144,
+      "step": 8796
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6444876522385358e-05,
+      "loss": 1.0658,
+      "step": 8797
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6443647039063535e-05,
+      "loss": 1.1132,
+      "step": 8798
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.644241747203742e-05,
+      "loss": 1.0549,
+      "step": 8799
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6441187821330175e-05,
+      "loss": 1.0941,
+      "step": 8800
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6439958086964968e-05,
+      "loss": 1.1088,
+      "step": 8801
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6438728268964957e-05,
+      "loss": 1.0909,
+      "step": 8802
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6437498367353315e-05,
+      "loss": 1.0166,
+      "step": 8803
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.64362683821532e-05,
+      "loss": 1.0459,
+      "step": 8804
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6435038313387786e-05,
+      "loss": 1.1341,
+      "step": 8805
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.643380816108025e-05,
+      "loss": 1.1099,
+      "step": 8806
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6432577925253753e-05,
+      "loss": 1.1195,
+      "step": 8807
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.643134760593147e-05,
+      "loss": 1.2103,
+      "step": 8808
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6430117203136584e-05,
+      "loss": 1.0989,
+      "step": 8809
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6428886716892267e-05,
+      "loss": 1.1133,
+      "step": 8810
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6427656147221697e-05,
+      "loss": 1.0647,
+      "step": 8811
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6426425494148062e-05,
+      "loss": 1.076,
+      "step": 8812
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.642519475769453e-05,
+      "loss": 1.0559,
+      "step": 8813
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6423963937884295e-05,
+      "loss": 1.0826,
+      "step": 8814
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6422733034740536e-05,
+      "loss": 1.1257,
+      "step": 8815
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.642150204828644e-05,
+      "loss": 1.0946,
+      "step": 8816
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6420270978545198e-05,
+      "loss": 1.0824,
+      "step": 8817
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.641903982554e-05,
+      "loss": 1.0957,
+      "step": 8818
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.641780858929403e-05,
+      "loss": 1.1747,
+      "step": 8819
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.641657726983049e-05,
+      "loss": 1.0592,
+      "step": 8820
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6415345867172563e-05,
+      "loss": 1.0495,
+      "step": 8821
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.641411438134346e-05,
+      "loss": 1.0887,
+      "step": 8822
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.641288281236636e-05,
+      "loss": 1.0252,
+      "step": 8823
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6411651160264478e-05,
+      "loss": 1.0946,
+      "step": 8824
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6410419425061004e-05,
+      "loss": 1.0738,
+      "step": 8825
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.640918760677915e-05,
+      "loss": 1.019,
+      "step": 8826
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6407955705442106e-05,
+      "loss": 1.0702,
+      "step": 8827
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.640672372107309e-05,
+      "loss": 1.1464,
+      "step": 8828
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.64054916536953e-05,
+      "loss": 1.0864,
+      "step": 8829
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6404259503331947e-05,
+      "loss": 1.0565,
+      "step": 8830
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6403027270006242e-05,
+      "loss": 1.014,
+      "step": 8831
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.64017949537414e-05,
+      "loss": 1.0683,
+      "step": 8832
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6400562554560625e-05,
+      "loss": 1.1312,
+      "step": 8833
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6399330072487138e-05,
+      "loss": 1.0797,
+      "step": 8834
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.639809750754415e-05,
+      "loss": 1.1422,
+      "step": 8835
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6396864859754884e-05,
+      "loss": 1.0154,
+      "step": 8836
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.639563212914256e-05,
+      "loss": 1.1304,
+      "step": 8837
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.639439931573039e-05,
+      "loss": 1.0873,
+      "step": 8838
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6393166419541607e-05,
+      "loss": 1.1246,
+      "step": 8839
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.639193344059943e-05,
+      "loss": 1.0686,
+      "step": 8840
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.639070037892708e-05,
+      "loss": 1.0409,
+      "step": 8841
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6389467234547793e-05,
+      "loss": 1.0738,
+      "step": 8842
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.638823400748479e-05,
+      "loss": 1.0919,
+      "step": 8843
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6387000697761304e-05,
+      "loss": 1.021,
+      "step": 8844
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.638576730540057e-05,
+      "loss": 1.1037,
+      "step": 8845
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6384533830425815e-05,
+      "loss": 1.0582,
+      "step": 8846
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6383300272860276e-05,
+      "loss": 1.1069,
+      "step": 8847
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.638206663272719e-05,
+      "loss": 1.1345,
+      "step": 8848
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.63808329100498e-05,
+      "loss": 1.1446,
+      "step": 8849
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6379599104851338e-05,
+      "loss": 1.0576,
+      "step": 8850
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.637836521715505e-05,
+      "loss": 1.1138,
+      "step": 8851
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.637713124698417e-05,
+      "loss": 1.1373,
+      "step": 8852
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.637589719436195e-05,
+      "loss": 1.0519,
+      "step": 8853
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6374663059311632e-05,
+      "loss": 1.0722,
+      "step": 8854
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.637342884185647e-05,
+      "loss": 1.0948,
+      "step": 8855
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6372194542019706e-05,
+      "loss": 1.0822,
+      "step": 8856
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.637096015982459e-05,
+      "loss": 1.0556,
+      "step": 8857
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6369725695294376e-05,
+      "loss": 1.0792,
+      "step": 8858
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.636849114845232e-05,
+      "loss": 1.0948,
+      "step": 8859
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6367256519321675e-05,
+      "loss": 1.132,
+      "step": 8860
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.636602180792569e-05,
+      "loss": 1.1562,
+      "step": 8861
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6364787014287637e-05,
+      "loss": 1.1671,
+      "step": 8862
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6363552138430763e-05,
+      "loss": 1.0834,
+      "step": 8863
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6362317180378337e-05,
+      "loss": 1.1609,
+      "step": 8864
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.636108214015362e-05,
+      "loss": 1.1106,
+      "step": 8865
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.635984701777988e-05,
+      "loss": 1.0254,
+      "step": 8866
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6358611813280368e-05,
+      "loss": 1.0851,
+      "step": 8867
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.635737652667837e-05,
+      "loss": 1.1899,
+      "step": 8868
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.635614115799714e-05,
+      "loss": 1.0515,
+      "step": 8869
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6354905707259963e-05,
+      "loss": 1.0815,
+      "step": 8870
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.63536701744901e-05,
+      "loss": 1.0202,
+      "step": 8871
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6352434559710826e-05,
+      "loss": 1.0909,
+      "step": 8872
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6351198862945418e-05,
+      "loss": 1.1146,
+      "step": 8873
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6349963084217155e-05,
+      "loss": 1.1071,
+      "step": 8874
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.634872722354931e-05,
+      "loss": 1.14,
+      "step": 8875
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6347491280965166e-05,
+      "loss": 1.0509,
+      "step": 8876
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6346255256488004e-05,
+      "loss": 1.0557,
+      "step": 8877
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6345019150141105e-05,
+      "loss": 1.0935,
+      "step": 8878
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6343782961947754e-05,
+      "loss": 1.1149,
+      "step": 8879
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6342546691931237e-05,
+      "loss": 1.1041,
+      "step": 8880
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6341310340114848e-05,
+      "loss": 1.2241,
+      "step": 8881
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6340073906521863e-05,
+      "loss": 1.0832,
+      "step": 8882
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.633883739117558e-05,
+      "loss": 1.0517,
+      "step": 8883
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6337600794099294e-05,
+      "loss": 1.1346,
+      "step": 8884
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6336364115316295e-05,
+      "loss": 1.1594,
+      "step": 8885
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6335127354849877e-05,
+      "loss": 1.1442,
+      "step": 8886
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.633389051272334e-05,
+      "loss": 1.1279,
+      "step": 8887
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6332653588959974e-05,
+      "loss": 1.044,
+      "step": 8888
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6331416583583093e-05,
+      "loss": 1.1302,
+      "step": 8889
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6330179496615984e-05,
+      "loss": 1.098,
+      "step": 8890
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.632894232808196e-05,
+      "loss": 1.1007,
+      "step": 8891
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6327705078004318e-05,
+      "loss": 1.1353,
+      "step": 8892
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6326467746406368e-05,
+      "loss": 1.1536,
+      "step": 8893
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6325230333311415e-05,
+      "loss": 1.0814,
+      "step": 8894
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6323992838742775e-05,
+      "loss": 1.0754,
+      "step": 8895
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6322755262723744e-05,
+      "loss": 1.1025,
+      "step": 8896
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.632151760527765e-05,
+      "loss": 1.1025,
+      "step": 8897
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.63202798664278e-05,
+      "loss": 1.0667,
+      "step": 8898
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6319042046197505e-05,
+      "loss": 1.1472,
+      "step": 8899
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6317804144610085e-05,
+      "loss": 1.0045,
+      "step": 8900
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.631656616168886e-05,
+      "loss": 1.0622,
+      "step": 8901
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6315328097457154e-05,
+      "loss": 1.1263,
+      "step": 8902
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6314089951938273e-05,
+      "loss": 1.0664,
+      "step": 8903
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6312851725155555e-05,
+      "loss": 1.0466,
+      "step": 8904
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6311613417132316e-05,
+      "loss": 1.0399,
+      "step": 8905
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6310375027891884e-05,
+      "loss": 1.0846,
+      "step": 8906
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6309136557457585e-05,
+      "loss": 1.0518,
+      "step": 8907
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6307898005852755e-05,
+      "loss": 1.0308,
+      "step": 8908
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6306659373100712e-05,
+      "loss": 1.0496,
+      "step": 8909
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6305420659224804e-05,
+      "loss": 0.9906,
+      "step": 8910
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6304181864248347e-05,
+      "loss": 1.0744,
+      "step": 8911
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.630294298819469e-05,
+      "loss": 1.0691,
+      "step": 8912
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.630170403108716e-05,
+      "loss": 1.1433,
+      "step": 8913
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.63004649929491e-05,
+      "loss": 1.0495,
+      "step": 8914
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6299225873803847e-05,
+      "loss": 1.1504,
+      "step": 8915
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6297986673674746e-05,
+      "loss": 0.9929,
+      "step": 8916
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6296747392585136e-05,
+      "loss": 1.0421,
+      "step": 8917
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6295508030558362e-05,
+      "loss": 1.1165,
+      "step": 8918
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.629426858761777e-05,
+      "loss": 1.1151,
+      "step": 8919
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6293029063786708e-05,
+      "loss": 1.0906,
+      "step": 8920
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6291789459088522e-05,
+      "loss": 1.0031,
+      "step": 8921
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6290549773546564e-05,
+      "loss": 1.0678,
+      "step": 8922
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.628931000718419e-05,
+      "loss": 1.1246,
+      "step": 8923
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6288070160024746e-05,
+      "loss": 1.1002,
+      "step": 8924
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6286830232091588e-05,
+      "loss": 1.0865,
+      "step": 8925
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.628559022340808e-05,
+      "loss": 1.148,
+      "step": 8926
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6284350133997566e-05,
+      "loss": 1.1298,
+      "step": 8927
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.628310996388342e-05,
+      "loss": 1.0467,
+      "step": 8928
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6281869713088994e-05,
+      "loss": 1.0687,
+      "step": 8929
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.628062938163765e-05,
+      "loss": 1.0018,
+      "step": 8930
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.627938896955276e-05,
+      "loss": 1.1385,
+      "step": 8931
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.627814847685768e-05,
+      "loss": 1.1034,
+      "step": 8932
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6276907903575787e-05,
+      "loss": 1.075,
+      "step": 8933
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6275667249730436e-05,
+      "loss": 1.0343,
+      "step": 8934
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6274426515345012e-05,
+      "loss": 1.1401,
+      "step": 8935
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6273185700442875e-05,
+      "loss": 1.0285,
+      "step": 8936
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.62719448050474e-05,
+      "loss": 1.0574,
+      "step": 8937
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.627070382918197e-05,
+      "loss": 1.0543,
+      "step": 8938
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.626946277286995e-05,
+      "loss": 1.0412,
+      "step": 8939
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6268221636134722e-05,
+      "loss": 1.1146,
+      "step": 8940
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6266980418999668e-05,
+      "loss": 1.0591,
+      "step": 8941
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6265739121488166e-05,
+      "loss": 1.0107,
+      "step": 8942
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.62644977436236e-05,
+      "loss": 1.0533,
+      "step": 8943
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6263256285429348e-05,
+      "loss": 1.0995,
+      "step": 8944
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6262014746928797e-05,
+      "loss": 1.1074,
+      "step": 8945
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6260773128145343e-05,
+      "loss": 1.0215,
+      "step": 8946
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6259531429102362e-05,
+      "loss": 1.0754,
+      "step": 8947
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6258289649823253e-05,
+      "loss": 1.075,
+      "step": 8948
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6257047790331395e-05,
+      "loss": 1.1199,
+      "step": 8949
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6255805850650196e-05,
+      "loss": 0.9931,
+      "step": 8950
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6254563830803038e-05,
+      "loss": 1.0307,
+      "step": 8951
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6253321730813325e-05,
+      "loss": 1.0885,
+      "step": 8952
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6252079550704446e-05,
+      "loss": 1.0764,
+      "step": 8953
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.625083729049981e-05,
+      "loss": 1.0843,
+      "step": 8954
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.624959495022281e-05,
+      "loss": 1.1113,
+      "step": 8955
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6248352529896846e-05,
+      "loss": 1.0398,
+      "step": 8956
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.624711002954533e-05,
+      "loss": 1.0786,
+      "step": 8957
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.624586744919166e-05,
+      "loss": 1.1089,
+      "step": 8958
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6244624788859242e-05,
+      "loss": 1.0604,
+      "step": 8959
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6243382048571487e-05,
+      "loss": 1.017,
+      "step": 8960
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.62421392283518e-05,
+      "loss": 1.0646,
+      "step": 8961
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6240896328223598e-05,
+      "loss": 1.0854,
+      "step": 8962
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.623965334821029e-05,
+      "loss": 1.1055,
+      "step": 8963
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6238410288335288e-05,
+      "loss": 1.0888,
+      "step": 8964
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.623716714862201e-05,
+      "loss": 1.1333,
+      "step": 8965
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6235923929093873e-05,
+      "loss": 1.1796,
+      "step": 8966
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6234680629774296e-05,
+      "loss": 1.0809,
+      "step": 8967
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6233437250686696e-05,
+      "loss": 1.0797,
+      "step": 8968
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6232193791854495e-05,
+      "loss": 1.109,
+      "step": 8969
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6230950253301114e-05,
+      "loss": 1.0701,
+      "step": 8970
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6229706635049984e-05,
+      "loss": 1.1076,
+      "step": 8971
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6228462937124522e-05,
+      "loss": 1.0286,
+      "step": 8972
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6227219159548164e-05,
+      "loss": 1.0461,
+      "step": 8973
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6225975302344333e-05,
+      "loss": 1.0837,
+      "step": 8974
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6224731365536464e-05,
+      "loss": 1.116,
+      "step": 8975
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.622348734914798e-05,
+      "loss": 0.9916,
+      "step": 8976
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6222243253202325e-05,
+      "loss": 1.0587,
+      "step": 8977
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6220999077722927e-05,
+      "loss": 1.1928,
+      "step": 8978
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6219754822733226e-05,
+      "loss": 1.1268,
+      "step": 8979
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6218510488256657e-05,
+      "loss": 1.076,
+      "step": 8980
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6217266074316663e-05,
+      "loss": 0.9695,
+      "step": 8981
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6216021580936682e-05,
+      "loss": 1.0638,
+      "step": 8982
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6214777008140156e-05,
+      "loss": 1.0529,
+      "step": 8983
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.621353235595053e-05,
+      "loss": 1.1207,
+      "step": 8984
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6212287624391247e-05,
+      "loss": 1.1604,
+      "step": 8985
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.621104281348576e-05,
+      "loss": 0.992,
+      "step": 8986
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6209797923257507e-05,
+      "loss": 1.1109,
+      "step": 8987
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.620855295372995e-05,
+      "loss": 1.1152,
+      "step": 8988
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.620730790492653e-05,
+      "loss": 1.097,
+      "step": 8989
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6206062776870708e-05,
+      "loss": 1.0972,
+      "step": 8990
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6204817569585927e-05,
+      "loss": 1.1643,
+      "step": 8991
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.620357228309566e-05,
+      "loss": 1.0239,
+      "step": 8992
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6202326917423344e-05,
+      "loss": 1.1158,
+      "step": 8993
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6201081472592456e-05,
+      "loss": 1.1374,
+      "step": 8994
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6199835948626444e-05,
+      "loss": 1.134,
+      "step": 8995
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.619859034554878e-05,
+      "loss": 1.0869,
+      "step": 8996
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6197344663382915e-05,
+      "loss": 1.0516,
+      "step": 8997
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6196098902152322e-05,
+      "loss": 1.1257,
+      "step": 8998
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6194853061880464e-05,
+      "loss": 1.0794,
+      "step": 8999
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6193607142590813e-05,
+      "loss": 1.0483,
+      "step": 9000
+    },
+    {
+      "epoch": 0.38,
+      "eval_text_loss": 0.8115671277046204,
+      "eval_text_runtime": 1.3986,
+      "eval_text_samples_per_second": 52.911,
+      "eval_text_steps_per_second": 1.43,
+      "step": 9000
+    },
+    {
+      "epoch": 0.38,
+      "eval_svit_loss": 1.1737525463104248,
+      "eval_svit_runtime": 5.1438,
+      "eval_svit_samples_per_second": 132.391,
+      "eval_svit_steps_per_second": 2.138,
+      "step": 9000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6192361144306832e-05,
+      "loss": 1.0985,
+      "step": 9001
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6191115067051998e-05,
+      "loss": 1.0374,
+      "step": 9002
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.618986891084978e-05,
+      "loss": 1.1634,
+      "step": 9003
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6188622675723647e-05,
+      "loss": 1.1057,
+      "step": 9004
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6187376361697083e-05,
+      "loss": 1.0725,
+      "step": 9005
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6186129968793557e-05,
+      "loss": 1.1115,
+      "step": 9006
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6184883497036553e-05,
+      "loss": 1.099,
+      "step": 9007
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6183636946449547e-05,
+      "loss": 1.1165,
+      "step": 9008
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.618239031705602e-05,
+      "loss": 1.1149,
+      "step": 9009
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.618114360887946e-05,
+      "loss": 1.0925,
+      "step": 9010
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6179896821943343e-05,
+      "loss": 1.0569,
+      "step": 9011
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6178649956271163e-05,
+      "loss": 1.1319,
+      "step": 9012
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6177403011886397e-05,
+      "loss": 1.0994,
+      "step": 9013
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.617615598881254e-05,
+      "loss": 1.074,
+      "step": 9014
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6174908887073084e-05,
+      "loss": 1.1576,
+      "step": 9015
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6173661706691515e-05,
+      "loss": 0.97,
+      "step": 9016
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.617241444769133e-05,
+      "loss": 1.084,
+      "step": 9017
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6171167110096016e-05,
+      "loss": 1.1467,
+      "step": 9018
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6169919693929083e-05,
+      "loss": 1.0989,
+      "step": 9019
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.616867219921401e-05,
+      "loss": 1.1094,
+      "step": 9020
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6167424625974315e-05,
+      "loss": 1.1027,
+      "step": 9021
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6166176974233483e-05,
+      "loss": 1.0606,
+      "step": 9022
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6164929244015028e-05,
+      "loss": 1.0735,
+      "step": 9023
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.616368143534244e-05,
+      "loss": 1.0823,
+      "step": 9024
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6162433548239236e-05,
+      "loss": 1.0853,
+      "step": 9025
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6161185582728912e-05,
+      "loss": 1.0889,
+      "step": 9026
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.615993753883499e-05,
+      "loss": 1.1232,
+      "step": 9027
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6158689416580966e-05,
+      "loss": 1.0891,
+      "step": 9028
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.615744121599035e-05,
+      "loss": 1.0725,
+      "step": 9029
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6156192937086666e-05,
+      "loss": 1.0771,
+      "step": 9030
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6154944579893417e-05,
+      "loss": 1.0601,
+      "step": 9031
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6153696144434123e-05,
+      "loss": 1.0444,
+      "step": 9032
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6152447630732296e-05,
+      "loss": 1.1421,
+      "step": 9033
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6151199038811462e-05,
+      "loss": 1.0767,
+      "step": 9034
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6149950368695135e-05,
+      "loss": 1.0363,
+      "step": 9035
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.614870162040684e-05,
+      "loss": 1.1132,
+      "step": 9036
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.614745279397009e-05,
+      "loss": 1.1099,
+      "step": 9037
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.614620388940842e-05,
+      "loss": 1.0777,
+      "step": 9038
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.614495490674535e-05,
+      "loss": 1.1463,
+      "step": 9039
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6143705846004405e-05,
+      "loss": 1.1066,
+      "step": 9040
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6142456707209122e-05,
+      "loss": 1.0774,
+      "step": 9041
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.614120749038302e-05,
+      "loss": 1.0389,
+      "step": 9042
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6139958195549634e-05,
+      "loss": 1.1,
+      "step": 9043
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.61387088227325e-05,
+      "loss": 1.0517,
+      "step": 9044
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6137459371955153e-05,
+      "loss": 1.0464,
+      "step": 9045
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6136209843241124e-05,
+      "loss": 1.1347,
+      "step": 9046
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6134960236613952e-05,
+      "loss": 1.0703,
+      "step": 9047
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6133710552097175e-05,
+      "loss": 0.9958,
+      "step": 9048
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6132460789714335e-05,
+      "loss": 1.1379,
+      "step": 9049
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.613121094948897e-05,
+      "loss": 1.1062,
+      "step": 9050
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6129961031444627e-05,
+      "loss": 1.0291,
+      "step": 9051
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.612871103560485e-05,
+      "loss": 1.1099,
+      "step": 9052
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6127460961993185e-05,
+      "loss": 1.0527,
+      "step": 9053
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6126210810633176e-05,
+      "loss": 1.0393,
+      "step": 9054
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6124960581548376e-05,
+      "loss": 1.0207,
+      "step": 9055
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6123710274762333e-05,
+      "loss": 1.1084,
+      "step": 9056
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6122459890298602e-05,
+      "loss": 1.0709,
+      "step": 9057
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.612120942818073e-05,
+      "loss": 1.0594,
+      "step": 9058
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6119958888432282e-05,
+      "loss": 1.1155,
+      "step": 9059
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6118708271076802e-05,
+      "loss": 1.0348,
+      "step": 9060
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6117457576137853e-05,
+      "loss": 1.0142,
+      "step": 9061
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6116206803639e-05,
+      "loss": 1.1107,
+      "step": 9062
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6114955953603794e-05,
+      "loss": 1.0893,
+      "step": 9063
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.61137050260558e-05,
+      "loss": 1.0158,
+      "step": 9064
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6112454021018587e-05,
+      "loss": 1.0524,
+      "step": 9065
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6111202938515712e-05,
+      "loss": 1.1341,
+      "step": 9066
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.610995177857075e-05,
+      "loss": 1.0902,
+      "step": 9067
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.610870054120726e-05,
+      "loss": 1.1051,
+      "step": 9068
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.610744922644882e-05,
+      "loss": 1.1378,
+      "step": 9069
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6106197834318993e-05,
+      "loss": 1.0388,
+      "step": 9070
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6104946364841358e-05,
+      "loss": 1.0981,
+      "step": 9071
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6103694818039484e-05,
+      "loss": 1.1977,
+      "step": 9072
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6102443193936943e-05,
+      "loss": 1.0778,
+      "step": 9073
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.610119149255732e-05,
+      "loss": 1.0526,
+      "step": 9074
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6099939713924188e-05,
+      "loss": 1.0502,
+      "step": 9075
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.609868785806113e-05,
+      "loss": 1.0934,
+      "step": 9076
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6097435924991724e-05,
+      "loss": 1.0321,
+      "step": 9077
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6096183914739554e-05,
+      "loss": 1.0672,
+      "step": 9078
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6094931827328202e-05,
+      "loss": 1.0824,
+      "step": 9079
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6093679662781256e-05,
+      "loss": 1.108,
+      "step": 9080
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6092427421122304e-05,
+      "loss": 1.1602,
+      "step": 9081
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6091175102374927e-05,
+      "loss": 1.0723,
+      "step": 9082
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.608992270656272e-05,
+      "loss": 1.0322,
+      "step": 9083
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6088670233709277e-05,
+      "loss": 1.0829,
+      "step": 9084
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6087417683838187e-05,
+      "loss": 1.0708,
+      "step": 9085
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6086165056973045e-05,
+      "loss": 1.0453,
+      "step": 9086
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6084912353137443e-05,
+      "loss": 1.0881,
+      "step": 9087
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6083659572354983e-05,
+      "loss": 1.0616,
+      "step": 9088
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6082406714649263e-05,
+      "loss": 1.0902,
+      "step": 9089
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.608115378004388e-05,
+      "loss": 1.1052,
+      "step": 9090
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6079900768562434e-05,
+      "loss": 1.129,
+      "step": 9091
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6078647680228532e-05,
+      "loss": 0.9999,
+      "step": 9092
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6077394515065776e-05,
+      "loss": 1.1645,
+      "step": 9093
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6076141273097773e-05,
+      "loss": 1.1091,
+      "step": 9094
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.607488795434813e-05,
+      "loss": 1.0991,
+      "step": 9095
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6073634558840454e-05,
+      "loss": 1.0516,
+      "step": 9096
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.607238108659836e-05,
+      "loss": 1.0372,
+      "step": 9097
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6071127537645452e-05,
+      "loss": 1.0693,
+      "step": 9098
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6069873912005348e-05,
+      "loss": 1.0149,
+      "step": 9099
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.606862020970166e-05,
+      "loss": 1.0134,
+      "step": 9100
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6067366430758004e-05,
+      "loss": 1.0509,
+      "step": 9101
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6066112575197995e-05,
+      "loss": 1.0626,
+      "step": 9102
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6064858643045263e-05,
+      "loss": 1.0444,
+      "step": 9103
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6063604634323412e-05,
+      "loss": 1.1532,
+      "step": 9104
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6062350549056074e-05,
+      "loss": 1.1067,
+      "step": 9105
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.606109638726687e-05,
+      "loss": 1.0651,
+      "step": 9106
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6059842148979424e-05,
+      "loss": 1.12,
+      "step": 9107
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6058587834217356e-05,
+      "loss": 1.123,
+      "step": 9108
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6057333443004306e-05,
+      "loss": 1.1151,
+      "step": 9109
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.605607897536389e-05,
+      "loss": 1.0016,
+      "step": 9110
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.605482443131975e-05,
+      "loss": 1.0818,
+      "step": 9111
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.605356981089551e-05,
+      "loss": 1.0792,
+      "step": 9112
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6052315114114803e-05,
+      "loss": 1.1937,
+      "step": 9113
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6051060341001265e-05,
+      "loss": 1.1597,
+      "step": 9114
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.604980549157853e-05,
+      "loss": 1.0225,
+      "step": 9115
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.604855056587024e-05,
+      "loss": 1.0123,
+      "step": 9116
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.604729556390003e-05,
+      "loss": 1.1359,
+      "step": 9117
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6046040485691543e-05,
+      "loss": 1.0327,
+      "step": 9118
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6044785331268417e-05,
+      "loss": 1.1212,
+      "step": 9119
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6043530100654303e-05,
+      "loss": 1.0739,
+      "step": 9120
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6042274793872834e-05,
+      "loss": 1.0279,
+      "step": 9121
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6041019410947667e-05,
+      "loss": 1.172,
+      "step": 9122
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.603976395190244e-05,
+      "loss": 1.0466,
+      "step": 9123
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6038508416760807e-05,
+      "loss": 1.1409,
+      "step": 9124
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6037252805546424e-05,
+      "loss": 1.0866,
+      "step": 9125
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.603599711828293e-05,
+      "loss": 1.0923,
+      "step": 9126
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6034741354993988e-05,
+      "loss": 1.0411,
+      "step": 9127
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6033485515703244e-05,
+      "loss": 1.0706,
+      "step": 9128
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6032229600434363e-05,
+      "loss": 1.0715,
+      "step": 9129
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6030973609211e-05,
+      "loss": 1.0039,
+      "step": 9130
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6029717542056813e-05,
+      "loss": 1.0721,
+      "step": 9131
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.602846139899546e-05,
+      "loss": 1.0211,
+      "step": 9132
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6027205180050607e-05,
+      "loss": 1.1445,
+      "step": 9133
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6025948885245914e-05,
+      "loss": 1.0788,
+      "step": 9134
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.602469251460505e-05,
+      "loss": 1.1252,
+      "step": 9135
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6023436068151673e-05,
+      "loss": 1.0773,
+      "step": 9136
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6022179545909455e-05,
+      "loss": 1.0892,
+      "step": 9137
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6020922947902067e-05,
+      "loss": 1.0582,
+      "step": 9138
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.601966627415318e-05,
+      "loss": 1.1007,
+      "step": 9139
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.601840952468646e-05,
+      "loss": 1.1179,
+      "step": 9140
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6017152699525583e-05,
+      "loss": 1.1068,
+      "step": 9141
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6015895798694224e-05,
+      "loss": 1.1432,
+      "step": 9142
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6014638822216062e-05,
+      "loss": 1.0917,
+      "step": 9143
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.601338177011477e-05,
+      "loss": 1.1151,
+      "step": 9144
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.601212464241403e-05,
+      "loss": 1.0786,
+      "step": 9145
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6010867439137515e-05,
+      "loss": 1.1171,
+      "step": 9146
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.600961016030892e-05,
+      "loss": 1.1426,
+      "step": 9147
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6008352805951917e-05,
+      "loss": 1.1194,
+      "step": 9148
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6007095376090195e-05,
+      "loss": 1.0792,
+      "step": 9149
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.600583787074744e-05,
+      "loss": 1.036,
+      "step": 9150
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6004580289947338e-05,
+      "loss": 1.1034,
+      "step": 9151
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6003322633713578e-05,
+      "loss": 1.0821,
+      "step": 9152
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6002064902069854e-05,
+      "loss": 1.0701,
+      "step": 9153
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.600080709503985e-05,
+      "loss": 1.1192,
+      "step": 9154
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5999549212647268e-05,
+      "loss": 1.1959,
+      "step": 9155
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5998291254915796e-05,
+      "loss": 1.0849,
+      "step": 9156
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5997033221869133e-05,
+      "loss": 1.1377,
+      "step": 9157
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5995775113530975e-05,
+      "loss": 1.0878,
+      "step": 9158
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5994516929925022e-05,
+      "loss": 1.166,
+      "step": 9159
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.599325867107497e-05,
+      "loss": 1.1089,
+      "step": 9160
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.599200033700453e-05,
+      "loss": 1.0915,
+      "step": 9161
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5990741927737394e-05,
+      "loss": 1.1387,
+      "step": 9162
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5989483443297275e-05,
+      "loss": 1.1567,
+      "step": 9163
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5988224883707874e-05,
+      "loss": 1.0902,
+      "step": 9164
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5986966248992902e-05,
+      "loss": 1.0411,
+      "step": 9165
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5985707539176062e-05,
+      "loss": 1.0851,
+      "step": 9166
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.598444875428107e-05,
+      "loss": 1.1152,
+      "step": 9167
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5983189894331636e-05,
+      "loss": 1.0875,
+      "step": 9168
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.598193095935147e-05,
+      "loss": 1.1415,
+      "step": 9169
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.598067194936429e-05,
+      "loss": 1.1108,
+      "step": 9170
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.597941286439381e-05,
+      "loss": 1.0986,
+      "step": 9171
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.597815370446375e-05,
+      "loss": 1.1324,
+      "step": 9172
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5976894469597824e-05,
+      "loss": 1.1333,
+      "step": 9173
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5975635159819758e-05,
+      "loss": 1.0577,
+      "step": 9174
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.597437577515326e-05,
+      "loss": 1.0945,
+      "step": 9175
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.597311631562207e-05,
+      "loss": 1.0961,
+      "step": 9176
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.59718567812499e-05,
+      "loss": 1.0547,
+      "step": 9177
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5970597172060484e-05,
+      "loss": 1.0839,
+      "step": 9178
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5969337488077543e-05,
+      "loss": 1.1914,
+      "step": 9179
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5968077729324808e-05,
+      "loss": 1.0305,
+      "step": 9180
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.596681789582601e-05,
+      "loss": 1.1055,
+      "step": 9181
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.596555798760488e-05,
+      "loss": 1.0659,
+      "step": 9182
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.596429800468515e-05,
+      "loss": 1.1074,
+      "step": 9183
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.596303794709055e-05,
+      "loss": 1.1076,
+      "step": 9184
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5961777814844825e-05,
+      "loss": 1.0715,
+      "step": 9185
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.59605176079717e-05,
+      "loss": 1.0983,
+      "step": 9186
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5959257326494925e-05,
+      "loss": 1.1154,
+      "step": 9187
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5957996970438225e-05,
+      "loss": 1.0896,
+      "step": 9188
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.595673653982536e-05,
+      "loss": 1.1649,
+      "step": 9189
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5955476034680057e-05,
+      "loss": 1.1488,
+      "step": 9190
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5954215455026074e-05,
+      "loss": 1.1003,
+      "step": 9191
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5952954800887136e-05,
+      "loss": 1.0406,
+      "step": 9192
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5951694072287013e-05,
+      "loss": 1.0598,
+      "step": 9193
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5950433269249435e-05,
+      "loss": 1.0607,
+      "step": 9194
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5949172391798166e-05,
+      "loss": 1.1121,
+      "step": 9195
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5947911439956943e-05,
+      "loss": 1.0906,
+      "step": 9196
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5946650413749527e-05,
+      "loss": 1.0405,
+      "step": 9197
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.594538931319967e-05,
+      "loss": 1.138,
+      "step": 9198
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5944128138331124e-05,
+      "loss": 1.0303,
+      "step": 9199
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5942866889167652e-05,
+      "loss": 1.0688,
+      "step": 9200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5941605565733008e-05,
+      "loss": 1.0882,
+      "step": 9201
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.594034416805095e-05,
+      "loss": 1.0894,
+      "step": 9202
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5939082696145244e-05,
+      "loss": 1.078,
+      "step": 9203
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5937821150039644e-05,
+      "loss": 1.0304,
+      "step": 9204
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.593655952975792e-05,
+      "loss": 1.0717,
+      "step": 9205
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5935297835323836e-05,
+      "loss": 1.0173,
+      "step": 9206
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.593403606676116e-05,
+      "loss": 0.9965,
+      "step": 9207
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5932774224093652e-05,
+      "loss": 1.0498,
+      "step": 9208
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.593151230734509e-05,
+      "loss": 1.0176,
+      "step": 9209
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5930250316539237e-05,
+      "loss": 1.1347,
+      "step": 9210
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5928988251699872e-05,
+      "loss": 1.1157,
+      "step": 9211
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5927726112850763e-05,
+      "loss": 1.0599,
+      "step": 9212
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.592646390001569e-05,
+      "loss": 1.1174,
+      "step": 9213
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5925201613218424e-05,
+      "loss": 1.0598,
+      "step": 9214
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5923939252482742e-05,
+      "loss": 1.0667,
+      "step": 9215
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5922676817832432e-05,
+      "loss": 1.1029,
+      "step": 9216
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5921414309291263e-05,
+      "loss": 1.0764,
+      "step": 9217
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.592015172688302e-05,
+      "loss": 1.0681,
+      "step": 9218
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.591888907063149e-05,
+      "loss": 1.0861,
+      "step": 9219
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5917626340560455e-05,
+      "loss": 1.0353,
+      "step": 9220
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5916363536693697e-05,
+      "loss": 1.0411,
+      "step": 9221
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.591510065905501e-05,
+      "loss": 1.0561,
+      "step": 9222
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5913837707668178e-05,
+      "loss": 1.0627,
+      "step": 9223
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5912574682556994e-05,
+      "loss": 1.0893,
+      "step": 9224
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.591131158374525e-05,
+      "loss": 1.0826,
+      "step": 9225
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5910048411256735e-05,
+      "loss": 1.0397,
+      "step": 9226
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5908785165115244e-05,
+      "loss": 1.002,
+      "step": 9227
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.590752184534457e-05,
+      "loss": 1.0659,
+      "step": 9228
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5906258451968523e-05,
+      "loss": 1.081,
+      "step": 9229
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5904994985010882e-05,
+      "loss": 1.0288,
+      "step": 9230
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5903731444495463e-05,
+      "loss": 1.0754,
+      "step": 9231
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5902467830446055e-05,
+      "loss": 1.1048,
+      "step": 9232
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5901204142886473e-05,
+      "loss": 1.0722,
+      "step": 9233
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.589994038184051e-05,
+      "loss": 1.1163,
+      "step": 9234
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5898676547331977e-05,
+      "loss": 1.0887,
+      "step": 9235
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5897412639384674e-05,
+      "loss": 1.0976,
+      "step": 9236
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.589614865802242e-05,
+      "loss": 0.9956,
+      "step": 9237
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5894884603269017e-05,
+      "loss": 1.0027,
+      "step": 9238
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5893620475148276e-05,
+      "loss": 1.1405,
+      "step": 9239
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5892356273684013e-05,
+      "loss": 1.2178,
+      "step": 9240
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.589109199890004e-05,
+      "loss": 1.1291,
+      "step": 9241
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5889827650820166e-05,
+      "loss": 1.0634,
+      "step": 9242
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5888563229468213e-05,
+      "loss": 1.1157,
+      "step": 9243
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5887298734868e-05,
+      "loss": 1.1002,
+      "step": 9244
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.588603416704334e-05,
+      "loss": 1.0113,
+      "step": 9245
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5884769526018064e-05,
+      "loss": 1.085,
+      "step": 9246
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5883504811815983e-05,
+      "loss": 1.1083,
+      "step": 9247
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5882240024460927e-05,
+      "loss": 1.0971,
+      "step": 9248
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5880975163976716e-05,
+      "loss": 1.0757,
+      "step": 9249
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5879710230387182e-05,
+      "loss": 1.0031,
+      "step": 9250
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5878445223716143e-05,
+      "loss": 1.1168,
+      "step": 9251
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5877180143987435e-05,
+      "loss": 1.1248,
+      "step": 9252
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5875914991224887e-05,
+      "loss": 1.1029,
+      "step": 9253
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5874649765452334e-05,
+      "loss": 1.1482,
+      "step": 9254
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.58733844666936e-05,
+      "loss": 1.1069,
+      "step": 9255
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5872119094972524e-05,
+      "loss": 1.0651,
+      "step": 9256
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5870853650312942e-05,
+      "loss": 1.1653,
+      "step": 9257
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5869588132738698e-05,
+      "loss": 1.1445,
+      "step": 9258
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5868322542273612e-05,
+      "loss": 0.9877,
+      "step": 9259
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.586705687894154e-05,
+      "loss": 1.1088,
+      "step": 9260
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5865791142766318e-05,
+      "loss": 1.1282,
+      "step": 9261
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5864525333771787e-05,
+      "loss": 1.0004,
+      "step": 9262
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.586325945198179e-05,
+      "loss": 1.046,
+      "step": 9263
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5861993497420175e-05,
+      "loss": 1.0176,
+      "step": 9264
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.586072747011079e-05,
+      "loss": 1.1198,
+      "step": 9265
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.585946137007748e-05,
+      "loss": 1.0957,
+      "step": 9266
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5858195197344097e-05,
+      "loss": 1.0655,
+      "step": 9267
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.585692895193449e-05,
+      "loss": 1.0757,
+      "step": 9268
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.585566263387251e-05,
+      "loss": 1.0281,
+      "step": 9269
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5854396243182008e-05,
+      "loss": 1.0891,
+      "step": 9270
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5853129779886848e-05,
+      "loss": 1.0283,
+      "step": 9271
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.585186324401087e-05,
+      "loss": 1.1152,
+      "step": 9272
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5850596635577954e-05,
+      "loss": 1.0456,
+      "step": 9273
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5849329954611942e-05,
+      "loss": 1.1029,
+      "step": 9274
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5848063201136698e-05,
+      "loss": 1.1096,
+      "step": 9275
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5846796375176084e-05,
+      "loss": 1.1548,
+      "step": 9276
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5845529476753967e-05,
+      "loss": 1.1369,
+      "step": 9277
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5844262505894205e-05,
+      "loss": 1.0366,
+      "step": 9278
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.584299546262067e-05,
+      "loss": 1.0095,
+      "step": 9279
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5841728346957226e-05,
+      "loss": 1.0906,
+      "step": 9280
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.584046115892774e-05,
+      "loss": 1.0643,
+      "step": 9281
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5839193898556084e-05,
+      "loss": 1.0302,
+      "step": 9282
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5837926565866128e-05,
+      "loss": 1.13,
+      "step": 9283
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5836659160881743e-05,
+      "loss": 1.0162,
+      "step": 9284
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5835391683626807e-05,
+      "loss": 1.1577,
+      "step": 9285
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.583412413412519e-05,
+      "loss": 1.0465,
+      "step": 9286
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5832856512400776e-05,
+      "loss": 1.0761,
+      "step": 9287
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5831588818477437e-05,
+      "loss": 1.0833,
+      "step": 9288
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5830321052379054e-05,
+      "loss": 1.0486,
+      "step": 9289
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5829053214129508e-05,
+      "loss": 1.1157,
+      "step": 9290
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5827785303752683e-05,
+      "loss": 1.0443,
+      "step": 9291
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5826517321272456e-05,
+      "loss": 1.043,
+      "step": 9292
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.582524926671272e-05,
+      "loss": 1.0179,
+      "step": 9293
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5823981140097353e-05,
+      "loss": 1.1249,
+      "step": 9294
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.582271294145025e-05,
+      "loss": 1.0897,
+      "step": 9295
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5821444670795296e-05,
+      "loss": 1.042,
+      "step": 9296
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5820176328156382e-05,
+      "loss": 1.1096,
+      "step": 9297
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.58189079135574e-05,
+      "loss": 1.0446,
+      "step": 9298
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.581763942702224e-05,
+      "loss": 1.049,
+      "step": 9299
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5816370868574806e-05,
+      "loss": 1.1139,
+      "step": 9300
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5815102238238976e-05,
+      "loss": 1.0654,
+      "step": 9301
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5813833536038663e-05,
+      "loss": 1.0657,
+      "step": 9302
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5812564761997758e-05,
+      "loss": 1.0609,
+      "step": 9303
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5811295916140165e-05,
+      "loss": 1.0563,
+      "step": 9304
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5810026998489777e-05,
+      "loss": 1.0533,
+      "step": 9305
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5808758009070505e-05,
+      "loss": 1.1014,
+      "step": 9306
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5807488947906248e-05,
+      "loss": 1.1537,
+      "step": 9307
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5806219815020916e-05,
+      "loss": 1.0728,
+      "step": 9308
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5804950610438412e-05,
+      "loss": 0.9823,
+      "step": 9309
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.580368133418264e-05,
+      "loss": 1.0448,
+      "step": 9310
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5802411986277517e-05,
+      "loss": 1.1003,
+      "step": 9311
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5801142566746945e-05,
+      "loss": 1.1098,
+      "step": 9312
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.579987307561485e-05,
+      "loss": 1.0393,
+      "step": 9313
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5798603512905124e-05,
+      "loss": 1.0282,
+      "step": 9314
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.57973338786417e-05,
+      "loss": 1.1393,
+      "step": 9315
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.579606417284848e-05,
+      "loss": 1.1441,
+      "step": 9316
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5794794395549397e-05,
+      "loss": 1.1195,
+      "step": 9317
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.579352454676836e-05,
+      "loss": 1.063,
+      "step": 9318
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5792254626529287e-05,
+      "loss": 1.0813,
+      "step": 9319
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.57909846348561e-05,
+      "loss": 1.0517,
+      "step": 9320
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.578971457177273e-05,
+      "loss": 1.0115,
+      "step": 9321
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.578844443730309e-05,
+      "loss": 1.1175,
+      "step": 9322
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5787174231471112e-05,
+      "loss": 1.0582,
+      "step": 9323
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.578590395430072e-05,
+      "loss": 1.1034,
+      "step": 9324
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.578463360581584e-05,
+      "loss": 1.0659,
+      "step": 9325
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.578336318604041e-05,
+      "loss": 1.0563,
+      "step": 9326
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5782092694998354e-05,
+      "loss": 1.0913,
+      "step": 9327
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.57808221327136e-05,
+      "loss": 1.0759,
+      "step": 9328
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5779551499210092e-05,
+      "loss": 1.1298,
+      "step": 9329
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5778280794511754e-05,
+      "loss": 1.1061,
+      "step": 9330
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.577701001864253e-05,
+      "loss": 1.1082,
+      "step": 9331
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5775739171626353e-05,
+      "loss": 1.033,
+      "step": 9332
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5774468253487164e-05,
+      "loss": 1.1408,
+      "step": 9333
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5773197264248904e-05,
+      "loss": 1.0285,
+      "step": 9334
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.577192620393551e-05,
+      "loss": 1.0897,
+      "step": 9335
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5770655072570928e-05,
+      "loss": 1.0515,
+      "step": 9336
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5769383870179104e-05,
+      "loss": 1.0751,
+      "step": 9337
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.576811259678398e-05,
+      "loss": 1.0865,
+      "step": 9338
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.57668412524095e-05,
+      "loss": 1.0351,
+      "step": 9339
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.576556983707962e-05,
+      "loss": 1.0498,
+      "step": 9340
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5764298350818287e-05,
+      "loss": 1.0459,
+      "step": 9341
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.576302679364945e-05,
+      "loss": 0.9906,
+      "step": 9342
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5761755165597058e-05,
+      "loss": 1.0891,
+      "step": 9343
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.576048346668507e-05,
+      "loss": 1.0186,
+      "step": 9344
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5759211696937435e-05,
+      "loss": 1.0978,
+      "step": 9345
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.575793985637812e-05,
+      "loss": 1.0545,
+      "step": 9346
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.575666794503107e-05,
+      "loss": 1.0064,
+      "step": 9347
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.575539596292025e-05,
+      "loss": 1.0322,
+      "step": 9348
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.575412391006962e-05,
+      "loss": 1.03,
+      "step": 9349
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5752851786503138e-05,
+      "loss": 1.0562,
+      "step": 9350
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.575157959224477e-05,
+      "loss": 1.0557,
+      "step": 9351
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.575030732731848e-05,
+      "loss": 1.086,
+      "step": 9352
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5749034991748232e-05,
+      "loss": 1.1171,
+      "step": 9353
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5747762585557994e-05,
+      "loss": 1.0009,
+      "step": 9354
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5746490108771738e-05,
+      "loss": 1.086,
+      "step": 9355
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5745217561413423e-05,
+      "loss": 1.0048,
+      "step": 9356
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5743944943507032e-05,
+      "loss": 1.116,
+      "step": 9357
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.574267225507653e-05,
+      "loss": 1.0936,
+      "step": 9358
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.574139949614589e-05,
+      "loss": 1.0294,
+      "step": 9359
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5740126666739087e-05,
+      "loss": 1.0226,
+      "step": 9360
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.57388537668801e-05,
+      "loss": 1.1057,
+      "step": 9361
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5737580796592906e-05,
+      "loss": 1.0579,
+      "step": 9362
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5736307755901482e-05,
+      "loss": 1.0634,
+      "step": 9363
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5735034644829805e-05,
+      "loss": 1.0838,
+      "step": 9364
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5733761463401864e-05,
+      "loss": 1.0681,
+      "step": 9365
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5732488211641637e-05,
+      "loss": 1.0639,
+      "step": 9366
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.573121488957311e-05,
+      "loss": 1.0943,
+      "step": 9367
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.572994149722027e-05,
+      "loss": 1.0981,
+      "step": 9368
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5728668034607093e-05,
+      "loss": 1.0569,
+      "step": 9369
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.572739450175758e-05,
+      "loss": 1.0482,
+      "step": 9370
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5726120898695716e-05,
+      "loss": 1.0903,
+      "step": 9371
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5724847225445487e-05,
+      "loss": 1.0587,
+      "step": 9372
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.572357348203089e-05,
+      "loss": 1.1598,
+      "step": 9373
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.572229966847592e-05,
+      "loss": 1.1339,
+      "step": 9374
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5721025784804566e-05,
+      "loss": 1.0531,
+      "step": 9375
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.571975183104083e-05,
+      "loss": 1.0796,
+      "step": 9376
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5718477807208706e-05,
+      "loss": 1.1083,
+      "step": 9377
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.571720371333219e-05,
+      "loss": 1.0352,
+      "step": 9378
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5715929549435286e-05,
+      "loss": 1.1262,
+      "step": 9379
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5714655315541993e-05,
+      "loss": 1.0512,
+      "step": 9380
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5713381011676314e-05,
+      "loss": 1.0401,
+      "step": 9381
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5712106637862255e-05,
+      "loss": 0.9694,
+      "step": 9382
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5710832194123817e-05,
+      "loss": 1.1181,
+      "step": 9383
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5709557680485015e-05,
+      "loss": 1.0941,
+      "step": 9384
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.570828309696984e-05,
+      "loss": 1.0684,
+      "step": 9385
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5707008443602324e-05,
+      "loss": 1.0887,
+      "step": 9386
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5705733720406456e-05,
+      "loss": 1.1249,
+      "step": 9387
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.570445892740626e-05,
+      "loss": 1.0853,
+      "step": 9388
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5703184064625746e-05,
+      "loss": 1.1303,
+      "step": 9389
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.570190913208893e-05,
+      "loss": 1.0846,
+      "step": 9390
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5700634129819827e-05,
+      "loss": 1.0674,
+      "step": 9391
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5699359057842452e-05,
+      "loss": 1.086,
+      "step": 9392
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5698083916180822e-05,
+      "loss": 1.0591,
+      "step": 9393
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5696808704858966e-05,
+      "loss": 1.0935,
+      "step": 9394
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.569553342390089e-05,
+      "loss": 1.1436,
+      "step": 9395
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5694258073330625e-05,
+      "loss": 1.0654,
+      "step": 9396
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.56929826531722e-05,
+      "loss": 1.0846,
+      "step": 9397
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.569170716344963e-05,
+      "loss": 1.0416,
+      "step": 9398
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.569043160418695e-05,
+      "loss": 1.0803,
+      "step": 9399
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5689155975408176e-05,
+      "loss": 1.065,
+      "step": 9400
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.568788027713735e-05,
+      "loss": 1.0795,
+      "step": 9401
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.568660450939849e-05,
+      "loss": 1.0192,
+      "step": 9402
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5685328672215634e-05,
+      "loss": 1.0618,
+      "step": 9403
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5684052765612813e-05,
+      "loss": 0.9827,
+      "step": 9404
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5682776789614067e-05,
+      "loss": 1.1131,
+      "step": 9405
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5681500744243424e-05,
+      "loss": 1.0367,
+      "step": 9406
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.568022462952492e-05,
+      "loss": 1.0044,
+      "step": 9407
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5678948445482597e-05,
+      "loss": 1.0794,
+      "step": 9408
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5677672192140494e-05,
+      "loss": 1.1442,
+      "step": 9409
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5676395869522652e-05,
+      "loss": 1.1228,
+      "step": 9410
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.567511947765311e-05,
+      "loss": 1.0827,
+      "step": 9411
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5673843016555913e-05,
+      "loss": 1.1592,
+      "step": 9412
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5672566486255106e-05,
+      "loss": 0.9951,
+      "step": 9413
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5671289886774734e-05,
+      "loss": 1.1172,
+      "step": 9414
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5670013218138844e-05,
+      "loss": 1.0642,
+      "step": 9415
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5668736480371483e-05,
+      "loss": 1.0425,
+      "step": 9416
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5667459673496703e-05,
+      "loss": 1.0932,
+      "step": 9417
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5666182797538556e-05,
+      "loss": 1.1347,
+      "step": 9418
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.566490585252109e-05,
+      "loss": 1.0225,
+      "step": 9419
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5663628838468364e-05,
+      "loss": 1.1372,
+      "step": 9420
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.566235175540443e-05,
+      "loss": 1.0381,
+      "step": 9421
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5661074603353343e-05,
+      "loss": 0.9969,
+      "step": 9422
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5659797382339163e-05,
+      "loss": 1.0189,
+      "step": 9423
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5658520092385945e-05,
+      "loss": 1.014,
+      "step": 9424
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5657242733517754e-05,
+      "loss": 1.1868,
+      "step": 9425
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.565596530575865e-05,
+      "loss": 1.1175,
+      "step": 9426
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5654687809132694e-05,
+      "loss": 1.0674,
+      "step": 9427
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5653410243663954e-05,
+      "loss": 1.0577,
+      "step": 9428
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5652132609376488e-05,
+      "loss": 1.1186,
+      "step": 9429
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5650854906294368e-05,
+      "loss": 1.0559,
+      "step": 9430
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.564957713444166e-05,
+      "loss": 1.0638,
+      "step": 9431
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.564829929384244e-05,
+      "loss": 1.1159,
+      "step": 9432
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5647021384520763e-05,
+      "loss": 1.0528,
+      "step": 9433
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5645743406500716e-05,
+      "loss": 1.0269,
+      "step": 9434
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5644465359806365e-05,
+      "loss": 1.0542,
+      "step": 9435
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5643187244461787e-05,
+      "loss": 1.1304,
+      "step": 9436
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5641909060491056e-05,
+      "loss": 1.1729,
+      "step": 9437
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5640630807918246e-05,
+      "loss": 1.0256,
+      "step": 9438
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5639352486767443e-05,
+      "loss": 1.053,
+      "step": 9439
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5638074097062716e-05,
+      "loss": 1.0833,
+      "step": 9440
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.563679563882816e-05,
+      "loss": 1.079,
+      "step": 9441
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5635517112087842e-05,
+      "loss": 1.0422,
+      "step": 9442
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5634238516865858e-05,
+      "loss": 1.1802,
+      "step": 9443
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.563295985318628e-05,
+      "loss": 1.0429,
+      "step": 9444
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5631681121073212e-05,
+      "loss": 1.0191,
+      "step": 9445
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.563040232055072e-05,
+      "loss": 1.0712,
+      "step": 9446
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5629123451642906e-05,
+      "loss": 1.1375,
+      "step": 9447
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.562784451437386e-05,
+      "loss": 1.0993,
+      "step": 9448
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5626565508767668e-05,
+      "loss": 1.0619,
+      "step": 9449
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5625286434848425e-05,
+      "loss": 1.1258,
+      "step": 9450
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.562400729264022e-05,
+      "loss": 1.1187,
+      "step": 9451
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5622728082167158e-05,
+      "loss": 1.0469,
+      "step": 9452
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5621448803453326e-05,
+      "loss": 1.0598,
+      "step": 9453
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5620169456522825e-05,
+      "loss": 1.1898,
+      "step": 9454
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5618890041399755e-05,
+      "loss": 1.0784,
+      "step": 9455
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5617610558108215e-05,
+      "loss": 1.1347,
+      "step": 9456
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5616331006672305e-05,
+      "loss": 1.0696,
+      "step": 9457
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5615051387116132e-05,
+      "loss": 1.1067,
+      "step": 9458
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5613771699463792e-05,
+      "loss": 1.0899,
+      "step": 9459
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5612491943739402e-05,
+      "loss": 1.0998,
+      "step": 9460
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.561121211996706e-05,
+      "loss": 1.1306,
+      "step": 9461
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5609932228170874e-05,
+      "loss": 1.1511,
+      "step": 9462
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5608652268374953e-05,
+      "loss": 1.0144,
+      "step": 9463
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.560737224060341e-05,
+      "loss": 1.1484,
+      "step": 9464
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.560609214488036e-05,
+      "loss": 1.0686,
+      "step": 9465
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5604811981229905e-05,
+      "loss": 1.094,
+      "step": 9466
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5603531749676175e-05,
+      "loss": 1.1146,
+      "step": 9467
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5602251450243274e-05,
+      "loss": 1.1218,
+      "step": 9468
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5600971082955316e-05,
+      "loss": 1.0753,
+      "step": 9469
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.559969064783643e-05,
+      "loss": 1.0767,
+      "step": 9470
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.559841014491073e-05,
+      "loss": 1.0525,
+      "step": 9471
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5597129574202333e-05,
+      "loss": 1.0267,
+      "step": 9472
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5595848935735364e-05,
+      "loss": 1.0693,
+      "step": 9473
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5594568229533952e-05,
+      "loss": 1.1084,
+      "step": 9474
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5593287455622214e-05,
+      "loss": 1.1478,
+      "step": 9475
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.559200661402428e-05,
+      "loss": 1.1171,
+      "step": 9476
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5590725704764268e-05,
+      "loss": 1.0108,
+      "step": 9477
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5589444727866322e-05,
+      "loss": 1.1635,
+      "step": 9478
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5588163683354557e-05,
+      "loss": 1.1502,
+      "step": 9479
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.558688257125311e-05,
+      "loss": 1.073,
+      "step": 9480
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5585601391586115e-05,
+      "loss": 1.1572,
+      "step": 9481
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.55843201443777e-05,
+      "loss": 1.1932,
+      "step": 9482
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5583038829652004e-05,
+      "loss": 1.0804,
+      "step": 9483
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5581757447433163e-05,
+      "loss": 1.1202,
+      "step": 9484
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5580475997745312e-05,
+      "loss": 1.079,
+      "step": 9485
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.557919448061259e-05,
+      "loss": 1.0439,
+      "step": 9486
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5577912896059138e-05,
+      "loss": 1.1029,
+      "step": 9487
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5576631244109092e-05,
+      "loss": 1.1352,
+      "step": 9488
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.557534952478661e-05,
+      "loss": 1.1524,
+      "step": 9489
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5574067738115813e-05,
+      "loss": 1.164,
+      "step": 9490
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5572785884120857e-05,
+      "loss": 1.1579,
+      "step": 9491
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.557150396282589e-05,
+      "loss": 1.061,
+      "step": 9492
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.557022197425506e-05,
+      "loss": 1.1584,
+      "step": 9493
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.556893991843251e-05,
+      "loss": 1.1157,
+      "step": 9494
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5567657795382395e-05,
+      "loss": 1.1603,
+      "step": 9495
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.556637560512886e-05,
+      "loss": 1.0582,
+      "step": 9496
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5565093347696063e-05,
+      "loss": 1.1539,
+      "step": 9497
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5563811023108158e-05,
+      "loss": 1.1068,
+      "step": 9498
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5562528631389295e-05,
+      "loss": 1.1097,
+      "step": 9499
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5561246172563632e-05,
+      "loss": 1.1261,
+      "step": 9500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5559963646655332e-05,
+      "loss": 1.0161,
+      "step": 9501
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.555868105368855e-05,
+      "loss": 1.0994,
+      "step": 9502
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.555739839368744e-05,
+      "loss": 1.2116,
+      "step": 9503
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.555611566667617e-05,
+      "loss": 1.1361,
+      "step": 9504
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5554832872678902e-05,
+      "loss": 1.0846,
+      "step": 9505
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.55535500117198e-05,
+      "loss": 0.9993,
+      "step": 9506
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5552267083823027e-05,
+      "loss": 1.2229,
+      "step": 9507
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.555098408901275e-05,
+      "loss": 1.1966,
+      "step": 9508
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5549701027313137e-05,
+      "loss": 1.108,
+      "step": 9509
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5548417898748362e-05,
+      "loss": 1.0697,
+      "step": 9510
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.554713470334258e-05,
+      "loss": 1.1556,
+      "step": 9511
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.554585144111998e-05,
+      "loss": 1.0394,
+      "step": 9512
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5544568112104722e-05,
+      "loss": 1.1518,
+      "step": 9513
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5543284716320988e-05,
+      "loss": 1.1582,
+      "step": 9514
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5542001253792945e-05,
+      "loss": 1.0802,
+      "step": 9515
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.554071772454478e-05,
+      "loss": 1.0143,
+      "step": 9516
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.553943412860066e-05,
+      "loss": 1.1184,
+      "step": 9517
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.553815046598477e-05,
+      "loss": 1.0302,
+      "step": 9518
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.553686673672129e-05,
+      "loss": 1.1254,
+      "step": 9519
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.55355829408344e-05,
+      "loss": 1.1931,
+      "step": 9520
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5534299078348283e-05,
+      "loss": 1.077,
+      "step": 9521
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5533015149287124e-05,
+      "loss": 1.0365,
+      "step": 9522
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5531731153675104e-05,
+      "loss": 1.1295,
+      "step": 9523
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5530447091536413e-05,
+      "loss": 1.1381,
+      "step": 9524
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.552916296289524e-05,
+      "loss": 1.0835,
+      "step": 9525
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.552787876777577e-05,
+      "loss": 0.9897,
+      "step": 9526
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5526594506202195e-05,
+      "loss": 1.0839,
+      "step": 9527
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5525310178198706e-05,
+      "loss": 1.0932,
+      "step": 9528
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.55240257837895e-05,
+      "loss": 1.0572,
+      "step": 9529
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5522741322998764e-05,
+      "loss": 1.1765,
+      "step": 9530
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.55214567958507e-05,
+      "loss": 1.0327,
+      "step": 9531
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.55201722023695e-05,
+      "loss": 1.1212,
+      "step": 9532
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.551888754257936e-05,
+      "loss": 1.0945,
+      "step": 9533
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.551760281650448e-05,
+      "loss": 1.0091,
+      "step": 9534
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5516318024169067e-05,
+      "loss": 1.0287,
+      "step": 9535
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5515033165597316e-05,
+      "loss": 1.0323,
+      "step": 9536
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.551374824081343e-05,
+      "loss": 1.0632,
+      "step": 9537
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5512463249841614e-05,
+      "loss": 1.0415,
+      "step": 9538
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.551117819270607e-05,
+      "loss": 1.1021,
+      "step": 9539
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5509893069431012e-05,
+      "loss": 1.0513,
+      "step": 9540
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5508607880040637e-05,
+      "loss": 1.1369,
+      "step": 9541
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5507322624559164e-05,
+      "loss": 1.0886,
+      "step": 9542
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.55060373030108e-05,
+      "loss": 1.0785,
+      "step": 9543
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5504751915419755e-05,
+      "loss": 1.0741,
+      "step": 9544
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.550346646181024e-05,
+      "loss": 1.0249,
+      "step": 9545
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.550218094220647e-05,
+      "loss": 1.0663,
+      "step": 9546
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5500895356632664e-05,
+      "loss": 1.063,
+      "step": 9547
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5499609705113035e-05,
+      "loss": 1.0504,
+      "step": 9548
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.54983239876718e-05,
+      "loss": 1.1191,
+      "step": 9549
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.549703820433318e-05,
+      "loss": 1.0604,
+      "step": 9550
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5495752355121396e-05,
+      "loss": 1.0448,
+      "step": 9551
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5494466440060666e-05,
+      "loss": 1.1453,
+      "step": 9552
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5493180459175214e-05,
+      "loss": 1.1353,
+      "step": 9553
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5491894412489264e-05,
+      "loss": 1.0938,
+      "step": 9554
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.549060830002704e-05,
+      "loss": 1.1069,
+      "step": 9555
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.548932212181277e-05,
+      "loss": 1.0956,
+      "step": 9556
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.548803587787068e-05,
+      "loss": 0.9602,
+      "step": 9557
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5486749568225e-05,
+      "loss": 1.028,
+      "step": 9558
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5485463192899958e-05,
+      "loss": 1.0802,
+      "step": 9559
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.548417675191979e-05,
+      "loss": 1.1421,
+      "step": 9560
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5482890245308724e-05,
+      "loss": 1.0707,
+      "step": 9561
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5481603673090994e-05,
+      "loss": 0.9916,
+      "step": 9562
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5480317035290836e-05,
+      "loss": 1.1694,
+      "step": 9563
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5479030331932487e-05,
+      "loss": 1.1055,
+      "step": 9564
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5477743563040188e-05,
+      "loss": 1.0349,
+      "step": 9565
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5476456728638168e-05,
+      "loss": 1.1219,
+      "step": 9566
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5475169828750672e-05,
+      "loss": 1.0335,
+      "step": 9567
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5473882863401942e-05,
+      "loss": 1.0947,
+      "step": 9568
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.547259583261622e-05,
+      "loss": 1.1927,
+      "step": 9569
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.547130873641775e-05,
+      "loss": 1.1062,
+      "step": 9570
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5470021574830775e-05,
+      "loss": 1.1288,
+      "step": 9571
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.546873434787954e-05,
+      "loss": 1.0848,
+      "step": 9572
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5467447055588298e-05,
+      "loss": 1.0658,
+      "step": 9573
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5466159697981294e-05,
+      "loss": 1.0086,
+      "step": 9574
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5464872275082777e-05,
+      "loss": 0.9909,
+      "step": 9575
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5463584786916998e-05,
+      "loss": 1.0884,
+      "step": 9576
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.546229723350821e-05,
+      "loss": 1.0542,
+      "step": 9577
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.546100961488067e-05,
+      "loss": 1.1202,
+      "step": 9578
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5459721931058624e-05,
+      "loss": 1.0652,
+      "step": 9579
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5458434182066334e-05,
+      "loss": 1.1171,
+      "step": 9580
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5457146367928057e-05,
+      "loss": 1.0835,
+      "step": 9581
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.545585848866805e-05,
+      "loss": 1.109,
+      "step": 9582
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5454570544310576e-05,
+      "loss": 1.0606,
+      "step": 9583
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5453282534879887e-05,
+      "loss": 1.0805,
+      "step": 9584
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5451994460400256e-05,
+      "loss": 1.1205,
+      "step": 9585
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.545070632089594e-05,
+      "loss": 1.092,
+      "step": 9586
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.54494181163912e-05,
+      "loss": 1.058,
+      "step": 9587
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5448129846910312e-05,
+      "loss": 1.1547,
+      "step": 9588
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5446841512477537e-05,
+      "loss": 1.0874,
+      "step": 9589
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5445553113117142e-05,
+      "loss": 1.093,
+      "step": 9590
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5444264648853396e-05,
+      "loss": 1.0734,
+      "step": 9591
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5442976119710572e-05,
+      "loss": 1.0979,
+      "step": 9592
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5441687525712943e-05,
+      "loss": 1.0945,
+      "step": 9593
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.544039886688478e-05,
+      "loss": 1.0405,
+      "step": 9594
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5439110143250358e-05,
+      "loss": 1.052,
+      "step": 9595
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.543782135483395e-05,
+      "loss": 1.0155,
+      "step": 9596
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5436532501659835e-05,
+      "loss": 1.0527,
+      "step": 9597
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5435243583752293e-05,
+      "loss": 1.1016,
+      "step": 9598
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.54339546011356e-05,
+      "loss": 1.0772,
+      "step": 9599
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5432665553834037e-05,
+      "loss": 1.0335,
+      "step": 9600
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.543137644187188e-05,
+      "loss": 1.0359,
+      "step": 9601
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5430087265273427e-05,
+      "loss": 1.0009,
+      "step": 9602
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5428798024062947e-05,
+      "loss": 1.026,
+      "step": 9603
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.542750871826473e-05,
+      "loss": 1.0496,
+      "step": 9604
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.542621934790307e-05,
+      "loss": 1.1052,
+      "step": 9605
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.542492991300224e-05,
+      "loss": 1.1127,
+      "step": 9606
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5423640413586542e-05,
+      "loss": 1.0529,
+      "step": 9607
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5422350849680257e-05,
+      "loss": 1.1132,
+      "step": 9608
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5421061221307683e-05,
+      "loss": 1.0213,
+      "step": 9609
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5419771528493107e-05,
+      "loss": 1.0769,
+      "step": 9610
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.541848177126083e-05,
+      "loss": 1.1046,
+      "step": 9611
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5417191949635137e-05,
+      "loss": 0.9805,
+      "step": 9612
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5415902063640335e-05,
+      "loss": 1.0754,
+      "step": 9613
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5414612113300715e-05,
+      "loss": 1.0203,
+      "step": 9614
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5413322098640575e-05,
+      "loss": 1.0701,
+      "step": 9615
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5412032019684214e-05,
+      "loss": 1.075,
+      "step": 9616
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5410741876455938e-05,
+      "loss": 1.1841,
+      "step": 9617
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5409451668980047e-05,
+      "loss": 1.026,
+      "step": 9618
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5408161397280844e-05,
+      "loss": 1.0125,
+      "step": 9619
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.540687106138263e-05,
+      "loss": 1.0466,
+      "step": 9620
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.540558066130972e-05,
+      "loss": 1.0497,
+      "step": 9621
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5404290197086413e-05,
+      "loss": 1.0661,
+      "step": 9622
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5402999668737024e-05,
+      "loss": 0.9954,
+      "step": 9623
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5401709076285854e-05,
+      "loss": 1.0875,
+      "step": 9624
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.540041841975722e-05,
+      "loss": 1.109,
+      "step": 9625
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.539912769917543e-05,
+      "loss": 1.0607,
+      "step": 9626
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5397836914564802e-05,
+      "loss": 1.01,
+      "step": 9627
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5396546065949643e-05,
+      "loss": 1.0105,
+      "step": 9628
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5395255153354273e-05,
+      "loss": 1.0499,
+      "step": 9629
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5393964176803013e-05,
+      "loss": 1.0665,
+      "step": 9630
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5392673136320177e-05,
+      "loss": 1.0972,
+      "step": 9631
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.539138203193008e-05,
+      "loss": 1.1202,
+      "step": 9632
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.539009086365705e-05,
+      "loss": 1.1506,
+      "step": 9633
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5388799631525402e-05,
+      "loss": 1.0665,
+      "step": 9634
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5387508335559458e-05,
+      "loss": 1.0616,
+      "step": 9635
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.538621697578355e-05,
+      "loss": 1.0445,
+      "step": 9636
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5384925552221996e-05,
+      "loss": 1.1479,
+      "step": 9637
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.538363406489913e-05,
+      "loss": 1.0128,
+      "step": 9638
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5382342513839268e-05,
+      "loss": 1.0929,
+      "step": 9639
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5381050899066748e-05,
+      "loss": 1.0217,
+      "step": 9640
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.537975922060589e-05,
+      "loss": 1.0328,
+      "step": 9641
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5378467478481043e-05,
+      "loss": 1.1339,
+      "step": 9642
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5377175672716522e-05,
+      "loss": 1.1351,
+      "step": 9643
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5375883803336672e-05,
+      "loss": 1.1028,
+      "step": 9644
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5374591870365817e-05,
+      "loss": 1.0837,
+      "step": 9645
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5373299873828304e-05,
+      "loss": 1.0248,
+      "step": 9646
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.537200781374846e-05,
+      "loss": 1.11,
+      "step": 9647
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.537071569015063e-05,
+      "loss": 1.1263,
+      "step": 9648
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.536942350305915e-05,
+      "loss": 1.0666,
+      "step": 9649
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5368131252498368e-05,
+      "loss": 1.1094,
+      "step": 9650
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5366838938492614e-05,
+      "loss": 1.0428,
+      "step": 9651
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5365546561066242e-05,
+      "loss": 1.1391,
+      "step": 9652
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.536425412024359e-05,
+      "loss": 1.0702,
+      "step": 9653
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5362961616049006e-05,
+      "loss": 1.0949,
+      "step": 9654
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5361669048506833e-05,
+      "loss": 1.1189,
+      "step": 9655
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5360376417641427e-05,
+      "loss": 1.0331,
+      "step": 9656
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5359083723477126e-05,
+      "loss": 1.0699,
+      "step": 9657
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.535779096603829e-05,
+      "loss": 1.0853,
+      "step": 9658
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5356498145349267e-05,
+      "loss": 1.0193,
+      "step": 9659
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5355205261434407e-05,
+      "loss": 1.0482,
+      "step": 9660
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.535391231431807e-05,
+      "loss": 1.1481,
+      "step": 9661
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5352619304024605e-05,
+      "loss": 1.0309,
+      "step": 9662
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.535132623057837e-05,
+      "loss": 1.0767,
+      "step": 9663
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5350033094003724e-05,
+      "loss": 1.0076,
+      "step": 9664
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5348739894325024e-05,
+      "loss": 1.0548,
+      "step": 9665
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.534744663156663e-05,
+      "loss": 1.0137,
+      "step": 9666
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5346153305752907e-05,
+      "loss": 1.1003,
+      "step": 9667
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5344859916908206e-05,
+      "loss": 1.0355,
+      "step": 9668
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5343566465056904e-05,
+      "loss": 1.1372,
+      "step": 9669
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.534227295022336e-05,
+      "loss": 1.0564,
+      "step": 9670
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5340979372431937e-05,
+      "loss": 1.0948,
+      "step": 9671
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5339685731707003e-05,
+      "loss": 1.1119,
+      "step": 9672
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5338392028072925e-05,
+      "loss": 1.069,
+      "step": 9673
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.533709826155408e-05,
+      "loss": 1.0348,
+      "step": 9674
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.533580443217483e-05,
+      "loss": 1.0837,
+      "step": 9675
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.533451053995955e-05,
+      "loss": 1.042,
+      "step": 9676
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5333216584932613e-05,
+      "loss": 1.0598,
+      "step": 9677
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5331922567118396e-05,
+      "loss": 1.108,
+      "step": 9678
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5330628486541263e-05,
+      "loss": 1.0757,
+      "step": 9679
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5329334343225606e-05,
+      "loss": 1.032,
+      "step": 9680
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.532804013719579e-05,
+      "loss": 1.1351,
+      "step": 9681
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.53267458684762e-05,
+      "loss": 1.0026,
+      "step": 9682
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5325451537091212e-05,
+      "loss": 1.0325,
+      "step": 9683
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5324157143065214e-05,
+      "loss": 1.0328,
+      "step": 9684
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5322862686422576e-05,
+      "loss": 1.0695,
+      "step": 9685
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.53215681671877e-05,
+      "loss": 1.0868,
+      "step": 9686
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.532027358538495e-05,
+      "loss": 1.0612,
+      "step": 9687
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.531897894103873e-05,
+      "loss": 1.1001,
+      "step": 9688
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5317684234173416e-05,
+      "loss": 1.0997,
+      "step": 9689
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5316389464813397e-05,
+      "loss": 1.0689,
+      "step": 9690
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5315094632983065e-05,
+      "loss": 1.0803,
+      "step": 9691
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.531379973870681e-05,
+      "loss": 1.0879,
+      "step": 9692
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5312504782009027e-05,
+      "loss": 1.1055,
+      "step": 9693
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5311209762914103e-05,
+      "loss": 1.0841,
+      "step": 9694
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5309914681446435e-05,
+      "loss": 1.0919,
+      "step": 9695
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5308619537630416e-05,
+      "loss": 1.2024,
+      "step": 9696
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5307324331490444e-05,
+      "loss": 1.1339,
+      "step": 9697
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.530602906305092e-05,
+      "loss": 1.0996,
+      "step": 9698
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5304733732336236e-05,
+      "loss": 1.0778,
+      "step": 9699
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.53034383393708e-05,
+      "loss": 1.0103,
+      "step": 9700
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5302142884179007e-05,
+      "loss": 1.1005,
+      "step": 9701
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5300847366785262e-05,
+      "loss": 0.9719,
+      "step": 9702
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5299551787213964e-05,
+      "loss": 0.9879,
+      "step": 9703
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5298256145489525e-05,
+      "loss": 1.0723,
+      "step": 9704
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5296960441636345e-05,
+      "loss": 1.0755,
+      "step": 9705
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.529566467567883e-05,
+      "loss": 1.111,
+      "step": 9706
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5294368847641394e-05,
+      "loss": 1.1109,
+      "step": 9707
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5293072957548444e-05,
+      "loss": 1.0075,
+      "step": 9708
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.529177700542439e-05,
+      "loss": 1.0347,
+      "step": 9709
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5290480991293643e-05,
+      "loss": 1.0133,
+      "step": 9710
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.528918491518062e-05,
+      "loss": 1.0743,
+      "step": 9711
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.528788877710972e-05,
+      "loss": 1.1395,
+      "step": 9712
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.528659257710538e-05,
+      "loss": 1.1172,
+      "step": 9713
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5285296315192e-05,
+      "loss": 1.1284,
+      "step": 9714
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5283999991394008e-05,
+      "loss": 1.1162,
+      "step": 9715
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5282703605735812e-05,
+      "loss": 1.042,
+      "step": 9716
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5281407158241843e-05,
+      "loss": 1.0475,
+      "step": 9717
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5280110648936512e-05,
+      "loss": 1.108,
+      "step": 9718
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5278814077844247e-05,
+      "loss": 1.0068,
+      "step": 9719
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5277517444989478e-05,
+      "loss": 1.0721,
+      "step": 9720
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.527622075039661e-05,
+      "loss": 1.0536,
+      "step": 9721
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.527492399409009e-05,
+      "loss": 1.119,
+      "step": 9722
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5273627176094327e-05,
+      "loss": 1.085,
+      "step": 9723
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.527233029643376e-05,
+      "loss": 1.0649,
+      "step": 9724
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5271033355132814e-05,
+      "loss": 1.0588,
+      "step": 9725
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5269736352215926e-05,
+      "loss": 1.081,
+      "step": 9726
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5268439287707515e-05,
+      "loss": 1.0914,
+      "step": 9727
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5267142161632026e-05,
+      "loss": 1.0703,
+      "step": 9728
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.526584497401388e-05,
+      "loss": 1.0276,
+      "step": 9729
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5264547724877526e-05,
+      "loss": 1.1184,
+      "step": 9730
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5263250414247392e-05,
+      "loss": 1.0675,
+      "step": 9731
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5261953042147914e-05,
+      "loss": 0.9743,
+      "step": 9732
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5260655608603535e-05,
+      "loss": 1.0765,
+      "step": 9733
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.525935811363869e-05,
+      "loss": 1.0448,
+      "step": 9734
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5258060557277822e-05,
+      "loss": 1.0497,
+      "step": 9735
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5256762939545375e-05,
+      "loss": 1.0482,
+      "step": 9736
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.525546526046579e-05,
+      "loss": 1.0669,
+      "step": 9737
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.525416752006351e-05,
+      "loss": 1.1739,
+      "step": 9738
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5252869718362984e-05,
+      "loss": 1.1343,
+      "step": 9739
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5251571855388652e-05,
+      "loss": 1.1253,
+      "step": 9740
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5250273931164967e-05,
+      "loss": 1.0575,
+      "step": 9741
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5248975945716378e-05,
+      "loss": 1.1489,
+      "step": 9742
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5247677899067332e-05,
+      "loss": 1.1277,
+      "step": 9743
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5246379791242283e-05,
+      "loss": 1.1433,
+      "step": 9744
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.524508162226568e-05,
+      "loss": 1.0548,
+      "step": 9745
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5243783392161977e-05,
+      "loss": 1.075,
+      "step": 9746
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5242485100955633e-05,
+      "loss": 1.0597,
+      "step": 9747
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.52411867486711e-05,
+      "loss": 1.1796,
+      "step": 9748
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5239888335332832e-05,
+      "loss": 1.1788,
+      "step": 9749
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5238589860965294e-05,
+      "loss": 1.0713,
+      "step": 9750
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.523729132559294e-05,
+      "loss": 0.9659,
+      "step": 9751
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5235992729240232e-05,
+      "loss": 1.0427,
+      "step": 9752
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.523469407193163e-05,
+      "loss": 1.0262,
+      "step": 9753
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5233395353691597e-05,
+      "loss": 1.089,
+      "step": 9754
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5232096574544601e-05,
+      "loss": 1.0848,
+      "step": 9755
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5230797734515101e-05,
+      "loss": 1.1154,
+      "step": 9756
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.522949883362757e-05,
+      "loss": 1.0189,
+      "step": 9757
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5228199871906464e-05,
+      "loss": 1.1269,
+      "step": 9758
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5226900849376262e-05,
+      "loss": 1.0403,
+      "step": 9759
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5225601766061429e-05,
+      "loss": 1.0817,
+      "step": 9760
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5224302621986434e-05,
+      "loss": 1.1251,
+      "step": 9761
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5223003417175754e-05,
+      "loss": 1.0504,
+      "step": 9762
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5221704151653856e-05,
+      "loss": 1.0829,
+      "step": 9763
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.522040482544522e-05,
+      "loss": 1.0368,
+      "step": 9764
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5219105438574316e-05,
+      "loss": 1.0711,
+      "step": 9765
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5217805991065623e-05,
+      "loss": 1.0747,
+      "step": 9766
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5216506482943618e-05,
+      "loss": 1.069,
+      "step": 9767
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.521520691423278e-05,
+      "loss": 1.0516,
+      "step": 9768
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5213907284957587e-05,
+      "loss": 1.1547,
+      "step": 9769
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5212607595142525e-05,
+      "loss": 1.0486,
+      "step": 9770
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5211307844812068e-05,
+      "loss": 1.1103,
+      "step": 9771
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5210008033990706e-05,
+      "loss": 1.0377,
+      "step": 9772
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.520870816270292e-05,
+      "loss": 1.1097,
+      "step": 9773
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5207408230973198e-05,
+      "loss": 1.1093,
+      "step": 9774
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5206108238826022e-05,
+      "loss": 1.0101,
+      "step": 9775
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5204808186285883e-05,
+      "loss": 1.009,
+      "step": 9776
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5203508073377273e-05,
+      "loss": 1.0754,
+      "step": 9777
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5202207900124677e-05,
+      "loss": 1.0341,
+      "step": 9778
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5200907666552586e-05,
+      "loss": 1.0404,
+      "step": 9779
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5199607372685496e-05,
+      "loss": 1.1153,
+      "step": 9780
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.51983070185479e-05,
+      "loss": 1.0645,
+      "step": 9781
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5197006604164288e-05,
+      "loss": 1.0832,
+      "step": 9782
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.519570612955916e-05,
+      "loss": 1.1048,
+      "step": 9783
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5194405594757011e-05,
+      "loss": 1.0812,
+      "step": 9784
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.519310499978234e-05,
+      "loss": 1.0431,
+      "step": 9785
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5191804344659646e-05,
+      "loss": 1.0574,
+      "step": 9786
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5190503629413428e-05,
+      "loss": 1.0775,
+      "step": 9787
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.518920285406819e-05,
+      "loss": 0.986,
+      "step": 9788
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5187902018648431e-05,
+      "loss": 0.9797,
+      "step": 9789
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5186601123178657e-05,
+      "loss": 1.1231,
+      "step": 9790
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5185300167683374e-05,
+      "loss": 1.0926,
+      "step": 9791
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5183999152187083e-05,
+      "loss": 1.0967,
+      "step": 9792
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.51826980767143e-05,
+      "loss": 1.0279,
+      "step": 9793
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.518139694128952e-05,
+      "loss": 1.0518,
+      "step": 9794
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5180095745937265e-05,
+      "loss": 0.9934,
+      "step": 9795
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5178794490682036e-05,
+      "loss": 1.0898,
+      "step": 9796
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5177493175548352e-05,
+      "loss": 1.0406,
+      "step": 9797
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.517619180056072e-05,
+      "loss": 1.09,
+      "step": 9798
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5174890365743657e-05,
+      "loss": 1.045,
+      "step": 9799
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5173588871121674e-05,
+      "loss": 1.0826,
+      "step": 9800
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5172287316719297e-05,
+      "loss": 1.0991,
+      "step": 9801
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5170985702561029e-05,
+      "loss": 1.1419,
+      "step": 9802
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5169684028671399e-05,
+      "loss": 1.1752,
+      "step": 9803
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5168382295074923e-05,
+      "loss": 1.0949,
+      "step": 9804
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5167080501796119e-05,
+      "loss": 1.0557,
+      "step": 9805
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5165778648859514e-05,
+      "loss": 1.1001,
+      "step": 9806
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5164476736289627e-05,
+      "loss": 1.0918,
+      "step": 9807
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5163174764110984e-05,
+      "loss": 1.1002,
+      "step": 9808
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5161872732348105e-05,
+      "loss": 0.9811,
+      "step": 9809
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5160570641025525e-05,
+      "loss": 1.0713,
+      "step": 9810
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.515926849016776e-05,
+      "loss": 1.0427,
+      "step": 9811
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5157966279799351e-05,
+      "loss": 1.0829,
+      "step": 9812
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5156664009944817e-05,
+      "loss": 0.9851,
+      "step": 9813
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5155361680628697e-05,
+      "loss": 1.0512,
+      "step": 9814
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5154059291875518e-05,
+      "loss": 1.0059,
+      "step": 9815
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5152756843709813e-05,
+      "loss": 1.054,
+      "step": 9816
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5151454336156114e-05,
+      "loss": 1.0648,
+      "step": 9817
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5150151769238963e-05,
+      "loss": 1.0073,
+      "step": 9818
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5148849142982889e-05,
+      "loss": 1.0796,
+      "step": 9819
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5147546457412434e-05,
+      "loss": 1.0839,
+      "step": 9820
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5146243712552135e-05,
+      "loss": 1.0579,
+      "step": 9821
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5144940908426531e-05,
+      "loss": 1.1044,
+      "step": 9822
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5143638045060163e-05,
+      "loss": 1.0209,
+      "step": 9823
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5142335122477574e-05,
+      "loss": 1.096,
+      "step": 9824
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5141032140703305e-05,
+      "loss": 1.1223,
+      "step": 9825
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.51397290997619e-05,
+      "loss": 1.0667,
+      "step": 9826
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5138425999677906e-05,
+      "loss": 0.9997,
+      "step": 9827
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.513712284047587e-05,
+      "loss": 1.0559,
+      "step": 9828
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5135819622180337e-05,
+      "loss": 1.0287,
+      "step": 9829
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5134516344815857e-05,
+      "loss": 1.0477,
+      "step": 9830
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.513321300840698e-05,
+      "loss": 1.053,
+      "step": 9831
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5131909612978253e-05,
+      "loss": 1.0435,
+      "step": 9832
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5130606158554232e-05,
+      "loss": 1.1034,
+      "step": 9833
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.512930264515947e-05,
+      "loss": 1.1089,
+      "step": 9834
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.512799907281852e-05,
+      "loss": 1.0743,
+      "step": 9835
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5126695441555934e-05,
+      "loss": 1.1373,
+      "step": 9836
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5125391751396275e-05,
+      "loss": 1.0555,
+      "step": 9837
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5124088002364093e-05,
+      "loss": 0.9773,
+      "step": 9838
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5122784194483952e-05,
+      "loss": 1.0606,
+      "step": 9839
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5121480327780408e-05,
+      "loss": 1.0814,
+      "step": 9840
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5120176402278028e-05,
+      "loss": 1.1176,
+      "step": 9841
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5118872418001364e-05,
+      "loss": 1.0379,
+      "step": 9842
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5117568374974989e-05,
+      "loss": 1.0967,
+      "step": 9843
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5116264273223458e-05,
+      "loss": 1.0685,
+      "step": 9844
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5114960112771343e-05,
+      "loss": 1.053,
+      "step": 9845
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5113655893643208e-05,
+      "loss": 1.0388,
+      "step": 9846
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5112351615863621e-05,
+      "loss": 1.0151,
+      "step": 9847
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5111047279457149e-05,
+      "loss": 1.0522,
+      "step": 9848
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.510974288444836e-05,
+      "loss": 1.0884,
+      "step": 9849
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5108438430861831e-05,
+      "loss": 1.0418,
+      "step": 9850
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5107133918722125e-05,
+      "loss": 1.0373,
+      "step": 9851
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5105829348053823e-05,
+      "loss": 1.0332,
+      "step": 9852
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5104524718881494e-05,
+      "loss": 1.0631,
+      "step": 9853
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5103220031229716e-05,
+      "loss": 1.1129,
+      "step": 9854
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5101915285123063e-05,
+      "loss": 1.0233,
+      "step": 9855
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5100610480586114e-05,
+      "loss": 1.1489,
+      "step": 9856
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5099305617643449e-05,
+      "loss": 1.0526,
+      "step": 9857
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5098000696319642e-05,
+      "loss": 1.0439,
+      "step": 9858
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5096695716639276e-05,
+      "loss": 1.1126,
+      "step": 9859
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5095390678626935e-05,
+      "loss": 1.1002,
+      "step": 9860
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5094085582307198e-05,
+      "loss": 1.0855,
+      "step": 9861
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5092780427704654e-05,
+      "loss": 1.1104,
+      "step": 9862
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5091475214843884e-05,
+      "loss": 1.1314,
+      "step": 9863
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5090169943749475e-05,
+      "loss": 1.0821,
+      "step": 9864
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5088864614446015e-05,
+      "loss": 1.0869,
+      "step": 9865
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5087559226958091e-05,
+      "loss": 1.1046,
+      "step": 9866
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5086253781310293e-05,
+      "loss": 1.0521,
+      "step": 9867
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5084948277527211e-05,
+      "loss": 1.0639,
+      "step": 9868
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.508364271563344e-05,
+      "loss": 1.1104,
+      "step": 9869
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5082337095653568e-05,
+      "loss": 1.1211,
+      "step": 9870
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5081031417612188e-05,
+      "loss": 1.0828,
+      "step": 9871
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5079725681533901e-05,
+      "loss": 1.0554,
+      "step": 9872
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5078419887443298e-05,
+      "loss": 1.0932,
+      "step": 9873
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5077114035364977e-05,
+      "loss": 1.0551,
+      "step": 9874
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5075808125323536e-05,
+      "loss": 1.0356,
+      "step": 9875
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5074502157343575e-05,
+      "loss": 1.0329,
+      "step": 9876
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5073196131449699e-05,
+      "loss": 0.9827,
+      "step": 9877
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5071890047666498e-05,
+      "loss": 1.1164,
+      "step": 9878
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5070583906018585e-05,
+      "loss": 0.9781,
+      "step": 9879
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5069277706530554e-05,
+      "loss": 1.0804,
+      "step": 9880
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5067971449227021e-05,
+      "loss": 1.127,
+      "step": 9881
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5066665134132584e-05,
+      "loss": 1.0804,
+      "step": 9882
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5065358761271852e-05,
+      "loss": 1.09,
+      "step": 9883
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5064052330669429e-05,
+      "loss": 1.0213,
+      "step": 9884
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5062745842349935e-05,
+      "loss": 1.0526,
+      "step": 9885
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5061439296337968e-05,
+      "loss": 1.076,
+      "step": 9886
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5060132692658148e-05,
+      "loss": 1.1087,
+      "step": 9887
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.505882603133508e-05,
+      "loss": 1.0187,
+      "step": 9888
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.505751931239338e-05,
+      "loss": 1.0294,
+      "step": 9889
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5056212535857665e-05,
+      "loss": 0.9704,
+      "step": 9890
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5054905701752549e-05,
+      "loss": 1.199,
+      "step": 9891
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5053598810102647e-05,
+      "loss": 0.9813,
+      "step": 9892
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5052291860932577e-05,
+      "loss": 1.0828,
+      "step": 9893
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5050984854266963e-05,
+      "loss": 1.1138,
+      "step": 9894
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5049677790130419e-05,
+      "loss": 1.099,
+      "step": 9895
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.504837066854757e-05,
+      "loss": 1.0623,
+      "step": 9896
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5047063489543031e-05,
+      "loss": 0.9813,
+      "step": 9897
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5045756253141435e-05,
+      "loss": 1.0691,
+      "step": 9898
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.50444489593674e-05,
+      "loss": 1.0557,
+      "step": 9899
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.504314160824555e-05,
+      "loss": 1.0867,
+      "step": 9900
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5041834199800515e-05,
+      "loss": 1.027,
+      "step": 9901
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5040526734056921e-05,
+      "loss": 1.1545,
+      "step": 9902
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5039219211039397e-05,
+      "loss": 1.04,
+      "step": 9903
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5037911630772575e-05,
+      "loss": 0.9984,
+      "step": 9904
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.503660399328108e-05,
+      "loss": 0.9976,
+      "step": 9905
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5035296298589548e-05,
+      "loss": 1.0655,
+      "step": 9906
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.503398854672261e-05,
+      "loss": 1.0225,
+      "step": 9907
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5032680737704902e-05,
+      "loss": 1.064,
+      "step": 9908
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5031372871561056e-05,
+      "loss": 1.0112,
+      "step": 9909
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.503006494831571e-05,
+      "loss": 1.052,
+      "step": 9910
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5028756967993501e-05,
+      "loss": 1.052,
+      "step": 9911
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5027448930619069e-05,
+      "loss": 1.078,
+      "step": 9912
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5026140836217049e-05,
+      "loss": 1.0779,
+      "step": 9913
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5024832684812085e-05,
+      "loss": 1.0944,
+      "step": 9914
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5023524476428816e-05,
+      "loss": 1.0576,
+      "step": 9915
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5022216211091887e-05,
+      "loss": 1.0619,
+      "step": 9916
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5020907888825939e-05,
+      "loss": 1.016,
+      "step": 9917
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.501959950965562e-05,
+      "loss": 1.1247,
+      "step": 9918
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5018291073605573e-05,
+      "loss": 1.0343,
+      "step": 9919
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5016982580700443e-05,
+      "loss": 1.0846,
+      "step": 9920
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5015674030964887e-05,
+      "loss": 1.0056,
+      "step": 9921
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5014365424423538e-05,
+      "loss": 1.0765,
+      "step": 9922
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5013056761101063e-05,
+      "loss": 1.0207,
+      "step": 9923
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.50117480410221e-05,
+      "loss": 1.073,
+      "step": 9924
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5010439264211312e-05,
+      "loss": 0.9779,
+      "step": 9925
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5009130430693342e-05,
+      "loss": 1.0513,
+      "step": 9926
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5007821540492853e-05,
+      "loss": 1.0868,
+      "step": 9927
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5006512593634492e-05,
+      "loss": 1.0294,
+      "step": 9928
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5005203590142925e-05,
+      "loss": 1.0624,
+      "step": 9929
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5003894530042802e-05,
+      "loss": 1.0668,
+      "step": 9930
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5002585413358783e-05,
+      "loss": 1.1077,
+      "step": 9931
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.500127624011553e-05,
+      "loss": 1.0567,
+      "step": 9932
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4999967010337702e-05,
+      "loss": 1.0493,
+      "step": 9933
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.499865772404996e-05,
+      "loss": 1.0376,
+      "step": 9934
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4997348381276968e-05,
+      "loss": 1.0682,
+      "step": 9935
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.499603898204339e-05,
+      "loss": 1.0812,
+      "step": 9936
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.499472952637389e-05,
+      "loss": 1.0464,
+      "step": 9937
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4993420014293138e-05,
+      "loss": 1.0175,
+      "step": 9938
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4992110445825794e-05,
+      "loss": 1.1144,
+      "step": 9939
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4990800820996534e-05,
+      "loss": 1.1283,
+      "step": 9940
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4989491139830022e-05,
+      "loss": 1.0899,
+      "step": 9941
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4988181402350926e-05,
+      "loss": 1.1011,
+      "step": 9942
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.498687160858392e-05,
+      "loss": 1.1076,
+      "step": 9943
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4985561758553682e-05,
+      "loss": 1.0881,
+      "step": 9944
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.498425185228488e-05,
+      "loss": 1.0977,
+      "step": 9945
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4982941889802188e-05,
+      "loss": 0.9862,
+      "step": 9946
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.498163187113028e-05,
+      "loss": 1.0533,
+      "step": 9947
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4980321796293838e-05,
+      "loss": 1.0426,
+      "step": 9948
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4979011665317536e-05,
+      "loss": 1.0611,
+      "step": 9949
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4977701478226054e-05,
+      "loss": 1.1406,
+      "step": 9950
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4976391235044072e-05,
+      "loss": 1.0732,
+      "step": 9951
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.497508093579627e-05,
+      "loss": 1.0862,
+      "step": 9952
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4973770580507327e-05,
+      "loss": 0.9565,
+      "step": 9953
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4972460169201933e-05,
+      "loss": 1.1028,
+      "step": 9954
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4971149701904767e-05,
+      "loss": 1.0475,
+      "step": 9955
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4969839178640514e-05,
+      "loss": 0.985,
+      "step": 9956
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4968528599433864e-05,
+      "loss": 1.0106,
+      "step": 9957
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.49672179643095e-05,
+      "loss": 1.0873,
+      "step": 9958
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4965907273292113e-05,
+      "loss": 1.1072,
+      "step": 9959
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4964596526406389e-05,
+      "loss": 1.109,
+      "step": 9960
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4963285723677024e-05,
+      "loss": 1.1057,
+      "step": 9961
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4961974865128699e-05,
+      "loss": 1.0007,
+      "step": 9962
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.496066395078612e-05,
+      "loss": 1.0103,
+      "step": 9963
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4959352980673968e-05,
+      "loss": 1.0321,
+      "step": 9964
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4958041954816948e-05,
+      "loss": 1.0728,
+      "step": 9965
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4956730873239746e-05,
+      "loss": 1.1571,
+      "step": 9966
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.495541973596707e-05,
+      "loss": 0.9506,
+      "step": 9967
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4954108543023605e-05,
+      "loss": 1.079,
+      "step": 9968
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.495279729443406e-05,
+      "loss": 1.0874,
+      "step": 9969
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4951485990223126e-05,
+      "loss": 0.9885,
+      "step": 9970
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4950174630415515e-05,
+      "loss": 1.0926,
+      "step": 9971
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4948863215035918e-05,
+      "loss": 1.0579,
+      "step": 9972
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4947551744109044e-05,
+      "loss": 1.1009,
+      "step": 9973
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4946240217659595e-05,
+      "loss": 1.1124,
+      "step": 9974
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4944928635712274e-05,
+      "loss": 1.1034,
+      "step": 9975
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4943616998291792e-05,
+      "loss": 1.1307,
+      "step": 9976
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4942305305422853e-05,
+      "loss": 0.9946,
+      "step": 9977
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4940993557130165e-05,
+      "loss": 1.0226,
+      "step": 9978
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4939681753438437e-05,
+      "loss": 1.0528,
+      "step": 9979
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4938369894372381e-05,
+      "loss": 1.0651,
+      "step": 9980
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4937057979956707e-05,
+      "loss": 1.1035,
+      "step": 9981
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.493574601021613e-05,
+      "loss": 1.1088,
+      "step": 9982
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4934433985175358e-05,
+      "loss": 1.0777,
+      "step": 9983
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4933121904859111e-05,
+      "loss": 1.0791,
+      "step": 9984
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.49318097692921e-05,
+      "loss": 1.0718,
+      "step": 9985
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4930497578499046e-05,
+      "loss": 1.0155,
+      "step": 9986
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.492918533250466e-05,
+      "loss": 1.0846,
+      "step": 9987
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4927873031333666e-05,
+      "loss": 1.0593,
+      "step": 9988
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4926560675010784e-05,
+      "loss": 1.1617,
+      "step": 9989
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4925248263560733e-05,
+      "loss": 1.0758,
+      "step": 9990
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.492393579700823e-05,
+      "loss": 1.0809,
+      "step": 9991
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.492262327537801e-05,
+      "loss": 1.0681,
+      "step": 9992
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4921310698694782e-05,
+      "loss": 1.0788,
+      "step": 9993
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4919998066983278e-05,
+      "loss": 1.0028,
+      "step": 9994
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4918685380268227e-05,
+      "loss": 1.1039,
+      "step": 9995
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.491737263857435e-05,
+      "loss": 1.0731,
+      "step": 9996
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4916059841926378e-05,
+      "loss": 1.0423,
+      "step": 9997
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.491474699034904e-05,
+      "loss": 1.1046,
+      "step": 9998
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4913434083867066e-05,
+      "loss": 1.0404,
+      "step": 9999
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4912121122505186e-05,
+      "loss": 1.0791,
+      "step": 10000
+    },
+    {
+      "epoch": 0.42,
+      "eval_text_loss": 0.7981115579605103,
+      "eval_text_runtime": 1.4033,
+      "eval_text_samples_per_second": 52.733,
+      "eval_text_steps_per_second": 1.425,
+      "step": 10000
+    },
+    {
+      "epoch": 0.42,
+      "eval_svit_loss": 1.1445456743240356,
+      "eval_svit_runtime": 5.0004,
+      "eval_svit_samples_per_second": 136.189,
+      "eval_svit_steps_per_second": 2.2,
+      "step": 10000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.491080810628813e-05,
+      "loss": 1.0323,
+      "step": 10001
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4909495035240638e-05,
+      "loss": 1.0483,
+      "step": 10002
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4908181909387437e-05,
+      "loss": 1.0607,
+      "step": 10003
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4906868728753263e-05,
+      "loss": 1.0105,
+      "step": 10004
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4905555493362858e-05,
+      "loss": 1.0463,
+      "step": 10005
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.490424220324095e-05,
+      "loss": 1.0718,
+      "step": 10006
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4902928858412289e-05,
+      "loss": 1.0353,
+      "step": 10007
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4901615458901601e-05,
+      "loss": 1.0331,
+      "step": 10008
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4900302004733638e-05,
+      "loss": 1.0944,
+      "step": 10009
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4898988495933132e-05,
+      "loss": 1.0545,
+      "step": 10010
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4897674932524832e-05,
+      "loss": 1.061,
+      "step": 10011
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4896361314533478e-05,
+      "loss": 1.0673,
+      "step": 10012
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4895047641983819e-05,
+      "loss": 1.0261,
+      "step": 10013
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4893733914900594e-05,
+      "loss": 1.0044,
+      "step": 10014
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4892420133308554e-05,
+      "loss": 1.0232,
+      "step": 10015
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4891106297232442e-05,
+      "loss": 0.9817,
+      "step": 10016
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.488979240669701e-05,
+      "loss": 0.9738,
+      "step": 10017
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4888478461727009e-05,
+      "loss": 0.9884,
+      "step": 10018
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4887164462347185e-05,
+      "loss": 1.0807,
+      "step": 10019
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4885850408582294e-05,
+      "loss": 1.0274,
+      "step": 10020
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4884536300457086e-05,
+      "loss": 1.0866,
+      "step": 10021
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4883222137996316e-05,
+      "loss": 1.0331,
+      "step": 10022
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4881907921224739e-05,
+      "loss": 0.9413,
+      "step": 10023
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4880593650167107e-05,
+      "loss": 1.0423,
+      "step": 10024
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4879279324848181e-05,
+      "loss": 1.0199,
+      "step": 10025
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4877964945292717e-05,
+      "loss": 1.0526,
+      "step": 10026
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4876650511525475e-05,
+      "loss": 1.0337,
+      "step": 10027
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4875336023571214e-05,
+      "loss": 1.0673,
+      "step": 10028
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4874021481454692e-05,
+      "loss": 1.0901,
+      "step": 10029
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4872706885200676e-05,
+      "loss": 0.9743,
+      "step": 10030
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4871392234833926e-05,
+      "loss": 1.0454,
+      "step": 10031
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4870077530379205e-05,
+      "loss": 1.0792,
+      "step": 10032
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4868762771861277e-05,
+      "loss": 1.0355,
+      "step": 10033
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4867447959304918e-05,
+      "loss": 1.0244,
+      "step": 10034
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4866133092734878e-05,
+      "loss": 1.0461,
+      "step": 10035
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4864818172175938e-05,
+      "loss": 1.055,
+      "step": 10036
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4863503197652859e-05,
+      "loss": 1.0669,
+      "step": 10037
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4862188169190418e-05,
+      "loss": 1.0608,
+      "step": 10038
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4860873086813382e-05,
+      "loss": 1.005,
+      "step": 10039
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4859557950546526e-05,
+      "loss": 1.1063,
+      "step": 10040
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4858242760414619e-05,
+      "loss": 1.0896,
+      "step": 10041
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4856927516442434e-05,
+      "loss": 1.012,
+      "step": 10042
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4855612218654754e-05,
+      "loss": 1.1028,
+      "step": 10043
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4854296867076346e-05,
+      "loss": 0.999,
+      "step": 10044
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4852981461731994e-05,
+      "loss": 1.0958,
+      "step": 10045
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4851666002646467e-05,
+      "loss": 1.0579,
+      "step": 10046
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4850350489844558e-05,
+      "loss": 1.073,
+      "step": 10047
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4849034923351034e-05,
+      "loss": 1.1195,
+      "step": 10048
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4847719303190684e-05,
+      "loss": 0.9903,
+      "step": 10049
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4846403629388284e-05,
+      "loss": 1.0814,
+      "step": 10050
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4845087901968625e-05,
+      "loss": 1.1746,
+      "step": 10051
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4843772120956483e-05,
+      "loss": 1.1075,
+      "step": 10052
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4842456286376649e-05,
+      "loss": 1.0262,
+      "step": 10053
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4841140398253906e-05,
+      "loss": 1.086,
+      "step": 10054
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4839824456613043e-05,
+      "loss": 1.1059,
+      "step": 10055
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.483850846147885e-05,
+      "loss": 1.0951,
+      "step": 10056
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4837192412876109e-05,
+      "loss": 1.0584,
+      "step": 10057
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4835876310829618e-05,
+      "loss": 0.993,
+      "step": 10058
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4834560155364163e-05,
+      "loss": 1.059,
+      "step": 10059
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.483324394650454e-05,
+      "loss": 1.0993,
+      "step": 10060
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.483192768427554e-05,
+      "loss": 1.0555,
+      "step": 10061
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4830611368701956e-05,
+      "loss": 1.0698,
+      "step": 10062
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4829294999808588e-05,
+      "loss": 1.1027,
+      "step": 10063
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4827978577620228e-05,
+      "loss": 1.1014,
+      "step": 10064
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4826662102161673e-05,
+      "loss": 0.9793,
+      "step": 10065
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4825345573457722e-05,
+      "loss": 1.037,
+      "step": 10066
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4824028991533179e-05,
+      "loss": 1.0818,
+      "step": 10067
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4822712356412839e-05,
+      "loss": 1.0565,
+      "step": 10068
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4821395668121503e-05,
+      "loss": 0.983,
+      "step": 10069
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4820078926683975e-05,
+      "loss": 1.0886,
+      "step": 10070
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.481876213212506e-05,
+      "loss": 1.1014,
+      "step": 10071
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4817445284469555e-05,
+      "loss": 1.1056,
+      "step": 10072
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4816128383742273e-05,
+      "loss": 1.1191,
+      "step": 10073
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4814811429968021e-05,
+      "loss": 0.9957,
+      "step": 10074
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4813494423171599e-05,
+      "loss": 1.0654,
+      "step": 10075
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4812177363377825e-05,
+      "loss": 1.0304,
+      "step": 10076
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4810860250611495e-05,
+      "loss": 1.0618,
+      "step": 10077
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4809543084897433e-05,
+      "loss": 1.1248,
+      "step": 10078
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.480822586626044e-05,
+      "loss": 1.0718,
+      "step": 10079
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4806908594725335e-05,
+      "loss": 1.053,
+      "step": 10080
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4805591270316923e-05,
+      "loss": 1.084,
+      "step": 10081
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4804273893060027e-05,
+      "loss": 1.0189,
+      "step": 10082
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4802956462979461e-05,
+      "loss": 1.0635,
+      "step": 10083
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4801638980100037e-05,
+      "loss": 1.0768,
+      "step": 10084
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4800321444446575e-05,
+      "loss": 1.104,
+      "step": 10085
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4799003856043892e-05,
+      "loss": 1.0603,
+      "step": 10086
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4797686214916811e-05,
+      "loss": 1.139,
+      "step": 10087
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4796368521090143e-05,
+      "loss": 1.0684,
+      "step": 10088
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4795050774588721e-05,
+      "loss": 1.089,
+      "step": 10089
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4793732975437356e-05,
+      "loss": 1.0616,
+      "step": 10090
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4792415123660881e-05,
+      "loss": 1.0035,
+      "step": 10091
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4791097219284114e-05,
+      "loss": 1.0167,
+      "step": 10092
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4789779262331882e-05,
+      "loss": 1.115,
+      "step": 10093
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4788461252829009e-05,
+      "loss": 1.0205,
+      "step": 10094
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.478714319080033e-05,
+      "loss": 1.1367,
+      "step": 10095
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4785825076270662e-05,
+      "loss": 1.0778,
+      "step": 10096
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4784506909264841e-05,
+      "loss": 1.0946,
+      "step": 10097
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4783188689807696e-05,
+      "loss": 1.1487,
+      "step": 10098
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4781870417924055e-05,
+      "loss": 1.0563,
+      "step": 10099
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4780552093638756e-05,
+      "loss": 1.0395,
+      "step": 10100
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4779233716976628e-05,
+      "loss": 1.0283,
+      "step": 10101
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4777915287962505e-05,
+      "loss": 1.1108,
+      "step": 10102
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4776596806621223e-05,
+      "loss": 1.0256,
+      "step": 10103
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.477527827297762e-05,
+      "loss": 1.0546,
+      "step": 10104
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4773959687056527e-05,
+      "loss": 1.023,
+      "step": 10105
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4772641048882787e-05,
+      "loss": 1.1137,
+      "step": 10106
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4771322358481242e-05,
+      "loss": 1.0534,
+      "step": 10107
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4770003615876725e-05,
+      "loss": 1.0358,
+      "step": 10108
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4768684821094079e-05,
+      "loss": 1.0441,
+      "step": 10109
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4767365974158145e-05,
+      "loss": 0.9869,
+      "step": 10110
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4766047075093773e-05,
+      "loss": 1.0663,
+      "step": 10111
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4764728123925795e-05,
+      "loss": 1.1032,
+      "step": 10112
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4763409120679069e-05,
+      "loss": 1.1307,
+      "step": 10113
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.476209006537843e-05,
+      "loss": 1.0964,
+      "step": 10114
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4760770958048728e-05,
+      "loss": 1.0777,
+      "step": 10115
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4759451798714816e-05,
+      "loss": 1.0693,
+      "step": 10116
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4758132587401536e-05,
+      "loss": 1.0399,
+      "step": 10117
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4756813324133745e-05,
+      "loss": 1.0451,
+      "step": 10118
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4755494008936284e-05,
+      "loss": 1.1585,
+      "step": 10119
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4754174641834016e-05,
+      "loss": 1.1094,
+      "step": 10120
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.475285522285178e-05,
+      "loss": 0.9827,
+      "step": 10121
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4751535752014443e-05,
+      "loss": 1.0231,
+      "step": 10122
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4750216229346849e-05,
+      "loss": 1.0832,
+      "step": 10123
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4748896654873865e-05,
+      "loss": 1.019,
+      "step": 10124
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4747577028620337e-05,
+      "loss": 1.0793,
+      "step": 10125
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4746257350611129e-05,
+      "loss": 1.0843,
+      "step": 10126
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4744937620871096e-05,
+      "loss": 1.0374,
+      "step": 10127
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.47436178394251e-05,
+      "loss": 1.0204,
+      "step": 10128
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4742298006298004e-05,
+      "loss": 1.0316,
+      "step": 10129
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4740978121514659e-05,
+      "loss": 1.0163,
+      "step": 10130
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.473965818509994e-05,
+      "loss": 1.0893,
+      "step": 10131
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.47383381970787e-05,
+      "loss": 1.0044,
+      "step": 10132
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4737018157475816e-05,
+      "loss": 1.0022,
+      "step": 10133
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4735698066316137e-05,
+      "loss": 1.0472,
+      "step": 10134
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4734377923624546e-05,
+      "loss": 1.0985,
+      "step": 10135
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4733057729425897e-05,
+      "loss": 1.0258,
+      "step": 10136
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4731737483745068e-05,
+      "loss": 1.0636,
+      "step": 10137
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.473041718660692e-05,
+      "loss": 1.0744,
+      "step": 10138
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4729096838036329e-05,
+      "loss": 1.0563,
+      "step": 10139
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4727776438058165e-05,
+      "loss": 1.0936,
+      "step": 10140
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4726455986697298e-05,
+      "loss": 1.0734,
+      "step": 10141
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4725135483978605e-05,
+      "loss": 1.0878,
+      "step": 10142
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4723814929926959e-05,
+      "loss": 1.0563,
+      "step": 10143
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4722494324567229e-05,
+      "loss": 1.0283,
+      "step": 10144
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.47211736679243e-05,
+      "loss": 1.0225,
+      "step": 10145
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4719852960023043e-05,
+      "loss": 1.0306,
+      "step": 10146
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4718532200888338e-05,
+      "loss": 1.0453,
+      "step": 10147
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4717211390545064e-05,
+      "loss": 1.1122,
+      "step": 10148
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4715890529018102e-05,
+      "loss": 1.1,
+      "step": 10149
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.471456961633233e-05,
+      "loss": 1.1663,
+      "step": 10150
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.471324865251263e-05,
+      "loss": 1.1001,
+      "step": 10151
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4711927637583891e-05,
+      "loss": 1.0667,
+      "step": 10152
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4710606571570988e-05,
+      "loss": 1.0379,
+      "step": 10153
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4709285454498812e-05,
+      "loss": 1.0533,
+      "step": 10154
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4707964286392245e-05,
+      "loss": 1.0575,
+      "step": 10155
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4706643067276176e-05,
+      "loss": 1.0447,
+      "step": 10156
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.470532179717549e-05,
+      "loss": 1.1029,
+      "step": 10157
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4704000476115079e-05,
+      "loss": 1.0652,
+      "step": 10158
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.470267910411983e-05,
+      "loss": 1.0688,
+      "step": 10159
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4701357681214637e-05,
+      "loss": 1.0457,
+      "step": 10160
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4700036207424384e-05,
+      "loss": 1.0632,
+      "step": 10161
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.469871468277397e-05,
+      "loss": 1.0004,
+      "step": 10162
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4697393107288285e-05,
+      "loss": 1.0528,
+      "step": 10163
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4696071480992228e-05,
+      "loss": 1.0788,
+      "step": 10164
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4694749803910689e-05,
+      "loss": 1.0842,
+      "step": 10165
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4693428076068567e-05,
+      "loss": 1.0317,
+      "step": 10166
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4692106297490755e-05,
+      "loss": 1.0312,
+      "step": 10167
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4690784468202159e-05,
+      "loss": 1.0385,
+      "step": 10168
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.468946258822767e-05,
+      "loss": 1.1004,
+      "step": 10169
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4688140657592194e-05,
+      "loss": 1.0647,
+      "step": 10170
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.468681867632063e-05,
+      "loss": 1.1192,
+      "step": 10171
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4685496644437878e-05,
+      "loss": 1.0873,
+      "step": 10172
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4684174561968845e-05,
+      "loss": 1.1049,
+      "step": 10173
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4682852428938428e-05,
+      "loss": 1.1311,
+      "step": 10174
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.468153024537154e-05,
+      "loss": 1.0318,
+      "step": 10175
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.468020801129308e-05,
+      "loss": 1.1136,
+      "step": 10176
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.467888572672796e-05,
+      "loss": 1.0373,
+      "step": 10177
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4677563391701083e-05,
+      "loss": 1.0409,
+      "step": 10178
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4676241006237367e-05,
+      "loss": 1.0741,
+      "step": 10179
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4674918570361705e-05,
+      "loss": 1.0709,
+      "step": 10180
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4673596084099023e-05,
+      "loss": 1.0415,
+      "step": 10181
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4672273547474224e-05,
+      "loss": 1.0811,
+      "step": 10182
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4670950960512225e-05,
+      "loss": 1.0315,
+      "step": 10183
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4669628323237936e-05,
+      "loss": 1.0423,
+      "step": 10184
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4668305635676274e-05,
+      "loss": 1.0923,
+      "step": 10185
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.466698289785215e-05,
+      "loss": 1.0691,
+      "step": 10186
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4665660109790484e-05,
+      "loss": 1.1136,
+      "step": 10187
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4664337271516193e-05,
+      "loss": 1.1114,
+      "step": 10188
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4663014383054194e-05,
+      "loss": 1.0919,
+      "step": 10189
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4661691444429408e-05,
+      "loss": 0.9831,
+      "step": 10190
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4660368455666751e-05,
+      "loss": 1.15,
+      "step": 10191
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4659045416791148e-05,
+      "loss": 1.1217,
+      "step": 10192
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.465772232782752e-05,
+      "loss": 1.0222,
+      "step": 10193
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4656399188800787e-05,
+      "loss": 1.0282,
+      "step": 10194
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4655075999735875e-05,
+      "loss": 1.079,
+      "step": 10195
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4653752760657708e-05,
+      "loss": 1.0559,
+      "step": 10196
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4652429471591214e-05,
+      "loss": 1.0584,
+      "step": 10197
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4651106132561316e-05,
+      "loss": 1.0998,
+      "step": 10198
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4649782743592944e-05,
+      "loss": 1.1131,
+      "step": 10199
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4648459304711026e-05,
+      "loss": 1.0654,
+      "step": 10200
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.464713581594049e-05,
+      "loss": 1.0253,
+      "step": 10201
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.464581227730627e-05,
+      "loss": 1.0254,
+      "step": 10202
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4644488688833292e-05,
+      "loss": 1.0577,
+      "step": 10203
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4643165050546495e-05,
+      "loss": 1.0232,
+      "step": 10204
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4641841362470804e-05,
+      "loss": 1.0294,
+      "step": 10205
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.464051762463116e-05,
+      "loss": 1.0145,
+      "step": 10206
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4639193837052491e-05,
+      "loss": 1.0812,
+      "step": 10207
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4637869999759741e-05,
+      "loss": 1.0284,
+      "step": 10208
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4636546112777843e-05,
+      "loss": 1.0393,
+      "step": 10209
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4635222176131735e-05,
+      "loss": 1.1033,
+      "step": 10210
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4633898189846353e-05,
+      "loss": 1.031,
+      "step": 10211
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4632574153946646e-05,
+      "loss": 1.0266,
+      "step": 10212
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4631250068457543e-05,
+      "loss": 1.135,
+      "step": 10213
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.462992593340399e-05,
+      "loss": 1.1406,
+      "step": 10214
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4628601748810936e-05,
+      "loss": 1.1024,
+      "step": 10215
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4627277514703313e-05,
+      "loss": 1.0814,
+      "step": 10216
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4625953231106075e-05,
+      "loss": 1.0567,
+      "step": 10217
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.462462889804416e-05,
+      "loss": 1.0558,
+      "step": 10218
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4623304515542523e-05,
+      "loss": 1.1267,
+      "step": 10219
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.46219800836261e-05,
+      "loss": 1.0311,
+      "step": 10220
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.462065560231985e-05,
+      "loss": 1.0317,
+      "step": 10221
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4619331071648715e-05,
+      "loss": 1.0937,
+      "step": 10222
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4618006491637648e-05,
+      "loss": 1.1047,
+      "step": 10223
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4616681862311599e-05,
+      "loss": 1.0585,
+      "step": 10224
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.461535718369552e-05,
+      "loss": 1.012,
+      "step": 10225
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4614032455814362e-05,
+      "loss": 1.1257,
+      "step": 10226
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4612707678693081e-05,
+      "loss": 1.0577,
+      "step": 10227
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4611382852356632e-05,
+      "loss": 0.9769,
+      "step": 10228
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.461005797682997e-05,
+      "loss": 1.0948,
+      "step": 10229
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4608733052138049e-05,
+      "loss": 1.1905,
+      "step": 10230
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.460740807830583e-05,
+      "loss": 1.0385,
+      "step": 10231
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.460608305535827e-05,
+      "loss": 1.0758,
+      "step": 10232
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4604757983320325e-05,
+      "loss": 1.0472,
+      "step": 10233
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4603432862216959e-05,
+      "loss": 1.0275,
+      "step": 10234
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4602107692073134e-05,
+      "loss": 1.0954,
+      "step": 10235
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.460078247291381e-05,
+      "loss": 1.059,
+      "step": 10236
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4599457204763949e-05,
+      "loss": 0.9962,
+      "step": 10237
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4598131887648516e-05,
+      "loss": 1.0674,
+      "step": 10238
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.459680652159248e-05,
+      "loss": 1.0536,
+      "step": 10239
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4595481106620798e-05,
+      "loss": 1.0967,
+      "step": 10240
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4594155642758443e-05,
+      "loss": 1.1536,
+      "step": 10241
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.459283013003038e-05,
+      "loss": 1.0671,
+      "step": 10242
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4591504568461578e-05,
+      "loss": 1.1353,
+      "step": 10243
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.459017895807701e-05,
+      "loss": 1.1029,
+      "step": 10244
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4588853298901639e-05,
+      "loss": 1.0738,
+      "step": 10245
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4587527590960443e-05,
+      "loss": 1.0022,
+      "step": 10246
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4586201834278388e-05,
+      "loss": 1.0567,
+      "step": 10247
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4584876028880455e-05,
+      "loss": 1.1194,
+      "step": 10248
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.458355017479161e-05,
+      "loss": 1.0157,
+      "step": 10249
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4582224272036833e-05,
+      "loss": 1.0886,
+      "step": 10250
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4580898320641097e-05,
+      "loss": 1.0736,
+      "step": 10251
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.457957232062938e-05,
+      "loss": 1.0128,
+      "step": 10252
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4578246272026658e-05,
+      "loss": 1.0293,
+      "step": 10253
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4576920174857916e-05,
+      "loss": 1.0322,
+      "step": 10254
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4575594029148125e-05,
+      "loss": 1.0203,
+      "step": 10255
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.457426783492227e-05,
+      "loss": 1.0737,
+      "step": 10256
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4572941592205331e-05,
+      "loss": 1.1163,
+      "step": 10257
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4571615301022288e-05,
+      "loss": 1.1097,
+      "step": 10258
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4570288961398132e-05,
+      "loss": 1.0235,
+      "step": 10259
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4568962573357837e-05,
+      "loss": 1.0308,
+      "step": 10260
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4567636136926397e-05,
+      "loss": 0.9906,
+      "step": 10261
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4566309652128788e-05,
+      "loss": 1.0651,
+      "step": 10262
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.456498311899001e-05,
+      "loss": 0.9789,
+      "step": 10263
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4563656537535035e-05,
+      "loss": 1.0527,
+      "step": 10264
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4562329907788868e-05,
+      "loss": 1.1062,
+      "step": 10265
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4561003229776484e-05,
+      "loss": 1.0591,
+      "step": 10266
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.455967650352288e-05,
+      "loss": 1.1151,
+      "step": 10267
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4558349729053052e-05,
+      "loss": 1.1358,
+      "step": 10268
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4557022906391983e-05,
+      "loss": 1.1044,
+      "step": 10269
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.455569603556467e-05,
+      "loss": 1.1766,
+      "step": 10270
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4554369116596109e-05,
+      "loss": 1.0154,
+      "step": 10271
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4553042149511294e-05,
+      "loss": 1.058,
+      "step": 10272
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4551715134335218e-05,
+      "loss": 0.9837,
+      "step": 10273
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4550388071092882e-05,
+      "loss": 1.0926,
+      "step": 10274
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.454906095980928e-05,
+      "loss": 1.1263,
+      "step": 10275
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4547733800509414e-05,
+      "loss": 1.1126,
+      "step": 10276
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4546406593218281e-05,
+      "loss": 1.002,
+      "step": 10277
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4545079337960882e-05,
+      "loss": 1.0552,
+      "step": 10278
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.454375203476222e-05,
+      "loss": 1.0586,
+      "step": 10279
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4542424683647294e-05,
+      "loss": 1.1144,
+      "step": 10280
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.454109728464111e-05,
+      "loss": 1.047,
+      "step": 10281
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4539769837768671e-05,
+      "loss": 1.102,
+      "step": 10282
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.453844234305498e-05,
+      "loss": 1.106,
+      "step": 10283
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4537114800525047e-05,
+      "loss": 1.0621,
+      "step": 10284
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4535787210203876e-05,
+      "loss": 1.0836,
+      "step": 10285
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4534459572116475e-05,
+      "loss": 1.143,
+      "step": 10286
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.453313188628785e-05,
+      "loss": 1.1589,
+      "step": 10287
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4531804152743017e-05,
+      "loss": 1.0572,
+      "step": 10288
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4530476371506975e-05,
+      "loss": 1.0283,
+      "step": 10289
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.452914854260475e-05,
+      "loss": 1.141,
+      "step": 10290
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.452782066606134e-05,
+      "loss": 1.0736,
+      "step": 10291
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.452649274190177e-05,
+      "loss": 1.0272,
+      "step": 10292
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4525164770151044e-05,
+      "loss": 1.0712,
+      "step": 10293
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4523836750834185e-05,
+      "loss": 1.036,
+      "step": 10294
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.45225086839762e-05,
+      "loss": 1.1096,
+      "step": 10295
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4521180569602116e-05,
+      "loss": 1.1064,
+      "step": 10296
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4519852407736943e-05,
+      "loss": 1.0357,
+      "step": 10297
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4518524198405698e-05,
+      "loss": 1.0787,
+      "step": 10298
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4517195941633409e-05,
+      "loss": 1.0579,
+      "step": 10299
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4515867637445087e-05,
+      "loss": 1.0093,
+      "step": 10300
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4514539285865759e-05,
+      "loss": 1.1,
+      "step": 10301
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.451321088692044e-05,
+      "loss": 1.0014,
+      "step": 10302
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4511882440634162e-05,
+      "loss": 1.0442,
+      "step": 10303
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4510553947031942e-05,
+      "loss": 1.0714,
+      "step": 10304
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4509225406138808e-05,
+      "loss": 1.0408,
+      "step": 10305
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4507896817979784e-05,
+      "loss": 1.0406,
+      "step": 10306
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.45065681825799e-05,
+      "loss": 1.014,
+      "step": 10307
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4505239499964178e-05,
+      "loss": 1.0226,
+      "step": 10308
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4503910770157649e-05,
+      "loss": 0.988,
+      "step": 10309
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.450258199318534e-05,
+      "loss": 1.1281,
+      "step": 10310
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4501253169072285e-05,
+      "loss": 1.0157,
+      "step": 10311
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4499924297843509e-05,
+      "loss": 1.0502,
+      "step": 10312
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4498595379524047e-05,
+      "loss": 1.0255,
+      "step": 10313
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4497266414138934e-05,
+      "loss": 1.0827,
+      "step": 10314
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4495937401713201e-05,
+      "loss": 1.0111,
+      "step": 10315
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.449460834227188e-05,
+      "loss": 0.9403,
+      "step": 10316
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4493279235840012e-05,
+      "loss": 1.1533,
+      "step": 10317
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.449195008244263e-05,
+      "loss": 1.1218,
+      "step": 10318
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.449062088210477e-05,
+      "loss": 1.0778,
+      "step": 10319
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4489291634851469e-05,
+      "loss": 1.0519,
+      "step": 10320
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.448796234070777e-05,
+      "loss": 1.0114,
+      "step": 10321
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4486632999698708e-05,
+      "loss": 1.0409,
+      "step": 10322
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.448530361184933e-05,
+      "loss": 1.0519,
+      "step": 10323
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4483974177184672e-05,
+      "loss": 1.0106,
+      "step": 10324
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4482644695729777e-05,
+      "loss": 1.0853,
+      "step": 10325
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.448131516750969e-05,
+      "loss": 1.0462,
+      "step": 10326
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4479985592549455e-05,
+      "loss": 1.0756,
+      "step": 10327
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4478655970874119e-05,
+      "loss": 1.1482,
+      "step": 10328
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4477326302508719e-05,
+      "loss": 0.9478,
+      "step": 10329
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.447599658747831e-05,
+      "loss": 1.05,
+      "step": 10330
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4474666825807939e-05,
+      "loss": 1.0572,
+      "step": 10331
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4473337017522652e-05,
+      "loss": 1.1202,
+      "step": 10332
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4472007162647499e-05,
+      "loss": 1.0315,
+      "step": 10333
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4470677261207533e-05,
+      "loss": 1.104,
+      "step": 10334
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4469347313227799e-05,
+      "loss": 1.0249,
+      "step": 10335
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4468017318733357e-05,
+      "loss": 0.9392,
+      "step": 10336
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4466687277749253e-05,
+      "loss": 1.0202,
+      "step": 10337
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4465357190300549e-05,
+      "loss": 1.1453,
+      "step": 10338
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4464027056412289e-05,
+      "loss": 1.0695,
+      "step": 10339
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4462696876109534e-05,
+      "loss": 1.0448,
+      "step": 10340
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.446136664941734e-05,
+      "loss": 1.0776,
+      "step": 10341
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4460036376360768e-05,
+      "loss": 0.9655,
+      "step": 10342
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.445870605696487e-05,
+      "loss": 1.0101,
+      "step": 10343
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4457375691254706e-05,
+      "loss": 1.0889,
+      "step": 10344
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4456045279255341e-05,
+      "loss": 1.0734,
+      "step": 10345
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4454714820991828e-05,
+      "loss": 1.1108,
+      "step": 10346
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4453384316489238e-05,
+      "loss": 1.0553,
+      "step": 10347
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4452053765772623e-05,
+      "loss": 1.0584,
+      "step": 10348
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4450723168867058e-05,
+      "loss": 1.085,
+      "step": 10349
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4449392525797597e-05,
+      "loss": 1.1179,
+      "step": 10350
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4448061836589309e-05,
+      "loss": 1.0926,
+      "step": 10351
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4446731101267261e-05,
+      "loss": 0.9855,
+      "step": 10352
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.444540031985652e-05,
+      "loss": 1.1201,
+      "step": 10353
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4444069492382151e-05,
+      "loss": 1.0371,
+      "step": 10354
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4442738618869224e-05,
+      "loss": 1.0161,
+      "step": 10355
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.444140769934281e-05,
+      "loss": 1.0245,
+      "step": 10356
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4440076733827978e-05,
+      "loss": 1.0687,
+      "step": 10357
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.44387457223498e-05,
+      "loss": 0.9982,
+      "step": 10358
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4437414664933347e-05,
+      "loss": 0.9849,
+      "step": 10359
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.443608356160369e-05,
+      "loss": 1.0233,
+      "step": 10360
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4434752412385908e-05,
+      "loss": 1.0373,
+      "step": 10361
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4433421217305072e-05,
+      "loss": 1.0524,
+      "step": 10362
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.443208997638626e-05,
+      "loss": 1.1154,
+      "step": 10363
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4430758689654542e-05,
+      "loss": 1.1621,
+      "step": 10364
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4429427357135001e-05,
+      "loss": 1.068,
+      "step": 10365
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4428095978852717e-05,
+      "loss": 1.0662,
+      "step": 10366
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4426764554832764e-05,
+      "loss": 1.1572,
+      "step": 10367
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4425433085100224e-05,
+      "loss": 1.1114,
+      "step": 10368
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4424101569680178e-05,
+      "loss": 0.9945,
+      "step": 10369
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4422770008597706e-05,
+      "loss": 1.0554,
+      "step": 10370
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4421438401877892e-05,
+      "loss": 1.1353,
+      "step": 10371
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4420106749545819e-05,
+      "loss": 1.0591,
+      "step": 10372
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4418775051626568e-05,
+      "loss": 1.0761,
+      "step": 10373
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.441744330814523e-05,
+      "loss": 1.051,
+      "step": 10374
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4416111519126885e-05,
+      "loss": 1.1306,
+      "step": 10375
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4414779684596625e-05,
+      "loss": 1.0938,
+      "step": 10376
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4413447804579532e-05,
+      "loss": 0.9399,
+      "step": 10377
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4412115879100701e-05,
+      "loss": 1.0965,
+      "step": 10378
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4410783908185212e-05,
+      "loss": 1.0713,
+      "step": 10379
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4409451891858169e-05,
+      "loss": 0.9767,
+      "step": 10380
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4408119830144649e-05,
+      "loss": 1.085,
+      "step": 10381
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.440678772306975e-05,
+      "loss": 0.992,
+      "step": 10382
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4405455570658565e-05,
+      "loss": 1.0727,
+      "step": 10383
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4404123372936185e-05,
+      "loss": 1.1722,
+      "step": 10384
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.440279112992771e-05,
+      "loss": 1.0251,
+      "step": 10385
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4401458841658226e-05,
+      "loss": 1.0921,
+      "step": 10386
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4400126508152838e-05,
+      "loss": 1.1369,
+      "step": 10387
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.439879412943664e-05,
+      "loss": 1.0464,
+      "step": 10388
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4397461705534727e-05,
+      "loss": 1.0221,
+      "step": 10389
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.43961292364722e-05,
+      "loss": 1.0797,
+      "step": 10390
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4394796722274163e-05,
+      "loss": 1.0363,
+      "step": 10391
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4393464162965707e-05,
+      "loss": 0.9746,
+      "step": 10392
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.439213155857194e-05,
+      "loss": 1.1428,
+      "step": 10393
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4390798909117963e-05,
+      "loss": 1.0933,
+      "step": 10394
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4389466214628876e-05,
+      "loss": 1.0538,
+      "step": 10395
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4388133475129783e-05,
+      "loss": 1.0764,
+      "step": 10396
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4386800690645794e-05,
+      "loss": 1.1273,
+      "step": 10397
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4385467861202009e-05,
+      "loss": 1.0283,
+      "step": 10398
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.438413498682354e-05,
+      "loss": 1.1548,
+      "step": 10399
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4382802067535486e-05,
+      "loss": 1.0588,
+      "step": 10400
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.438146910336296e-05,
+      "loss": 1.0545,
+      "step": 10401
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.438013609433107e-05,
+      "loss": 1.0284,
+      "step": 10402
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4378803040464927e-05,
+      "loss": 1.0077,
+      "step": 10403
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4377469941789638e-05,
+      "loss": 1.0529,
+      "step": 10404
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.437613679833032e-05,
+      "loss": 1.0849,
+      "step": 10405
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.437480361011208e-05,
+      "loss": 1.0973,
+      "step": 10406
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4373470377160033e-05,
+      "loss": 1.0595,
+      "step": 10407
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4372137099499292e-05,
+      "loss": 1.1402,
+      "step": 10408
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4370803777154975e-05,
+      "loss": 0.9437,
+      "step": 10409
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4369470410152194e-05,
+      "loss": 1.0964,
+      "step": 10410
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4368136998516066e-05,
+      "loss": 0.9806,
+      "step": 10411
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4366803542271712e-05,
+      "loss": 1.0797,
+      "step": 10412
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4365470041444243e-05,
+      "loss": 1.1082,
+      "step": 10413
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4364136496058786e-05,
+      "loss": 1.0637,
+      "step": 10414
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4362802906140451e-05,
+      "loss": 0.9938,
+      "step": 10415
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4361469271714371e-05,
+      "loss": 1.0552,
+      "step": 10416
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4360135592805655e-05,
+      "loss": 1.0618,
+      "step": 10417
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4358801869439439e-05,
+      "loss": 1.1103,
+      "step": 10418
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4357468101640828e-05,
+      "loss": 1.0208,
+      "step": 10419
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4356134289434964e-05,
+      "loss": 1.0455,
+      "step": 10420
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.435480043284696e-05,
+      "loss": 1.0276,
+      "step": 10421
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.435346653190195e-05,
+      "loss": 1.0804,
+      "step": 10422
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4352132586625051e-05,
+      "loss": 1.0161,
+      "step": 10423
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.43507985970414e-05,
+      "loss": 1.0476,
+      "step": 10424
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4349464563176118e-05,
+      "loss": 1.0487,
+      "step": 10425
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4348130485054337e-05,
+      "loss": 1.0487,
+      "step": 10426
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4346796362701187e-05,
+      "loss": 1.1349,
+      "step": 10427
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4345462196141796e-05,
+      "loss": 1.049,
+      "step": 10428
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4344127985401301e-05,
+      "loss": 1.0607,
+      "step": 10429
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4342793730504829e-05,
+      "loss": 1.0387,
+      "step": 10430
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4341459431477515e-05,
+      "loss": 1.0877,
+      "step": 10431
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4340125088344496e-05,
+      "loss": 1.0858,
+      "step": 10432
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.43387907011309e-05,
+      "loss": 1.0888,
+      "step": 10433
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4337456269861869e-05,
+      "loss": 1.0337,
+      "step": 10434
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4336121794562537e-05,
+      "loss": 1.0704,
+      "step": 10435
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.433478727525804e-05,
+      "loss": 1.1222,
+      "step": 10436
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4333452711973518e-05,
+      "loss": 1.0096,
+      "step": 10437
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.433211810473411e-05,
+      "loss": 1.0086,
+      "step": 10438
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4330783453564954e-05,
+      "loss": 1.0861,
+      "step": 10439
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4329448758491194e-05,
+      "loss": 1.0763,
+      "step": 10440
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4328114019537968e-05,
+      "loss": 1.0518,
+      "step": 10441
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4326779236730419e-05,
+      "loss": 1.0835,
+      "step": 10442
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4325444410093692e-05,
+      "loss": 1.0768,
+      "step": 10443
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4324109539652927e-05,
+      "loss": 1.1115,
+      "step": 10444
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4322774625433275e-05,
+      "loss": 1.0081,
+      "step": 10445
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4321439667459875e-05,
+      "loss": 1.0826,
+      "step": 10446
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4320104665757879e-05,
+      "loss": 1.0696,
+      "step": 10447
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4318769620352429e-05,
+      "loss": 1.1263,
+      "step": 10448
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4317434531268674e-05,
+      "loss": 1.1077,
+      "step": 10449
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4316099398531769e-05,
+      "loss": 1.0959,
+      "step": 10450
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4314764222166854e-05,
+      "loss": 1.0098,
+      "step": 10451
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4313429002199087e-05,
+      "loss": 1.0351,
+      "step": 10452
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4312093738653619e-05,
+      "loss": 1.0902,
+      "step": 10453
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.43107584315556e-05,
+      "loss": 0.978,
+      "step": 10454
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.430942308093018e-05,
+      "loss": 1.0769,
+      "step": 10455
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.430808768680252e-05,
+      "loss": 1.0095,
+      "step": 10456
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4306752249197766e-05,
+      "loss": 1.0351,
+      "step": 10457
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4305416768141082e-05,
+      "loss": 1.0465,
+      "step": 10458
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4304081243657615e-05,
+      "loss": 1.1245,
+      "step": 10459
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4302745675772534e-05,
+      "loss": 1.0732,
+      "step": 10460
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4301410064510984e-05,
+      "loss": 1.0616,
+      "step": 10461
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4300074409898133e-05,
+      "loss": 1.089,
+      "step": 10462
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4298738711959134e-05,
+      "loss": 1.1031,
+      "step": 10463
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4297402970719156e-05,
+      "loss": 1.1128,
+      "step": 10464
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4296067186203353e-05,
+      "loss": 1.0542,
+      "step": 10465
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4294731358436886e-05,
+      "loss": 0.985,
+      "step": 10466
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4293395487444922e-05,
+      "loss": 1.1996,
+      "step": 10467
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4292059573252625e-05,
+      "loss": 1.0821,
+      "step": 10468
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4290723615885155e-05,
+      "loss": 1.1265,
+      "step": 10469
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4289387615367683e-05,
+      "loss": 1.107,
+      "step": 10470
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4288051571725368e-05,
+      "loss": 1.0413,
+      "step": 10471
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4286715484983382e-05,
+      "loss": 1.085,
+      "step": 10472
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4285379355166893e-05,
+      "loss": 1.0015,
+      "step": 10473
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4284043182301065e-05,
+      "loss": 1.103,
+      "step": 10474
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4282706966411073e-05,
+      "loss": 1.0891,
+      "step": 10475
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4281370707522082e-05,
+      "loss": 1.0318,
+      "step": 10476
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4280034405659266e-05,
+      "loss": 1.0568,
+      "step": 10477
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4278698060847795e-05,
+      "loss": 1.0666,
+      "step": 10478
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4277361673112844e-05,
+      "loss": 1.1024,
+      "step": 10479
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4276025242479583e-05,
+      "loss": 0.9836,
+      "step": 10480
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4274688768973187e-05,
+      "loss": 1.0448,
+      "step": 10481
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4273352252618833e-05,
+      "loss": 1.016,
+      "step": 10482
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4272015693441695e-05,
+      "loss": 1.0114,
+      "step": 10483
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4270679091466952e-05,
+      "loss": 1.0405,
+      "step": 10484
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4269342446719778e-05,
+      "loss": 1.0219,
+      "step": 10485
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4268005759225351e-05,
+      "loss": 1.0208,
+      "step": 10486
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4266669029008854e-05,
+      "loss": 1.1043,
+      "step": 10487
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4265332256095463e-05,
+      "loss": 0.9891,
+      "step": 10488
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4263995440510362e-05,
+      "loss": 1.0493,
+      "step": 10489
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4262658582278725e-05,
+      "loss": 1.0668,
+      "step": 10490
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4261321681425745e-05,
+      "loss": 1.0131,
+      "step": 10491
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4259984737976597e-05,
+      "loss": 1.0118,
+      "step": 10492
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4258647751956466e-05,
+      "loss": 1.0562,
+      "step": 10493
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.425731072339054e-05,
+      "loss": 1.0507,
+      "step": 10494
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4255973652304003e-05,
+      "loss": 1.1183,
+      "step": 10495
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4254636538722039e-05,
+      "loss": 1.0558,
+      "step": 10496
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4253299382669836e-05,
+      "loss": 1.1022,
+      "step": 10497
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4251962184172586e-05,
+      "loss": 1.1591,
+      "step": 10498
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4250624943255466e-05,
+      "loss": 1.1444,
+      "step": 10499
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4249287659943681e-05,
+      "loss": 1.0496,
+      "step": 10500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4247950334262408e-05,
+      "loss": 1.0601,
+      "step": 10501
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.424661296623685e-05,
+      "loss": 0.974,
+      "step": 10502
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4245275555892186e-05,
+      "loss": 0.9637,
+      "step": 10503
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4243938103253618e-05,
+      "loss": 1.0804,
+      "step": 10504
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4242600608346333e-05,
+      "loss": 1.0533,
+      "step": 10505
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4241263071195534e-05,
+      "loss": 1.0598,
+      "step": 10506
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4239925491826407e-05,
+      "loss": 1.1162,
+      "step": 10507
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4238587870264151e-05,
+      "loss": 1.1232,
+      "step": 10508
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4237250206533963e-05,
+      "loss": 1.0562,
+      "step": 10509
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.423591250066104e-05,
+      "loss": 0.9585,
+      "step": 10510
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.423457475267058e-05,
+      "loss": 1.0489,
+      "step": 10511
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4233236962587782e-05,
+      "loss": 1.0696,
+      "step": 10512
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4231899130437847e-05,
+      "loss": 1.157,
+      "step": 10513
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4230561256245973e-05,
+      "loss": 1.1185,
+      "step": 10514
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4229223340037364e-05,
+      "loss": 1.1174,
+      "step": 10515
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.422788538183722e-05,
+      "loss": 1.133,
+      "step": 10516
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4226547381670748e-05,
+      "loss": 1.0267,
+      "step": 10517
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4225209339563144e-05,
+      "loss": 0.9657,
+      "step": 10518
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.422387125553962e-05,
+      "loss": 1.068,
+      "step": 10519
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4222533129625377e-05,
+      "loss": 1.0596,
+      "step": 10520
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4221194961845623e-05,
+      "loss": 1.0708,
+      "step": 10521
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4219856752225565e-05,
+      "loss": 1.0601,
+      "step": 10522
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4218518500790407e-05,
+      "loss": 1.0442,
+      "step": 10523
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4217180207565365e-05,
+      "loss": 1.0199,
+      "step": 10524
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.421584187257564e-05,
+      "loss": 1.0459,
+      "step": 10525
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4214503495846447e-05,
+      "loss": 0.9866,
+      "step": 10526
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4213165077402996e-05,
+      "loss": 1.0522,
+      "step": 10527
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4211826617270494e-05,
+      "loss": 1.0176,
+      "step": 10528
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4210488115474163e-05,
+      "loss": 0.9918,
+      "step": 10529
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4209149572039206e-05,
+      "loss": 1.0711,
+      "step": 10530
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4207810986990846e-05,
+      "loss": 1.017,
+      "step": 10531
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4206472360354286e-05,
+      "loss": 0.9783,
+      "step": 10532
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4205133692154756e-05,
+      "loss": 1.1876,
+      "step": 10533
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.420379498241746e-05,
+      "loss": 1.0599,
+      "step": 10534
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4202456231167622e-05,
+      "loss": 1.0084,
+      "step": 10535
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4201117438430456e-05,
+      "loss": 1.134,
+      "step": 10536
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4199778604231184e-05,
+      "loss": 1.0899,
+      "step": 10537
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4198439728595025e-05,
+      "loss": 1.0492,
+      "step": 10538
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4197100811547195e-05,
+      "loss": 1.1119,
+      "step": 10539
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4195761853112921e-05,
+      "loss": 1.1374,
+      "step": 10540
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.419442285331742e-05,
+      "loss": 1.0624,
+      "step": 10541
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4193083812185918e-05,
+      "loss": 1.034,
+      "step": 10542
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4191744729743635e-05,
+      "loss": 1.0683,
+      "step": 10543
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4190405606015798e-05,
+      "loss": 1.0756,
+      "step": 10544
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4189066441027628e-05,
+      "loss": 1.1568,
+      "step": 10545
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4187727234804355e-05,
+      "loss": 1.0354,
+      "step": 10546
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4186387987371204e-05,
+      "loss": 1.084,
+      "step": 10547
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4185048698753403e-05,
+      "loss": 1.0529,
+      "step": 10548
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4183709368976177e-05,
+      "loss": 1.1319,
+      "step": 10549
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4182369998064755e-05,
+      "loss": 1.0666,
+      "step": 10550
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.418103058604437e-05,
+      "loss": 1.0341,
+      "step": 10551
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4179691132940252e-05,
+      "loss": 1.0662,
+      "step": 10552
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4178351638777628e-05,
+      "loss": 1.1153,
+      "step": 10553
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4177012103581732e-05,
+      "loss": 1.1084,
+      "step": 10554
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.41756725273778e-05,
+      "loss": 1.1098,
+      "step": 10555
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.417433291019106e-05,
+      "loss": 1.0323,
+      "step": 10556
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4172993252046746e-05,
+      "loss": 1.0148,
+      "step": 10557
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.41716535529701e-05,
+      "loss": 1.049,
+      "step": 10558
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4170313812986349e-05,
+      "loss": 1.0306,
+      "step": 10559
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4168974032120736e-05,
+      "loss": 1.1718,
+      "step": 10560
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4167634210398496e-05,
+      "loss": 1.0406,
+      "step": 10561
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4166294347844866e-05,
+      "loss": 1.0296,
+      "step": 10562
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4164954444485086e-05,
+      "loss": 1.0753,
+      "step": 10563
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4163614500344396e-05,
+      "loss": 1.1452,
+      "step": 10564
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4162274515448033e-05,
+      "loss": 1.0154,
+      "step": 10565
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4160934489821244e-05,
+      "loss": 1.0628,
+      "step": 10566
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4159594423489263e-05,
+      "loss": 1.1052,
+      "step": 10567
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4158254316477339e-05,
+      "loss": 1.0787,
+      "step": 10568
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4156914168810715e-05,
+      "loss": 1.0807,
+      "step": 10569
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4155573980514631e-05,
+      "loss": 1.0701,
+      "step": 10570
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.415423375161434e-05,
+      "loss": 1.0482,
+      "step": 10571
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4152893482135074e-05,
+      "loss": 0.9652,
+      "step": 10572
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4151553172102094e-05,
+      "loss": 1.0408,
+      "step": 10573
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4150212821540637e-05,
+      "loss": 1.0011,
+      "step": 10574
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.414887243047596e-05,
+      "loss": 1.0558,
+      "step": 10575
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.41475319989333e-05,
+      "loss": 1.034,
+      "step": 10576
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4146191526937918e-05,
+      "loss": 1.0135,
+      "step": 10577
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4144851014515055e-05,
+      "loss": 1.048,
+      "step": 10578
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.414351046168997e-05,
+      "loss": 0.9952,
+      "step": 10579
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4142169868487909e-05,
+      "loss": 1.0615,
+      "step": 10580
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4140829234934125e-05,
+      "loss": 1.0363,
+      "step": 10581
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4139488561053878e-05,
+      "loss": 1.134,
+      "step": 10582
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4138147846872412e-05,
+      "loss": 1.0764,
+      "step": 10583
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4136807092414992e-05,
+      "loss": 1.1263,
+      "step": 10584
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4135466297706864e-05,
+      "loss": 1.0076,
+      "step": 10585
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4134125462773292e-05,
+      "loss": 1.0338,
+      "step": 10586
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4132784587639527e-05,
+      "loss": 0.999,
+      "step": 10587
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4131443672330834e-05,
+      "loss": 1.0996,
+      "step": 10588
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4130102716872461e-05,
+      "loss": 1.0155,
+      "step": 10589
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4128761721289679e-05,
+      "loss": 1.1086,
+      "step": 10590
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4127420685607742e-05,
+      "loss": 1.0646,
+      "step": 10591
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.412607960985191e-05,
+      "loss": 1.1189,
+      "step": 10592
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4124738494047448e-05,
+      "loss": 1.0768,
+      "step": 10593
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4123397338219615e-05,
+      "loss": 1.0433,
+      "step": 10594
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4122056142393678e-05,
+      "loss": 1.0726,
+      "step": 10595
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4120714906594896e-05,
+      "loss": 1.0881,
+      "step": 10596
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.411937363084854e-05,
+      "loss": 1.0628,
+      "step": 10597
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.411803231517987e-05,
+      "loss": 1.0686,
+      "step": 10598
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4116690959614154e-05,
+      "loss": 1.0539,
+      "step": 10599
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.411534956417666e-05,
+      "loss": 1.121,
+      "step": 10600
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.411400812889265e-05,
+      "loss": 1.0564,
+      "step": 10601
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4112666653787403e-05,
+      "loss": 1.1035,
+      "step": 10602
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.411132513888618e-05,
+      "loss": 1.0371,
+      "step": 10603
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4109983584214252e-05,
+      "loss": 1.0627,
+      "step": 10604
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.410864198979689e-05,
+      "loss": 1.0366,
+      "step": 10605
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4107300355659368e-05,
+      "loss": 1.0456,
+      "step": 10606
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4105958681826957e-05,
+      "loss": 1.0537,
+      "step": 10607
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4104616968324927e-05,
+      "loss": 0.9829,
+      "step": 10608
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4103275215178555e-05,
+      "loss": 1.0618,
+      "step": 10609
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4101933422413115e-05,
+      "loss": 1.0657,
+      "step": 10610
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4100591590053882e-05,
+      "loss": 1.075,
+      "step": 10611
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4099249718126126e-05,
+      "loss": 1.0475,
+      "step": 10612
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4097907806655136e-05,
+      "loss": 1.079,
+      "step": 10613
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4096565855666177e-05,
+      "loss": 1.0462,
+      "step": 10614
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4095223865184538e-05,
+      "loss": 1.0567,
+      "step": 10615
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4093881835235485e-05,
+      "loss": 0.9684,
+      "step": 10616
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.409253976584431e-05,
+      "loss": 1.0398,
+      "step": 10617
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4091197657036285e-05,
+      "loss": 1.1111,
+      "step": 10618
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4089855508836699e-05,
+      "loss": 1.0355,
+      "step": 10619
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4088513321270823e-05,
+      "loss": 0.9717,
+      "step": 10620
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.408717109436395e-05,
+      "loss": 1.0735,
+      "step": 10621
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.408582882814136e-05,
+      "loss": 1.0925,
+      "step": 10622
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4084486522628331e-05,
+      "loss": 1.1487,
+      "step": 10623
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4083144177850155e-05,
+      "loss": 1.0088,
+      "step": 10624
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4081801793832115e-05,
+      "loss": 0.9971,
+      "step": 10625
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4080459370599501e-05,
+      "loss": 1.0413,
+      "step": 10626
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4079116908177592e-05,
+      "loss": 1.0193,
+      "step": 10627
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4077774406591686e-05,
+      "loss": 1.1094,
+      "step": 10628
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4076431865867061e-05,
+      "loss": 1.1283,
+      "step": 10629
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4075089286029015e-05,
+      "loss": 1.1178,
+      "step": 10630
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4073746667102832e-05,
+      "loss": 1.096,
+      "step": 10631
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4072404009113807e-05,
+      "loss": 1.1003,
+      "step": 10632
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4071061312087228e-05,
+      "loss": 1.0632,
+      "step": 10633
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4069718576048387e-05,
+      "loss": 1.0622,
+      "step": 10634
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.406837580102258e-05,
+      "loss": 1.0973,
+      "step": 10635
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.40670329870351e-05,
+      "loss": 1.0925,
+      "step": 10636
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4065690134111239e-05,
+      "loss": 1.074,
+      "step": 10637
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4064347242276292e-05,
+      "loss": 1.1284,
+      "step": 10638
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4063004311555559e-05,
+      "loss": 1.0308,
+      "step": 10639
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4061661341974333e-05,
+      "loss": 1.0401,
+      "step": 10640
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4060318333557915e-05,
+      "loss": 0.9833,
+      "step": 10641
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4058975286331596e-05,
+      "loss": 0.998,
+      "step": 10642
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4057632200320681e-05,
+      "loss": 1.0301,
+      "step": 10643
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4056289075550466e-05,
+      "loss": 1.0779,
+      "step": 10644
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4054945912046256e-05,
+      "loss": 0.9855,
+      "step": 10645
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4053602709833347e-05,
+      "loss": 1.0853,
+      "step": 10646
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4052259468937042e-05,
+      "loss": 0.9905,
+      "step": 10647
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4050916189382645e-05,
+      "loss": 1.0327,
+      "step": 10648
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4049572871195458e-05,
+      "loss": 1.0861,
+      "step": 10649
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4048229514400784e-05,
+      "loss": 1.141,
+      "step": 10650
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.404688611902393e-05,
+      "loss": 1.1279,
+      "step": 10651
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4045542685090198e-05,
+      "loss": 1.0677,
+      "step": 10652
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.40441992126249e-05,
+      "loss": 1.0288,
+      "step": 10653
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4042855701653333e-05,
+      "loss": 1.1194,
+      "step": 10654
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4041512152200815e-05,
+      "loss": 1.095,
+      "step": 10655
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4040168564292644e-05,
+      "loss": 1.0174,
+      "step": 10656
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.403882493795414e-05,
+      "loss": 1.068,
+      "step": 10657
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4037481273210603e-05,
+      "loss": 1.089,
+      "step": 10658
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4036137570087352e-05,
+      "loss": 1.0605,
+      "step": 10659
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.403479382860969e-05,
+      "loss": 1.0446,
+      "step": 10660
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4033450048802935e-05,
+      "loss": 1.0311,
+      "step": 10661
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4032106230692393e-05,
+      "loss": 1.0791,
+      "step": 10662
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4030762374303386e-05,
+      "loss": 0.999,
+      "step": 10663
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4029418479661222e-05,
+      "loss": 1.0393,
+      "step": 10664
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4028074546791215e-05,
+      "loss": 1.036,
+      "step": 10665
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4026730575718683e-05,
+      "loss": 1.0248,
+      "step": 10666
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4025386566468941e-05,
+      "loss": 1.0467,
+      "step": 10667
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4024042519067309e-05,
+      "loss": 1.0309,
+      "step": 10668
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.40226984335391e-05,
+      "loss": 1.1268,
+      "step": 10669
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4021354309909637e-05,
+      "loss": 1.0463,
+      "step": 10670
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4020010148204232e-05,
+      "loss": 1.092,
+      "step": 10671
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4018665948448215e-05,
+      "loss": 1.0677,
+      "step": 10672
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4017321710666898e-05,
+      "loss": 1.0666,
+      "step": 10673
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4015977434885607e-05,
+      "loss": 1.0185,
+      "step": 10674
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4014633121129662e-05,
+      "loss": 1.0448,
+      "step": 10675
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4013288769424383e-05,
+      "loss": 1.0588,
+      "step": 10676
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4011944379795097e-05,
+      "loss": 0.9913,
+      "step": 10677
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4010599952267128e-05,
+      "loss": 1.0767,
+      "step": 10678
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4009255486865798e-05,
+      "loss": 1.0642,
+      "step": 10679
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4007910983616435e-05,
+      "loss": 1.0987,
+      "step": 10680
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4006566442544365e-05,
+      "loss": 1.0455,
+      "step": 10681
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4005221863674914e-05,
+      "loss": 1.0798,
+      "step": 10682
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.400387724703341e-05,
+      "loss": 1.0517,
+      "step": 10683
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4002532592645181e-05,
+      "loss": 1.1157,
+      "step": 10684
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4001187900535555e-05,
+      "loss": 1.0303,
+      "step": 10685
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3999843170729865e-05,
+      "loss": 0.9869,
+      "step": 10686
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.399849840325344e-05,
+      "loss": 1.0619,
+      "step": 10687
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.399715359813161e-05,
+      "loss": 0.964,
+      "step": 10688
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3995808755389708e-05,
+      "loss": 1.048,
+      "step": 10689
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3994463875053065e-05,
+      "loss": 0.9688,
+      "step": 10690
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3993118957147018e-05,
+      "loss": 1.1452,
+      "step": 10691
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3991774001696896e-05,
+      "loss": 1.0508,
+      "step": 10692
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3990429008728036e-05,
+      "loss": 1.0107,
+      "step": 10693
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3989083978265776e-05,
+      "loss": 1.0609,
+      "step": 10694
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.398773891033545e-05,
+      "loss": 0.9833,
+      "step": 10695
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3986393804962392e-05,
+      "loss": 1.0293,
+      "step": 10696
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3985048662171947e-05,
+      "loss": 1.0377,
+      "step": 10697
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3983703481989443e-05,
+      "loss": 1.0307,
+      "step": 10698
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3982358264440227e-05,
+      "loss": 1.0546,
+      "step": 10699
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3981013009549635e-05,
+      "loss": 1.0812,
+      "step": 10700
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.397966771734301e-05,
+      "loss": 1.1364,
+      "step": 10701
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3978322387845688e-05,
+      "loss": 1.1072,
+      "step": 10702
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.397697702108302e-05,
+      "loss": 1.1574,
+      "step": 10703
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3975631617080338e-05,
+      "loss": 1.0339,
+      "step": 10704
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3974286175862992e-05,
+      "loss": 1.0048,
+      "step": 10705
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3972940697456323e-05,
+      "loss": 1.0225,
+      "step": 10706
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3971595181885677e-05,
+      "loss": 1.119,
+      "step": 10707
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3970249629176398e-05,
+      "loss": 1.0529,
+      "step": 10708
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3968904039353832e-05,
+      "loss": 1.0127,
+      "step": 10709
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3967558412443327e-05,
+      "loss": 1.1545,
+      "step": 10710
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.396621274847023e-05,
+      "loss": 1.0523,
+      "step": 10711
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3964867047459887e-05,
+      "loss": 1.0679,
+      "step": 10712
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3963521309437646e-05,
+      "loss": 1.104,
+      "step": 10713
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3962175534428865e-05,
+      "loss": 1.0945,
+      "step": 10714
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3960829722458885e-05,
+      "loss": 1.0939,
+      "step": 10715
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3959483873553058e-05,
+      "loss": 1.0208,
+      "step": 10716
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3958137987736742e-05,
+      "loss": 1.1233,
+      "step": 10717
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3956792065035281e-05,
+      "loss": 1.1209,
+      "step": 10718
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3955446105474033e-05,
+      "loss": 1.0759,
+      "step": 10719
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3954100109078349e-05,
+      "loss": 1.0262,
+      "step": 10720
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3952754075873585e-05,
+      "loss": 0.991,
+      "step": 10721
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3951408005885097e-05,
+      "loss": 1.0235,
+      "step": 10722
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3950061899138235e-05,
+      "loss": 1.0604,
+      "step": 10723
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3948715755658364e-05,
+      "loss": 1.0798,
+      "step": 10724
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3947369575470837e-05,
+      "loss": 1.0646,
+      "step": 10725
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.394602335860101e-05,
+      "loss": 0.9959,
+      "step": 10726
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3944677105074242e-05,
+      "loss": 0.9531,
+      "step": 10727
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3943330814915897e-05,
+      "loss": 1.0845,
+      "step": 10728
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3941984488151328e-05,
+      "loss": 1.0608,
+      "step": 10729
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3940638124805899e-05,
+      "loss": 1.0583,
+      "step": 10730
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3939291724904968e-05,
+      "loss": 1.0132,
+      "step": 10731
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3937945288473905e-05,
+      "loss": 1.0955,
+      "step": 10732
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3936598815538064e-05,
+      "loss": 1.1039,
+      "step": 10733
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3935252306122812e-05,
+      "loss": 0.9965,
+      "step": 10734
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3933905760253515e-05,
+      "loss": 1.0088,
+      "step": 10735
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3932559177955533e-05,
+      "loss": 1.0999,
+      "step": 10736
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3931212559254236e-05,
+      "loss": 1.0856,
+      "step": 10737
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3929865904174984e-05,
+      "loss": 1.0824,
+      "step": 10738
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3928519212743152e-05,
+      "loss": 1.0652,
+      "step": 10739
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3927172484984099e-05,
+      "loss": 1.0421,
+      "step": 10740
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.39258257209232e-05,
+      "loss": 1.1681,
+      "step": 10741
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3924478920585817e-05,
+      "loss": 0.9598,
+      "step": 10742
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3923132083997328e-05,
+      "loss": 0.9933,
+      "step": 10743
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3921785211183095e-05,
+      "loss": 0.9815,
+      "step": 10744
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3920438302168494e-05,
+      "loss": 1.1216,
+      "step": 10745
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3919091356978894e-05,
+      "loss": 1.0266,
+      "step": 10746
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.391774437563967e-05,
+      "loss": 1.0577,
+      "step": 10747
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.391639735817619e-05,
+      "loss": 1.0808,
+      "step": 10748
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3915050304613832e-05,
+      "loss": 1.0424,
+      "step": 10749
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3913703214977968e-05,
+      "loss": 0.9748,
+      "step": 10750
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3912356089293975e-05,
+      "loss": 1.0051,
+      "step": 10751
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3911008927587224e-05,
+      "loss": 1.0713,
+      "step": 10752
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3909661729883097e-05,
+      "loss": 1.0763,
+      "step": 10753
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3908314496206966e-05,
+      "loss": 1.0612,
+      "step": 10754
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3906967226584214e-05,
+      "loss": 1.1417,
+      "step": 10755
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3905619921040215e-05,
+      "loss": 1.0646,
+      "step": 10756
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3904272579600346e-05,
+      "loss": 1.0305,
+      "step": 10757
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3902925202289996e-05,
+      "loss": 1.0587,
+      "step": 10758
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3901577789134537e-05,
+      "loss": 1.0423,
+      "step": 10759
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3900230340159351e-05,
+      "loss": 1.1233,
+      "step": 10760
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3898882855389823e-05,
+      "loss": 1.0606,
+      "step": 10761
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.389753533485133e-05,
+      "loss": 0.9769,
+      "step": 10762
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3896187778569263e-05,
+      "loss": 1.0154,
+      "step": 10763
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3894840186568999e-05,
+      "loss": 0.9164,
+      "step": 10764
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3893492558875923e-05,
+      "loss": 1.0234,
+      "step": 10765
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3892144895515422e-05,
+      "loss": 1.0554,
+      "step": 10766
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3890797196512883e-05,
+      "loss": 1.0323,
+      "step": 10767
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3889449461893692e-05,
+      "loss": 1.034,
+      "step": 10768
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3888101691683232e-05,
+      "loss": 1.0611,
+      "step": 10769
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3886753885906895e-05,
+      "loss": 1.0491,
+      "step": 10770
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3885406044590066e-05,
+      "loss": 1.024,
+      "step": 10771
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3884058167758136e-05,
+      "loss": 1.1134,
+      "step": 10772
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3882710255436496e-05,
+      "loss": 1.1517,
+      "step": 10773
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3881362307650534e-05,
+      "loss": 0.9968,
+      "step": 10774
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3880014324425642e-05,
+      "loss": 1.0,
+      "step": 10775
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3878666305787214e-05,
+      "loss": 0.9708,
+      "step": 10776
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3877318251760638e-05,
+      "loss": 1.0983,
+      "step": 10777
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3875970162371312e-05,
+      "loss": 1.0551,
+      "step": 10778
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3874622037644629e-05,
+      "loss": 1.0979,
+      "step": 10779
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3873273877605977e-05,
+      "loss": 1.0337,
+      "step": 10780
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3871925682280758e-05,
+      "loss": 0.9874,
+      "step": 10781
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3870577451694362e-05,
+      "loss": 0.9953,
+      "step": 10782
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3869229185872195e-05,
+      "loss": 1.0394,
+      "step": 10783
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3867880884839641e-05,
+      "loss": 1.0235,
+      "step": 10784
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3866532548622111e-05,
+      "loss": 1.0576,
+      "step": 10785
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3865184177244992e-05,
+      "loss": 1.0892,
+      "step": 10786
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3863835770733693e-05,
+      "loss": 1.0239,
+      "step": 10787
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3862487329113606e-05,
+      "loss": 1.0526,
+      "step": 10788
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3861138852410137e-05,
+      "loss": 1.0369,
+      "step": 10789
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3859790340648679e-05,
+      "loss": 1.0261,
+      "step": 10790
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3858441793854645e-05,
+      "loss": 1.0631,
+      "step": 10791
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3857093212053427e-05,
+      "loss": 1.0593,
+      "step": 10792
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3855744595270434e-05,
+      "loss": 1.0081,
+      "step": 10793
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.385439594353107e-05,
+      "loss": 1.0909,
+      "step": 10794
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3853047256860733e-05,
+      "loss": 1.0642,
+      "step": 10795
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3851698535284834e-05,
+      "loss": 1.0296,
+      "step": 10796
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3850349778828778e-05,
+      "loss": 1.0847,
+      "step": 10797
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.384900098751797e-05,
+      "loss": 1.0944,
+      "step": 10798
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3847652161377818e-05,
+      "loss": 1.0828,
+      "step": 10799
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.384630330043373e-05,
+      "loss": 0.9874,
+      "step": 10800
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.384495440471111e-05,
+      "loss": 1.121,
+      "step": 10801
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3843605474235373e-05,
+      "loss": 1.0927,
+      "step": 10802
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3842256509031926e-05,
+      "loss": 1.0585,
+      "step": 10803
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3840907509126177e-05,
+      "loss": 1.085,
+      "step": 10804
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.383955847454354e-05,
+      "loss": 1.029,
+      "step": 10805
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3838209405309426e-05,
+      "loss": 1.0627,
+      "step": 10806
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3836860301449246e-05,
+      "loss": 1.0709,
+      "step": 10807
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3835511162988412e-05,
+      "loss": 0.9452,
+      "step": 10808
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3834161989952344e-05,
+      "loss": 1.035,
+      "step": 10809
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3832812782366448e-05,
+      "loss": 1.0972,
+      "step": 10810
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3831463540256142e-05,
+      "loss": 1.0652,
+      "step": 10811
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3830114263646843e-05,
+      "loss": 1.0841,
+      "step": 10812
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3828764952563963e-05,
+      "loss": 1.0035,
+      "step": 10813
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3827415607032923e-05,
+      "loss": 1.0334,
+      "step": 10814
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.382606622707914e-05,
+      "loss": 1.0626,
+      "step": 10815
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3824716812728029e-05,
+      "loss": 1.045,
+      "step": 10816
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3823367364005011e-05,
+      "loss": 1.082,
+      "step": 10817
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3822017880935506e-05,
+      "loss": 1.0794,
+      "step": 10818
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.382066836354493e-05,
+      "loss": 1.0127,
+      "step": 10819
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.381931881185871e-05,
+      "loss": 1.1193,
+      "step": 10820
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3817969225902265e-05,
+      "loss": 1.0296,
+      "step": 10821
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3816619605701012e-05,
+      "loss": 1.0414,
+      "step": 10822
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.381526995128038e-05,
+      "loss": 1.0902,
+      "step": 10823
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3813920262665787e-05,
+      "loss": 1.1057,
+      "step": 10824
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3812570539882663e-05,
+      "loss": 1.0017,
+      "step": 10825
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3811220782956427e-05,
+      "loss": 1.0681,
+      "step": 10826
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3809870991912508e-05,
+      "loss": 1.0226,
+      "step": 10827
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3808521166776328e-05,
+      "loss": 1.034,
+      "step": 10828
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3807171307573318e-05,
+      "loss": 1.0876,
+      "step": 10829
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3805821414328897e-05,
+      "loss": 1.0844,
+      "step": 10830
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3804471487068504e-05,
+      "loss": 1.0747,
+      "step": 10831
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.380312152581756e-05,
+      "loss": 1.0002,
+      "step": 10832
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3801771530601495e-05,
+      "loss": 1.0005,
+      "step": 10833
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3800421501445743e-05,
+      "loss": 0.9979,
+      "step": 10834
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3799071438375726e-05,
+      "loss": 1.0367,
+      "step": 10835
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3797721341416881e-05,
+      "loss": 1.0025,
+      "step": 10836
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3796371210594637e-05,
+      "loss": 0.9872,
+      "step": 10837
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3795021045934429e-05,
+      "loss": 1.0059,
+      "step": 10838
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3793670847461686e-05,
+      "loss": 0.998,
+      "step": 10839
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3792320615201845e-05,
+      "loss": 1.0237,
+      "step": 10840
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3790970349180339e-05,
+      "loss": 0.9621,
+      "step": 10841
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3789620049422603e-05,
+      "loss": 1.0251,
+      "step": 10842
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3788269715954072e-05,
+      "loss": 0.9536,
+      "step": 10843
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.378691934880018e-05,
+      "loss": 1.1463,
+      "step": 10844
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3785568947986367e-05,
+      "loss": 1.0049,
+      "step": 10845
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.378421851353807e-05,
+      "loss": 1.0187,
+      "step": 10846
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3782868045480723e-05,
+      "loss": 1.0516,
+      "step": 10847
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3781517543839769e-05,
+      "loss": 1.054,
+      "step": 10848
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3780167008640646e-05,
+      "loss": 1.0389,
+      "step": 10849
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3778816439908794e-05,
+      "loss": 1.0836,
+      "step": 10850
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.377746583766965e-05,
+      "loss": 1.0667,
+      "step": 10851
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3776115201948658e-05,
+      "loss": 1.0417,
+      "step": 10852
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3774764532771261e-05,
+      "loss": 1.0602,
+      "step": 10853
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3773413830162902e-05,
+      "loss": 0.9554,
+      "step": 10854
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.377206309414902e-05,
+      "loss": 1.0119,
+      "step": 10855
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3770712324755062e-05,
+      "loss": 1.0161,
+      "step": 10856
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3769361522006465e-05,
+      "loss": 0.9788,
+      "step": 10857
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3768010685928685e-05,
+      "loss": 1.0617,
+      "step": 10858
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.376665981654716e-05,
+      "loss": 1.0467,
+      "step": 10859
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.376530891388734e-05,
+      "loss": 1.124,
+      "step": 10860
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3763957977974669e-05,
+      "loss": 1.0663,
+      "step": 10861
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3762607008834596e-05,
+      "loss": 1.0164,
+      "step": 10862
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.376125600649257e-05,
+      "loss": 0.9864,
+      "step": 10863
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3759904970974033e-05,
+      "loss": 1.0071,
+      "step": 10864
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3758553902304443e-05,
+      "loss": 1.0185,
+      "step": 10865
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3757202800509244e-05,
+      "loss": 1.0403,
+      "step": 10866
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3755851665613893e-05,
+      "loss": 1.0177,
+      "step": 10867
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.375450049764383e-05,
+      "loss": 1.0344,
+      "step": 10868
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3753149296624519e-05,
+      "loss": 0.9822,
+      "step": 10869
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3751798062581402e-05,
+      "loss": 1.0206,
+      "step": 10870
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3750446795539941e-05,
+      "loss": 1.0667,
+      "step": 10871
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3749095495525584e-05,
+      "loss": 1.0708,
+      "step": 10872
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3747744162563787e-05,
+      "loss": 1.0227,
+      "step": 10873
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3746392796680004e-05,
+      "loss": 0.9973,
+      "step": 10874
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3745041397899692e-05,
+      "loss": 1.044,
+      "step": 10875
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3743689966248303e-05,
+      "loss": 1.0169,
+      "step": 10876
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3742338501751299e-05,
+      "loss": 0.9809,
+      "step": 10877
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3740987004434137e-05,
+      "loss": 1.1066,
+      "step": 10878
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3739635474322268e-05,
+      "loss": 1.041,
+      "step": 10879
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3738283911441158e-05,
+      "loss": 1.0437,
+      "step": 10880
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3736932315816265e-05,
+      "loss": 1.0521,
+      "step": 10881
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3735580687473046e-05,
+      "loss": 1.0596,
+      "step": 10882
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3734229026436965e-05,
+      "loss": 1.0481,
+      "step": 10883
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3732877332733479e-05,
+      "loss": 1.0,
+      "step": 10884
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3731525606388052e-05,
+      "loss": 1.0652,
+      "step": 10885
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3730173847426147e-05,
+      "loss": 1.0379,
+      "step": 10886
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3728822055873226e-05,
+      "loss": 0.9318,
+      "step": 10887
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3727470231754748e-05,
+      "loss": 1.071,
+      "step": 10888
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3726118375096187e-05,
+      "loss": 1.0993,
+      "step": 10889
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3724766485922998e-05,
+      "loss": 1.0688,
+      "step": 10890
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3723414564260654e-05,
+      "loss": 0.9798,
+      "step": 10891
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3722062610134615e-05,
+      "loss": 1.0406,
+      "step": 10892
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.372071062357035e-05,
+      "loss": 1.0194,
+      "step": 10893
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3719358604593325e-05,
+      "loss": 1.0646,
+      "step": 10894
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.371800655322901e-05,
+      "loss": 1.009,
+      "step": 10895
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3716654469502874e-05,
+      "loss": 1.0379,
+      "step": 10896
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.371530235344038e-05,
+      "loss": 1.0151,
+      "step": 10897
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3713950205067008e-05,
+      "loss": 1.0041,
+      "step": 10898
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3712598024408216e-05,
+      "loss": 1.1066,
+      "step": 10899
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3711245811489485e-05,
+      "loss": 1.0871,
+      "step": 10900
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.370989356633628e-05,
+      "loss": 1.0299,
+      "step": 10901
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3708541288974073e-05,
+      "loss": 1.0592,
+      "step": 10902
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3707188979428343e-05,
+      "loss": 1.0997,
+      "step": 10903
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3705836637724557e-05,
+      "loss": 1.0181,
+      "step": 10904
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3704484263888194e-05,
+      "loss": 1.024,
+      "step": 10905
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.370313185794472e-05,
+      "loss": 1.0063,
+      "step": 10906
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3701779419919623e-05,
+      "loss": 1.0706,
+      "step": 10907
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3700426949838364e-05,
+      "loss": 1.0404,
+      "step": 10908
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3699074447726432e-05,
+      "loss": 1.0722,
+      "step": 10909
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3697721913609295e-05,
+      "loss": 1.0248,
+      "step": 10910
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3696369347512434e-05,
+      "loss": 1.0806,
+      "step": 10911
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3695016749461327e-05,
+      "loss": 1.0159,
+      "step": 10912
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3693664119481455e-05,
+      "loss": 1.104,
+      "step": 10913
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.369231145759829e-05,
+      "loss": 1.0118,
+      "step": 10914
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3690958763837324e-05,
+      "loss": 1.0918,
+      "step": 10915
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3689606038224024e-05,
+      "loss": 0.9491,
+      "step": 10916
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3688253280783879e-05,
+      "loss": 1.0897,
+      "step": 10917
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.368690049154237e-05,
+      "loss": 1.0197,
+      "step": 10918
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3685547670524976e-05,
+      "loss": 1.0482,
+      "step": 10919
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3684194817757183e-05,
+      "loss": 0.9829,
+      "step": 10920
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3682841933264477e-05,
+      "loss": 1.029,
+      "step": 10921
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3681489017072333e-05,
+      "loss": 0.9945,
+      "step": 10922
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3680136069206247e-05,
+      "loss": 1.0575,
+      "step": 10923
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3678783089691694e-05,
+      "loss": 1.1069,
+      "step": 10924
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3677430078554165e-05,
+      "loss": 1.0331,
+      "step": 10925
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3676077035819148e-05,
+      "loss": 1.0511,
+      "step": 10926
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3674723961512128e-05,
+      "loss": 1.0443,
+      "step": 10927
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3673370855658592e-05,
+      "loss": 1.0931,
+      "step": 10928
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.367201771828403e-05,
+      "loss": 1.1541,
+      "step": 10929
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3670664549413928e-05,
+      "loss": 1.0818,
+      "step": 10930
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3669311349073777e-05,
+      "loss": 0.9865,
+      "step": 10931
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3667958117289068e-05,
+      "loss": 1.0932,
+      "step": 10932
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3666604854085293e-05,
+      "loss": 1.0561,
+      "step": 10933
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3665251559487937e-05,
+      "loss": 1.026,
+      "step": 10934
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3663898233522499e-05,
+      "loss": 1.1193,
+      "step": 10935
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3662544876214468e-05,
+      "loss": 0.9134,
+      "step": 10936
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3661191487589335e-05,
+      "loss": 1.0797,
+      "step": 10937
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.36598380676726e-05,
+      "loss": 1.0825,
+      "step": 10938
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3658484616489747e-05,
+      "loss": 1.1374,
+      "step": 10939
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3657131134066281e-05,
+      "loss": 1.0219,
+      "step": 10940
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3655777620427691e-05,
+      "loss": 1.0149,
+      "step": 10941
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3654424075599478e-05,
+      "loss": 1.0864,
+      "step": 10942
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.365307049960713e-05,
+      "loss": 1.1246,
+      "step": 10943
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3651716892476154e-05,
+      "loss": 1.0427,
+      "step": 10944
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3650363254232041e-05,
+      "loss": 1.0655,
+      "step": 10945
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3649009584900293e-05,
+      "loss": 1.0758,
+      "step": 10946
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3647655884506408e-05,
+      "loss": 0.9564,
+      "step": 10947
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3646302153075881e-05,
+      "loss": 1.0264,
+      "step": 10948
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3644948390634221e-05,
+      "loss": 1.0809,
+      "step": 10949
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.364359459720692e-05,
+      "loss": 1.0783,
+      "step": 10950
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3642240772819484e-05,
+      "loss": 1.0319,
+      "step": 10951
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3640886917497412e-05,
+      "loss": 1.0767,
+      "step": 10952
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.363953303126621e-05,
+      "loss": 1.0639,
+      "step": 10953
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3638179114151378e-05,
+      "loss": 1.0263,
+      "step": 10954
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.363682516617842e-05,
+      "loss": 1.0308,
+      "step": 10955
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.363547118737284e-05,
+      "loss": 1.0774,
+      "step": 10956
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3634117177760144e-05,
+      "loss": 0.9718,
+      "step": 10957
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3632763137365837e-05,
+      "loss": 1.0901,
+      "step": 10958
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3631409066215421e-05,
+      "loss": 1.0336,
+      "step": 10959
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.363005496433441e-05,
+      "loss": 1.0649,
+      "step": 10960
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3628700831748303e-05,
+      "loss": 1.026,
+      "step": 10961
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3627346668482613e-05,
+      "loss": 1.0041,
+      "step": 10962
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3625992474562846e-05,
+      "loss": 1.0715,
+      "step": 10963
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.362463825001451e-05,
+      "loss": 0.9947,
+      "step": 10964
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3623283994863116e-05,
+      "loss": 1.1313,
+      "step": 10965
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3621929709134176e-05,
+      "loss": 0.9818,
+      "step": 10966
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3620575392853194e-05,
+      "loss": 1.0313,
+      "step": 10967
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3619221046045688e-05,
+      "loss": 1.0609,
+      "step": 10968
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3617866668737165e-05,
+      "loss": 0.9887,
+      "step": 10969
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3616512260953138e-05,
+      "loss": 1.0737,
+      "step": 10970
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3615157822719121e-05,
+      "loss": 1.0229,
+      "step": 10971
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3613803354060626e-05,
+      "loss": 1.0609,
+      "step": 10972
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3612448855003168e-05,
+      "loss": 1.0843,
+      "step": 10973
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3611094325572263e-05,
+      "loss": 1.054,
+      "step": 10974
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3609739765793424e-05,
+      "loss": 1.0859,
+      "step": 10975
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3608385175692164e-05,
+      "loss": 1.0086,
+      "step": 10976
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3607030555294005e-05,
+      "loss": 1.0719,
+      "step": 10977
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.360567590462446e-05,
+      "loss": 1.1079,
+      "step": 10978
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3604321223709047e-05,
+      "loss": 1.087,
+      "step": 10979
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3602966512573286e-05,
+      "loss": 0.9731,
+      "step": 10980
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.360161177124269e-05,
+      "loss": 1.0024,
+      "step": 10981
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3600256999742785e-05,
+      "loss": 1.0892,
+      "step": 10982
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3598902198099083e-05,
+      "loss": 1.0629,
+      "step": 10983
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3597547366337113e-05,
+      "loss": 1.0884,
+      "step": 10984
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3596192504482388e-05,
+      "loss": 0.9934,
+      "step": 10985
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3594837612560436e-05,
+      "loss": 1.1059,
+      "step": 10986
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.359348269059677e-05,
+      "loss": 1.0293,
+      "step": 10987
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3592127738616925e-05,
+      "loss": 1.0148,
+      "step": 10988
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3590772756646411e-05,
+      "loss": 1.0555,
+      "step": 10989
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3589417744710758e-05,
+      "loss": 0.9986,
+      "step": 10990
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3588062702835494e-05,
+      "loss": 1.0426,
+      "step": 10991
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3586707631046135e-05,
+      "loss": 1.028,
+      "step": 10992
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3585352529368213e-05,
+      "loss": 1.0457,
+      "step": 10993
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.358399739782725e-05,
+      "loss": 1.0626,
+      "step": 10994
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3582642236448776e-05,
+      "loss": 0.9676,
+      "step": 10995
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3581287045258311e-05,
+      "loss": 0.9848,
+      "step": 10996
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3579931824281393e-05,
+      "loss": 1.0902,
+      "step": 10997
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.357857657354354e-05,
+      "loss": 1.0059,
+      "step": 10998
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3577221293070288e-05,
+      "loss": 1.0858,
+      "step": 10999
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3575865982887163e-05,
+      "loss": 0.9978,
+      "step": 11000
+    },
+    {
+      "epoch": 0.47,
+      "eval_text_loss": 0.7868174314498901,
+      "eval_text_runtime": 1.3952,
+      "eval_text_samples_per_second": 53.038,
+      "eval_text_steps_per_second": 1.433,
+      "step": 11000
+    },
+    {
+      "epoch": 0.47,
+      "eval_svit_loss": 1.1269441843032837,
+      "eval_svit_runtime": 4.9594,
+      "eval_svit_samples_per_second": 137.315,
+      "eval_svit_steps_per_second": 2.218,
+      "step": 11000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3574510643019696e-05,
+      "loss": 1.1268,
+      "step": 11001
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3573155273493414e-05,
+      "loss": 1.025,
+      "step": 11002
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.357179987433385e-05,
+      "loss": 1.0422,
+      "step": 11003
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3570444445566537e-05,
+      "loss": 1.002,
+      "step": 11004
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3569088987217008e-05,
+      "loss": 1.0285,
+      "step": 11005
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.356773349931079e-05,
+      "loss": 1.1359,
+      "step": 11006
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.356637798187342e-05,
+      "loss": 0.9938,
+      "step": 11007
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3565022434930432e-05,
+      "loss": 1.0829,
+      "step": 11008
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3563666858507358e-05,
+      "loss": 1.0123,
+      "step": 11009
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3562311252629735e-05,
+      "loss": 1.1035,
+      "step": 11010
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3560955617323097e-05,
+      "loss": 1.0488,
+      "step": 11011
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.355959995261298e-05,
+      "loss": 1.0429,
+      "step": 11012
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3558244258524922e-05,
+      "loss": 1.0426,
+      "step": 11013
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3556888535084458e-05,
+      "loss": 1.0354,
+      "step": 11014
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3555532782317126e-05,
+      "loss": 1.1038,
+      "step": 11015
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3554177000248465e-05,
+      "loss": 1.0507,
+      "step": 11016
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3552821188904013e-05,
+      "loss": 1.0614,
+      "step": 11017
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.355146534830931e-05,
+      "loss": 1.1308,
+      "step": 11018
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3550109478489895e-05,
+      "loss": 1.0288,
+      "step": 11019
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3548753579471305e-05,
+      "loss": 1.0767,
+      "step": 11020
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3547397651279088e-05,
+      "loss": 1.0399,
+      "step": 11021
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3546041693938783e-05,
+      "loss": 1.0271,
+      "step": 11022
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3544685707475924e-05,
+      "loss": 1.1251,
+      "step": 11023
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3543329691916067e-05,
+      "loss": 1.0884,
+      "step": 11024
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3541973647284743e-05,
+      "loss": 0.9986,
+      "step": 11025
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3540617573607505e-05,
+      "loss": 1.0375,
+      "step": 11026
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3539261470909887e-05,
+      "loss": 1.0702,
+      "step": 11027
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3537905339217447e-05,
+      "loss": 1.1133,
+      "step": 11028
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3536549178555718e-05,
+      "loss": 1.101,
+      "step": 11029
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3535192988950253e-05,
+      "loss": 1.0523,
+      "step": 11030
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3533836770426594e-05,
+      "loss": 1.0743,
+      "step": 11031
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3532480523010292e-05,
+      "loss": 1.0551,
+      "step": 11032
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.353112424672689e-05,
+      "loss": 1.0641,
+      "step": 11033
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3529767941601936e-05,
+      "loss": 1.0125,
+      "step": 11034
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3528411607660987e-05,
+      "loss": 1.0181,
+      "step": 11035
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3527055244929578e-05,
+      "loss": 0.9787,
+      "step": 11036
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3525698853433275e-05,
+      "loss": 0.9721,
+      "step": 11037
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3524342433197613e-05,
+      "loss": 1.0791,
+      "step": 11038
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3522985984248153e-05,
+      "loss": 1.0266,
+      "step": 11039
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3521629506610438e-05,
+      "loss": 1.0092,
+      "step": 11040
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.352027300031003e-05,
+      "loss": 0.9931,
+      "step": 11041
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3518916465372474e-05,
+      "loss": 1.1247,
+      "step": 11042
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3517559901823323e-05,
+      "loss": 1.0601,
+      "step": 11043
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.351620330968813e-05,
+      "loss": 1.0577,
+      "step": 11044
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3514846688992451e-05,
+      "loss": 1.0778,
+      "step": 11045
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3513490039761842e-05,
+      "loss": 0.9992,
+      "step": 11046
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3512133362021855e-05,
+      "loss": 0.9816,
+      "step": 11047
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3510776655798045e-05,
+      "loss": 0.9993,
+      "step": 11048
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3509419921115974e-05,
+      "loss": 1.0406,
+      "step": 11049
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.350806315800119e-05,
+      "loss": 1.0382,
+      "step": 11050
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3506706366479256e-05,
+      "loss": 1.0685,
+      "step": 11051
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3505349546575727e-05,
+      "loss": 0.9594,
+      "step": 11052
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3503992698316164e-05,
+      "loss": 1.026,
+      "step": 11053
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3502635821726124e-05,
+      "loss": 1.1301,
+      "step": 11054
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3501278916831165e-05,
+      "loss": 1.0482,
+      "step": 11055
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3499921983656848e-05,
+      "loss": 1.0675,
+      "step": 11056
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3498565022228734e-05,
+      "loss": 1.0624,
+      "step": 11057
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3497208032572385e-05,
+      "loss": 0.9766,
+      "step": 11058
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3495851014713357e-05,
+      "loss": 1.0701,
+      "step": 11059
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3494493968677216e-05,
+      "loss": 1.0817,
+      "step": 11060
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3493136894489526e-05,
+      "loss": 1.0674,
+      "step": 11061
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3491779792175847e-05,
+      "loss": 1.0394,
+      "step": 11062
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3490422661761743e-05,
+      "loss": 1.0129,
+      "step": 11063
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3489065503272784e-05,
+      "loss": 1.0554,
+      "step": 11064
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3487708316734523e-05,
+      "loss": 1.0477,
+      "step": 11065
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3486351102172534e-05,
+      "loss": 1.0782,
+      "step": 11066
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3484993859612379e-05,
+      "loss": 1.1045,
+      "step": 11067
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3483636589079627e-05,
+      "loss": 1.1016,
+      "step": 11068
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3482279290599837e-05,
+      "loss": 1.0041,
+      "step": 11069
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3480921964198589e-05,
+      "loss": 1.1293,
+      "step": 11070
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.347956460990144e-05,
+      "loss": 1.0768,
+      "step": 11071
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3478207227733966e-05,
+      "loss": 1.0932,
+      "step": 11072
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3476849817721728e-05,
+      "loss": 0.9056,
+      "step": 11073
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.34754923798903e-05,
+      "loss": 1.0168,
+      "step": 11074
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3474134914265252e-05,
+      "loss": 0.9783,
+      "step": 11075
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3472777420872153e-05,
+      "loss": 1.051,
+      "step": 11076
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3471419899736572e-05,
+      "loss": 1.0371,
+      "step": 11077
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3470062350884086e-05,
+      "loss": 1.0264,
+      "step": 11078
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3468704774340264e-05,
+      "loss": 1.0687,
+      "step": 11079
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3467347170130673e-05,
+      "loss": 1.0386,
+      "step": 11080
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3465989538280896e-05,
+      "loss": 1.0937,
+      "step": 11081
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3464631878816501e-05,
+      "loss": 0.9777,
+      "step": 11082
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3463274191763063e-05,
+      "loss": 0.9904,
+      "step": 11083
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3461916477146156e-05,
+      "loss": 1.0128,
+      "step": 11084
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3460558734991354e-05,
+      "loss": 1.0346,
+      "step": 11085
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3459200965324232e-05,
+      "loss": 1.0484,
+      "step": 11086
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.345784316817037e-05,
+      "loss": 1.1057,
+      "step": 11087
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3456485343555343e-05,
+      "loss": 1.1107,
+      "step": 11088
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3455127491504728e-05,
+      "loss": 1.0057,
+      "step": 11089
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.34537696120441e-05,
+      "loss": 1.0568,
+      "step": 11090
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3452411705199042e-05,
+      "loss": 1.1071,
+      "step": 11091
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.345105377099513e-05,
+      "loss": 1.086,
+      "step": 11092
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3449695809457941e-05,
+      "loss": 1.012,
+      "step": 11093
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.344833782061306e-05,
+      "loss": 0.9789,
+      "step": 11094
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3446979804486062e-05,
+      "loss": 1.0693,
+      "step": 11095
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3445621761102532e-05,
+      "loss": 1.0623,
+      "step": 11096
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3444263690488046e-05,
+      "loss": 0.9919,
+      "step": 11097
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3442905592668191e-05,
+      "loss": 0.9906,
+      "step": 11098
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3441547467668547e-05,
+      "loss": 1.0859,
+      "step": 11099
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3440189315514698e-05,
+      "loss": 1.0381,
+      "step": 11100
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3438831136232226e-05,
+      "loss": 0.9736,
+      "step": 11101
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3437472929846715e-05,
+      "loss": 1.0507,
+      "step": 11102
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3436114696383748e-05,
+      "loss": 1.1089,
+      "step": 11103
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3434756435868916e-05,
+      "loss": 1.1366,
+      "step": 11104
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3433398148327794e-05,
+      "loss": 1.0375,
+      "step": 11105
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3432039833785978e-05,
+      "loss": 1.037,
+      "step": 11106
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3430681492269047e-05,
+      "loss": 1.0356,
+      "step": 11107
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3429323123802593e-05,
+      "loss": 1.0265,
+      "step": 11108
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.34279647284122e-05,
+      "loss": 1.0127,
+      "step": 11109
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3426606306123458e-05,
+      "loss": 1.0767,
+      "step": 11110
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.342524785696195e-05,
+      "loss": 1.1479,
+      "step": 11111
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3423889380953277e-05,
+      "loss": 1.0116,
+      "step": 11112
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3422530878123015e-05,
+      "loss": 1.0487,
+      "step": 11113
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3421172348496762e-05,
+      "loss": 1.0428,
+      "step": 11114
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3419813792100105e-05,
+      "loss": 1.0319,
+      "step": 11115
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3418455208958636e-05,
+      "loss": 1.0742,
+      "step": 11116
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3417096599097947e-05,
+      "loss": 1.0487,
+      "step": 11117
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3415737962543629e-05,
+      "loss": 1.055,
+      "step": 11118
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3414379299321273e-05,
+      "loss": 1.0747,
+      "step": 11119
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3413020609456477e-05,
+      "loss": 1.0633,
+      "step": 11120
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3411661892974829e-05,
+      "loss": 1.0798,
+      "step": 11121
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3410303149901924e-05,
+      "loss": 1.0468,
+      "step": 11122
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.340894438026336e-05,
+      "loss": 1.0211,
+      "step": 11123
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3407585584084726e-05,
+      "loss": 1.0418,
+      "step": 11124
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3406226761391624e-05,
+      "loss": 1.0423,
+      "step": 11125
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3404867912209646e-05,
+      "loss": 1.0711,
+      "step": 11126
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.340350903656439e-05,
+      "loss": 1.103,
+      "step": 11127
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3402150134481453e-05,
+      "loss": 1.0497,
+      "step": 11128
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3400791205986428e-05,
+      "loss": 1.078,
+      "step": 11129
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3399432251104918e-05,
+      "loss": 1.0894,
+      "step": 11130
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3398073269862521e-05,
+      "loss": 1.023,
+      "step": 11131
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3396714262284833e-05,
+      "loss": 0.9889,
+      "step": 11132
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3395355228397458e-05,
+      "loss": 1.0245,
+      "step": 11133
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3393996168225988e-05,
+      "loss": 0.9971,
+      "step": 11134
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3392637081796034e-05,
+      "loss": 1.0732,
+      "step": 11135
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.339127796913319e-05,
+      "loss": 1.0244,
+      "step": 11136
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3389918830263058e-05,
+      "loss": 1.066,
+      "step": 11137
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.338855966521124e-05,
+      "loss": 1.0908,
+      "step": 11138
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3387200474003338e-05,
+      "loss": 1.0288,
+      "step": 11139
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.338584125666496e-05,
+      "loss": 1.0864,
+      "step": 11140
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3384482013221703e-05,
+      "loss": 1.1071,
+      "step": 11141
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3383122743699172e-05,
+      "loss": 1.0087,
+      "step": 11142
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3381763448122976e-05,
+      "loss": 1.0764,
+      "step": 11143
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3380404126518713e-05,
+      "loss": 1.0114,
+      "step": 11144
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3379044778911993e-05,
+      "loss": 1.0183,
+      "step": 11145
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3377685405328423e-05,
+      "loss": 1.0812,
+      "step": 11146
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.33763260057936e-05,
+      "loss": 1.093,
+      "step": 11147
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3374966580333147e-05,
+      "loss": 0.9887,
+      "step": 11148
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3373607128972654e-05,
+      "loss": 1.0266,
+      "step": 11149
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3372247651737742e-05,
+      "loss": 1.0401,
+      "step": 11150
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.337088814865401e-05,
+      "loss": 0.9941,
+      "step": 11151
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3369528619747073e-05,
+      "loss": 1.0584,
+      "step": 11152
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3368169065042536e-05,
+      "loss": 1.045,
+      "step": 11153
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3366809484566014e-05,
+      "loss": 0.9096,
+      "step": 11154
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.336544987834311e-05,
+      "loss": 1.0712,
+      "step": 11155
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.336409024639944e-05,
+      "loss": 1.0398,
+      "step": 11156
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3362730588760617e-05,
+      "loss": 0.9346,
+      "step": 11157
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3361370905452246e-05,
+      "loss": 1.0485,
+      "step": 11158
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3360011196499942e-05,
+      "loss": 0.9753,
+      "step": 11159
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3358651461929318e-05,
+      "loss": 1.0263,
+      "step": 11160
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.335729170176599e-05,
+      "loss": 0.9616,
+      "step": 11161
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.335593191603557e-05,
+      "loss": 1.0261,
+      "step": 11162
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3354572104763668e-05,
+      "loss": 0.9451,
+      "step": 11163
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.33532122679759e-05,
+      "loss": 1.0234,
+      "step": 11164
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3351852405697884e-05,
+      "loss": 1.0145,
+      "step": 11165
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3350492517955233e-05,
+      "loss": 1.0658,
+      "step": 11166
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3349132604773569e-05,
+      "loss": 0.9865,
+      "step": 11167
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.33477726661785e-05,
+      "loss": 1.0197,
+      "step": 11168
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3346412702195645e-05,
+      "loss": 1.0807,
+      "step": 11169
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3345052712850625e-05,
+      "loss": 1.0433,
+      "step": 11170
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3343692698169057e-05,
+      "loss": 1.0764,
+      "step": 11171
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3342332658176555e-05,
+      "loss": 1.0378,
+      "step": 11172
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3340972592898745e-05,
+      "loss": 0.9949,
+      "step": 11173
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.333961250236124e-05,
+      "loss": 1.0016,
+      "step": 11174
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.333825238658966e-05,
+      "loss": 1.0868,
+      "step": 11175
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.333689224560963e-05,
+      "loss": 1.0776,
+      "step": 11176
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3335532079446768e-05,
+      "loss": 1.1351,
+      "step": 11177
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3334171888126697e-05,
+      "loss": 1.061,
+      "step": 11178
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3332811671675037e-05,
+      "loss": 1.0376,
+      "step": 11179
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3331451430117405e-05,
+      "loss": 1.0013,
+      "step": 11180
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3330091163479434e-05,
+      "loss": 1.0631,
+      "step": 11181
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3328730871786742e-05,
+      "loss": 1.0048,
+      "step": 11182
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3327370555064951e-05,
+      "loss": 1.0594,
+      "step": 11183
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3326010213339688e-05,
+      "loss": 1.0578,
+      "step": 11184
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3324649846636577e-05,
+      "loss": 1.0797,
+      "step": 11185
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3323289454981242e-05,
+      "loss": 1.0532,
+      "step": 11186
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3321929038399308e-05,
+      "loss": 0.9769,
+      "step": 11187
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3320568596916405e-05,
+      "loss": 1.0052,
+      "step": 11188
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3319208130558151e-05,
+      "loss": 1.0427,
+      "step": 11189
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3317847639350186e-05,
+      "loss": 1.059,
+      "step": 11190
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3316487123318122e-05,
+      "loss": 1.0537,
+      "step": 11191
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3315126582487599e-05,
+      "loss": 1.0447,
+      "step": 11192
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3313766016884238e-05,
+      "loss": 1.1282,
+      "step": 11193
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3312405426533671e-05,
+      "loss": 1.0598,
+      "step": 11194
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3311044811461526e-05,
+      "loss": 1.0296,
+      "step": 11195
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3309684171693435e-05,
+      "loss": 1.1299,
+      "step": 11196
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3308323507255025e-05,
+      "loss": 0.997,
+      "step": 11197
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3306962818171932e-05,
+      "loss": 0.9838,
+      "step": 11198
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3305602104469778e-05,
+      "loss": 1.0163,
+      "step": 11199
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3304241366174204e-05,
+      "loss": 1.0334,
+      "step": 11200
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3302880603310834e-05,
+      "loss": 1.0635,
+      "step": 11201
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3301519815905308e-05,
+      "loss": 1.011,
+      "step": 11202
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3300159003983253e-05,
+      "loss": 1.0326,
+      "step": 11203
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3298798167570305e-05,
+      "loss": 1.045,
+      "step": 11204
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3297437306692097e-05,
+      "loss": 1.0548,
+      "step": 11205
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3296076421374264e-05,
+      "loss": 1.0473,
+      "step": 11206
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3294715511642441e-05,
+      "loss": 1.0398,
+      "step": 11207
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3293354577522264e-05,
+      "loss": 1.0204,
+      "step": 11208
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3291993619039364e-05,
+      "loss": 1.1444,
+      "step": 11209
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3290632636219385e-05,
+      "loss": 1.0948,
+      "step": 11210
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3289271629087959e-05,
+      "loss": 1.0784,
+      "step": 11211
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3287910597670721e-05,
+      "loss": 1.0359,
+      "step": 11212
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3286549541993311e-05,
+      "loss": 1.1042,
+      "step": 11213
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3285188462081369e-05,
+      "loss": 1.0351,
+      "step": 11214
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.328382735796053e-05,
+      "loss": 1.0246,
+      "step": 11215
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3282466229656435e-05,
+      "loss": 0.956,
+      "step": 11216
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3281105077194723e-05,
+      "loss": 1.0876,
+      "step": 11217
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3279743900601032e-05,
+      "loss": 1.033,
+      "step": 11218
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3278382699901002e-05,
+      "loss": 1.0739,
+      "step": 11219
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.327702147512028e-05,
+      "loss": 1.0929,
+      "step": 11220
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.32756602262845e-05,
+      "loss": 1.0398,
+      "step": 11221
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3274298953419307e-05,
+      "loss": 1.0967,
+      "step": 11222
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3272937656550344e-05,
+      "loss": 1.1031,
+      "step": 11223
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.327157633570325e-05,
+      "loss": 1.0839,
+      "step": 11224
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3270214990903669e-05,
+      "loss": 1.0401,
+      "step": 11225
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3268853622177247e-05,
+      "loss": 1.0391,
+      "step": 11226
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3267492229549625e-05,
+      "loss": 1.0764,
+      "step": 11227
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3266130813046449e-05,
+      "loss": 0.9978,
+      "step": 11228
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3264769372693364e-05,
+      "loss": 0.9854,
+      "step": 11229
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3263407908516018e-05,
+      "loss": 1.1401,
+      "step": 11230
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3262046420540046e-05,
+      "loss": 1.0335,
+      "step": 11231
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3260684908791108e-05,
+      "loss": 1.0872,
+      "step": 11232
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3259323373294841e-05,
+      "loss": 1.0259,
+      "step": 11233
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3257961814076897e-05,
+      "loss": 1.0562,
+      "step": 11234
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3256600231162917e-05,
+      "loss": 1.036,
+      "step": 11235
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3255238624578558e-05,
+      "loss": 0.9768,
+      "step": 11236
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3253876994349461e-05,
+      "loss": 1.0964,
+      "step": 11237
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3252515340501282e-05,
+      "loss": 1.0556,
+      "step": 11238
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3251153663059659e-05,
+      "loss": 1.0157,
+      "step": 11239
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3249791962050255e-05,
+      "loss": 0.9731,
+      "step": 11240
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3248430237498713e-05,
+      "loss": 1.0655,
+      "step": 11241
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3247068489430682e-05,
+      "loss": 0.9711,
+      "step": 11242
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3245706717871816e-05,
+      "loss": 1.0708,
+      "step": 11243
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3244344922847764e-05,
+      "loss": 1.0246,
+      "step": 11244
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3242983104384183e-05,
+      "loss": 1.0131,
+      "step": 11245
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3241621262506721e-05,
+      "loss": 1.0034,
+      "step": 11246
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3240259397241033e-05,
+      "loss": 1.0069,
+      "step": 11247
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.323889750861277e-05,
+      "loss": 1.0536,
+      "step": 11248
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3237535596647588e-05,
+      "loss": 1.0437,
+      "step": 11249
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3236173661371138e-05,
+      "loss": 1.0141,
+      "step": 11250
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3234811702809078e-05,
+      "loss": 1.0332,
+      "step": 11251
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3233449720987062e-05,
+      "loss": 1.076,
+      "step": 11252
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3232087715930744e-05,
+      "loss": 0.9305,
+      "step": 11253
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3230725687665782e-05,
+      "loss": 0.9889,
+      "step": 11254
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.322936363621783e-05,
+      "loss": 1.0823,
+      "step": 11255
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3228001561612547e-05,
+      "loss": 1.0241,
+      "step": 11256
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3226639463875588e-05,
+      "loss": 1.0668,
+      "step": 11257
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3225277343032613e-05,
+      "loss": 1.1336,
+      "step": 11258
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3223915199109279e-05,
+      "loss": 1.063,
+      "step": 11259
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3222553032131244e-05,
+      "loss": 1.1106,
+      "step": 11260
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3221190842124167e-05,
+      "loss": 1.0528,
+      "step": 11261
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3219828629113706e-05,
+      "loss": 0.9698,
+      "step": 11262
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3218466393125529e-05,
+      "loss": 1.0423,
+      "step": 11263
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3217104134185281e-05,
+      "loss": 1.062,
+      "step": 11264
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3215741852318638e-05,
+      "loss": 1.1078,
+      "step": 11265
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.321437954755125e-05,
+      "loss": 1.0747,
+      "step": 11266
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3213017219908786e-05,
+      "loss": 1.1093,
+      "step": 11267
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3211654869416901e-05,
+      "loss": 1.0064,
+      "step": 11268
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3210292496101264e-05,
+      "loss": 1.0306,
+      "step": 11269
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3208930099987535e-05,
+      "loss": 1.0223,
+      "step": 11270
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3207567681101377e-05,
+      "loss": 1.0391,
+      "step": 11271
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3206205239468457e-05,
+      "loss": 1.0507,
+      "step": 11272
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.320484277511443e-05,
+      "loss": 1.0791,
+      "step": 11273
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3203480288064974e-05,
+      "loss": 1.084,
+      "step": 11274
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.320211777834574e-05,
+      "loss": 0.984,
+      "step": 11275
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3200755245982404e-05,
+      "loss": 1.0424,
+      "step": 11276
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3199392691000624e-05,
+      "loss": 1.0042,
+      "step": 11277
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3198030113426074e-05,
+      "loss": 1.0627,
+      "step": 11278
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3196667513284414e-05,
+      "loss": 0.9991,
+      "step": 11279
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3195304890601317e-05,
+      "loss": 1.0712,
+      "step": 11280
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3193942245402442e-05,
+      "loss": 1.0267,
+      "step": 11281
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3192579577713469e-05,
+      "loss": 1.0659,
+      "step": 11282
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3191216887560056e-05,
+      "loss": 0.9632,
+      "step": 11283
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3189854174967875e-05,
+      "loss": 1.0544,
+      "step": 11284
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3188491439962595e-05,
+      "loss": 1.0493,
+      "step": 11285
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3187128682569887e-05,
+      "loss": 0.9973,
+      "step": 11286
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3185765902815421e-05,
+      "loss": 1.009,
+      "step": 11287
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3184403100724867e-05,
+      "loss": 1.0963,
+      "step": 11288
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3183040276323894e-05,
+      "loss": 1.1132,
+      "step": 11289
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3181677429638175e-05,
+      "loss": 1.0577,
+      "step": 11290
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3180314560693383e-05,
+      "loss": 1.0143,
+      "step": 11291
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.317895166951519e-05,
+      "loss": 1.0784,
+      "step": 11292
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3177588756129264e-05,
+      "loss": 1.0467,
+      "step": 11293
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3176225820561283e-05,
+      "loss": 0.9774,
+      "step": 11294
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.317486286283692e-05,
+      "loss": 1.105,
+      "step": 11295
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3173499882981846e-05,
+      "loss": 1.0479,
+      "step": 11296
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3172136881021737e-05,
+      "loss": 1.0695,
+      "step": 11297
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3170773856982268e-05,
+      "loss": 0.9995,
+      "step": 11298
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3169410810889113e-05,
+      "loss": 1.0152,
+      "step": 11299
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3168047742767949e-05,
+      "loss": 0.9718,
+      "step": 11300
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.316668465264445e-05,
+      "loss": 1.0214,
+      "step": 11301
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3165321540544294e-05,
+      "loss": 1.0482,
+      "step": 11302
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3163958406493156e-05,
+      "loss": 1.0478,
+      "step": 11303
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3162595250516714e-05,
+      "loss": 1.0266,
+      "step": 11304
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3161232072640649e-05,
+      "loss": 1.0594,
+      "step": 11305
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.315986887289063e-05,
+      "loss": 1.0418,
+      "step": 11306
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3158505651292347e-05,
+      "loss": 1.0235,
+      "step": 11307
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3157142407871468e-05,
+      "loss": 1.0142,
+      "step": 11308
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3155779142653682e-05,
+      "loss": 1.0967,
+      "step": 11309
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3154415855664657e-05,
+      "loss": 1.132,
+      "step": 11310
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3153052546930087e-05,
+      "loss": 1.0132,
+      "step": 11311
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.315168921647564e-05,
+      "loss": 1.1344,
+      "step": 11312
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3150325864327002e-05,
+      "loss": 1.0837,
+      "step": 11313
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.314896249050986e-05,
+      "loss": 1.0042,
+      "step": 11314
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3147599095049883e-05,
+      "loss": 1.0161,
+      "step": 11315
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3146235677972764e-05,
+      "loss": 1.1142,
+      "step": 11316
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3144872239304179e-05,
+      "loss": 1.046,
+      "step": 11317
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3143508779069816e-05,
+      "loss": 0.995,
+      "step": 11318
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3142145297295351e-05,
+      "loss": 1.0488,
+      "step": 11319
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3140781794006478e-05,
+      "loss": 1.0882,
+      "step": 11320
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3139418269228871e-05,
+      "loss": 1.0203,
+      "step": 11321
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3138054722988223e-05,
+      "loss": 1.0765,
+      "step": 11322
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.313669115531021e-05,
+      "loss": 1.1565,
+      "step": 11323
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3135327566220527e-05,
+      "loss": 0.9739,
+      "step": 11324
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3133963955744853e-05,
+      "loss": 1.0793,
+      "step": 11325
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3132600323908876e-05,
+      "loss": 1.0446,
+      "step": 11326
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3131236670738282e-05,
+      "loss": 1.0006,
+      "step": 11327
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3129872996258756e-05,
+      "loss": 0.9897,
+      "step": 11328
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3128509300495991e-05,
+      "loss": 1.1263,
+      "step": 11329
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.312714558347567e-05,
+      "loss": 1.0841,
+      "step": 11330
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3125781845223484e-05,
+      "loss": 1.0708,
+      "step": 11331
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.312441808576512e-05,
+      "loss": 0.949,
+      "step": 11332
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3123054305126266e-05,
+      "loss": 1.0679,
+      "step": 11333
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3121690503332613e-05,
+      "loss": 1.0373,
+      "step": 11334
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3120326680409849e-05,
+      "loss": 1.0295,
+      "step": 11335
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3118962836383666e-05,
+      "loss": 1.0774,
+      "step": 11336
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3117598971279754e-05,
+      "loss": 1.0225,
+      "step": 11337
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3116235085123803e-05,
+      "loss": 1.1009,
+      "step": 11338
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3114871177941506e-05,
+      "loss": 1.0785,
+      "step": 11339
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3113507249758552e-05,
+      "loss": 1.0473,
+      "step": 11340
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3112143300600636e-05,
+      "loss": 1.0531,
+      "step": 11341
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3110779330493448e-05,
+      "loss": 1.0842,
+      "step": 11342
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3109415339462685e-05,
+      "loss": 1.0853,
+      "step": 11343
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3108051327534034e-05,
+      "loss": 1.0405,
+      "step": 11344
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3106687294733195e-05,
+      "loss": 1.0121,
+      "step": 11345
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3105323241085856e-05,
+      "loss": 1.0194,
+      "step": 11346
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3103959166617721e-05,
+      "loss": 1.0636,
+      "step": 11347
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.310259507135447e-05,
+      "loss": 1.1039,
+      "step": 11348
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3101230955321815e-05,
+      "loss": 1.0846,
+      "step": 11349
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3099866818545437e-05,
+      "loss": 1.105,
+      "step": 11350
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3098502661051042e-05,
+      "loss": 1.0291,
+      "step": 11351
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.309713848286432e-05,
+      "loss": 1.043,
+      "step": 11352
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3095774284010974e-05,
+      "loss": 1.0922,
+      "step": 11353
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3094410064516694e-05,
+      "loss": 1.138,
+      "step": 11354
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3093045824407186e-05,
+      "loss": 1.0739,
+      "step": 11355
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3091681563708141e-05,
+      "loss": 1.0797,
+      "step": 11356
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3090317282445258e-05,
+      "loss": 1.0589,
+      "step": 11357
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3088952980644241e-05,
+      "loss": 1.0487,
+      "step": 11358
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.308758865833078e-05,
+      "loss": 1.0433,
+      "step": 11359
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3086224315530589e-05,
+      "loss": 1.1247,
+      "step": 11360
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3084859952269353e-05,
+      "loss": 1.0383,
+      "step": 11361
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.308349556857278e-05,
+      "loss": 1.1589,
+      "step": 11362
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3082131164466569e-05,
+      "loss": 1.0715,
+      "step": 11363
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3080766739976423e-05,
+      "loss": 1.1174,
+      "step": 11364
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3079402295128038e-05,
+      "loss": 0.9994,
+      "step": 11365
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3078037829947122e-05,
+      "loss": 1.0725,
+      "step": 11366
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3076673344459376e-05,
+      "loss": 1.0633,
+      "step": 11367
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.30753088386905e-05,
+      "loss": 1.1002,
+      "step": 11368
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3073944312666197e-05,
+      "loss": 1.0702,
+      "step": 11369
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3072579766412172e-05,
+      "loss": 1.1463,
+      "step": 11370
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.307121519995413e-05,
+      "loss": 1.0768,
+      "step": 11371
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3069850613317772e-05,
+      "loss": 1.0704,
+      "step": 11372
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3068486006528805e-05,
+      "loss": 1.1301,
+      "step": 11373
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3067121379612933e-05,
+      "loss": 1.142,
+      "step": 11374
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.306575673259586e-05,
+      "loss": 0.9914,
+      "step": 11375
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3064392065503294e-05,
+      "loss": 1.0571,
+      "step": 11376
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.306302737836094e-05,
+      "loss": 1.0301,
+      "step": 11377
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3061662671194503e-05,
+      "loss": 1.0927,
+      "step": 11378
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3060297944029691e-05,
+      "loss": 1.1546,
+      "step": 11379
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3058933196892214e-05,
+      "loss": 1.1679,
+      "step": 11380
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3057568429807775e-05,
+      "loss": 1.0797,
+      "step": 11381
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.305620364280208e-05,
+      "loss": 1.1125,
+      "step": 11382
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3054838835900847e-05,
+      "loss": 1.1385,
+      "step": 11383
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3053474009129774e-05,
+      "loss": 1.0964,
+      "step": 11384
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3052109162514577e-05,
+      "loss": 1.0774,
+      "step": 11385
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3050744296080962e-05,
+      "loss": 1.0956,
+      "step": 11386
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.304937940985464e-05,
+      "loss": 1.1057,
+      "step": 11387
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.304801450386132e-05,
+      "loss": 1.0726,
+      "step": 11388
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3046649578126718e-05,
+      "loss": 1.0911,
+      "step": 11389
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3045284632676535e-05,
+      "loss": 1.0351,
+      "step": 11390
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.304391966753649e-05,
+      "loss": 1.1157,
+      "step": 11391
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3042554682732291e-05,
+      "loss": 1.1533,
+      "step": 11392
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3041189678289654e-05,
+      "loss": 1.0886,
+      "step": 11393
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3039824654234285e-05,
+      "loss": 1.0692,
+      "step": 11394
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3038459610591905e-05,
+      "loss": 1.0964,
+      "step": 11395
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3037094547388216e-05,
+      "loss": 1.0039,
+      "step": 11396
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3035729464648945e-05,
+      "loss": 1.0946,
+      "step": 11397
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3034364362399794e-05,
+      "loss": 1.0193,
+      "step": 11398
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3032999240666483e-05,
+      "loss": 0.9985,
+      "step": 11399
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3031634099474726e-05,
+      "loss": 1.0922,
+      "step": 11400
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3030268938850237e-05,
+      "loss": 1.0828,
+      "step": 11401
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3028903758818733e-05,
+      "loss": 1.101,
+      "step": 11402
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3027538559405928e-05,
+      "loss": 1.0845,
+      "step": 11403
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.302617334063754e-05,
+      "loss": 1.1101,
+      "step": 11404
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3024808102539279e-05,
+      "loss": 1.0474,
+      "step": 11405
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3023442845136874e-05,
+      "loss": 1.0153,
+      "step": 11406
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.302207756845603e-05,
+      "loss": 1.1543,
+      "step": 11407
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3020712272522473e-05,
+      "loss": 1.0248,
+      "step": 11408
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3019346957361916e-05,
+      "loss": 1.0877,
+      "step": 11409
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3017981623000078e-05,
+      "loss": 1.1169,
+      "step": 11410
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3016616269462675e-05,
+      "loss": 1.0262,
+      "step": 11411
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3015250896775435e-05,
+      "loss": 1.0205,
+      "step": 11412
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3013885504964068e-05,
+      "loss": 1.0507,
+      "step": 11413
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3012520094054297e-05,
+      "loss": 1.0299,
+      "step": 11414
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3011154664071843e-05,
+      "loss": 1.0573,
+      "step": 11415
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3009789215042424e-05,
+      "loss": 1.0621,
+      "step": 11416
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3008423746991767e-05,
+      "loss": 1.116,
+      "step": 11417
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3007058259945583e-05,
+      "loss": 1.0487,
+      "step": 11418
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3005692753929601e-05,
+      "loss": 1.0469,
+      "step": 11419
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3004327228969542e-05,
+      "loss": 1.0373,
+      "step": 11420
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3002961685091126e-05,
+      "loss": 1.1367,
+      "step": 11421
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3001596122320075e-05,
+      "loss": 1.0403,
+      "step": 11422
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3000230540682115e-05,
+      "loss": 1.059,
+      "step": 11423
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2998864940202967e-05,
+      "loss": 1.0886,
+      "step": 11424
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2997499320908355e-05,
+      "loss": 1.0456,
+      "step": 11425
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2996133682824004e-05,
+      "loss": 0.9715,
+      "step": 11426
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2994768025975635e-05,
+      "loss": 1.0904,
+      "step": 11427
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2993402350388977e-05,
+      "loss": 1.0372,
+      "step": 11428
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2992036656089753e-05,
+      "loss": 1.074,
+      "step": 11429
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2990670943103687e-05,
+      "loss": 1.0519,
+      "step": 11430
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2989305211456512e-05,
+      "loss": 1.0729,
+      "step": 11431
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2987939461173939e-05,
+      "loss": 1.0541,
+      "step": 11432
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2986573692281711e-05,
+      "loss": 1.0238,
+      "step": 11433
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2985207904805544e-05,
+      "loss": 1.0313,
+      "step": 11434
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2983842098771171e-05,
+      "loss": 0.986,
+      "step": 11435
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2982476274204314e-05,
+      "loss": 0.9998,
+      "step": 11436
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2981110431130708e-05,
+      "loss": 0.9981,
+      "step": 11437
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2979744569576071e-05,
+      "loss": 1.043,
+      "step": 11438
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2978378689566143e-05,
+      "loss": 1.095,
+      "step": 11439
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2977012791126646e-05,
+      "loss": 1.0904,
+      "step": 11440
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2975646874283309e-05,
+      "loss": 1.1446,
+      "step": 11441
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2974280939061866e-05,
+      "loss": 1.1411,
+      "step": 11442
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2972914985488042e-05,
+      "loss": 1.1071,
+      "step": 11443
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2971549013587572e-05,
+      "loss": 1.0179,
+      "step": 11444
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.297018302338618e-05,
+      "loss": 1.0646,
+      "step": 11445
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2968817014909604e-05,
+      "loss": 1.1417,
+      "step": 11446
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.296745098818357e-05,
+      "loss": 1.0124,
+      "step": 11447
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2966084943233817e-05,
+      "loss": 1.0407,
+      "step": 11448
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2964718880086064e-05,
+      "loss": 0.9871,
+      "step": 11449
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2963352798766058e-05,
+      "loss": 1.0472,
+      "step": 11450
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2961986699299523e-05,
+      "loss": 1.1272,
+      "step": 11451
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2960620581712193e-05,
+      "loss": 1.0988,
+      "step": 11452
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2959254446029804e-05,
+      "loss": 1.0677,
+      "step": 11453
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2957888292278084e-05,
+      "loss": 1.0253,
+      "step": 11454
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2956522120482773e-05,
+      "loss": 1.0964,
+      "step": 11455
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2955155930669603e-05,
+      "loss": 1.0974,
+      "step": 11456
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.295378972286431e-05,
+      "loss": 1.0022,
+      "step": 11457
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2952423497092628e-05,
+      "loss": 0.9744,
+      "step": 11458
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.295105725338029e-05,
+      "loss": 1.0486,
+      "step": 11459
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2949690991753036e-05,
+      "loss": 1.0397,
+      "step": 11460
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2948324712236602e-05,
+      "loss": 1.0468,
+      "step": 11461
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.294695841485672e-05,
+      "loss": 1.0857,
+      "step": 11462
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.294559209963913e-05,
+      "loss": 1.0536,
+      "step": 11463
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2944225766609568e-05,
+      "loss": 1.0771,
+      "step": 11464
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2942859415793773e-05,
+      "loss": 1.0633,
+      "step": 11465
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2941493047217481e-05,
+      "loss": 1.0788,
+      "step": 11466
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2940126660906428e-05,
+      "loss": 1.0774,
+      "step": 11467
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.293876025688636e-05,
+      "loss": 1.022,
+      "step": 11468
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2937393835183007e-05,
+      "loss": 1.0109,
+      "step": 11469
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2936027395822113e-05,
+      "loss": 1.049,
+      "step": 11470
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2934660938829415e-05,
+      "loss": 1.0187,
+      "step": 11471
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2933294464230656e-05,
+      "loss": 1.01,
+      "step": 11472
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2931927972051576e-05,
+      "loss": 1.081,
+      "step": 11473
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2930561462317907e-05,
+      "loss": 1.0912,
+      "step": 11474
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2929194935055402e-05,
+      "loss": 1.0586,
+      "step": 11475
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2927828390289794e-05,
+      "loss": 0.9509,
+      "step": 11476
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2926461828046828e-05,
+      "loss": 1.0076,
+      "step": 11477
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.292509524835224e-05,
+      "loss": 1.0338,
+      "step": 11478
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.292372865123178e-05,
+      "loss": 1.035,
+      "step": 11479
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2922362036711185e-05,
+      "loss": 1.0448,
+      "step": 11480
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2920995404816204e-05,
+      "loss": 1.0567,
+      "step": 11481
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.291962875557257e-05,
+      "loss": 1.0226,
+      "step": 11482
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2918262089006035e-05,
+      "loss": 1.0501,
+      "step": 11483
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2916895405142338e-05,
+      "loss": 1.0449,
+      "step": 11484
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2915528704007225e-05,
+      "loss": 1.1011,
+      "step": 11485
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.291416198562644e-05,
+      "loss": 1.0155,
+      "step": 11486
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2912795250025728e-05,
+      "loss": 1.0131,
+      "step": 11487
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2911428497230834e-05,
+      "loss": 1.0712,
+      "step": 11488
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.29100617272675e-05,
+      "loss": 1.0651,
+      "step": 11489
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2908694940161476e-05,
+      "loss": 1.036,
+      "step": 11490
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2907328135938506e-05,
+      "loss": 1.0244,
+      "step": 11491
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.290596131462434e-05,
+      "loss": 1.0608,
+      "step": 11492
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2904594476244719e-05,
+      "loss": 1.0655,
+      "step": 11493
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2903227620825391e-05,
+      "loss": 1.0348,
+      "step": 11494
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2901860748392105e-05,
+      "loss": 1.0099,
+      "step": 11495
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.290049385897061e-05,
+      "loss": 1.0084,
+      "step": 11496
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2899126952586652e-05,
+      "loss": 1.0604,
+      "step": 11497
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2897760029265977e-05,
+      "loss": 1.0376,
+      "step": 11498
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2896393089034335e-05,
+      "loss": 1.0624,
+      "step": 11499
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2895026131917478e-05,
+      "loss": 1.0788,
+      "step": 11500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2893659157941151e-05,
+      "loss": 1.0128,
+      "step": 11501
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2892292167131107e-05,
+      "loss": 1.0357,
+      "step": 11502
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2890925159513093e-05,
+      "loss": 1.1065,
+      "step": 11503
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2889558135112857e-05,
+      "loss": 1.0376,
+      "step": 11504
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2888191093956157e-05,
+      "loss": 1.0691,
+      "step": 11505
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2886824036068737e-05,
+      "loss": 1.1109,
+      "step": 11506
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2885456961476352e-05,
+      "loss": 1.0551,
+      "step": 11507
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.288408987020475e-05,
+      "loss": 1.0476,
+      "step": 11508
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2882722762279684e-05,
+      "loss": 1.1027,
+      "step": 11509
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2881355637726907e-05,
+      "loss": 1.0344,
+      "step": 11510
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.287998849657217e-05,
+      "loss": 1.0175,
+      "step": 11511
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2878621338841229e-05,
+      "loss": 1.0075,
+      "step": 11512
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2877254164559832e-05,
+      "loss": 1.0814,
+      "step": 11513
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2875886973753735e-05,
+      "loss": 1.1445,
+      "step": 11514
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2874519766448691e-05,
+      "loss": 1.0594,
+      "step": 11515
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2873152542670453e-05,
+      "loss": 1.0999,
+      "step": 11516
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2871785302444779e-05,
+      "loss": 1.1383,
+      "step": 11517
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2870418045797415e-05,
+      "loss": 1.085,
+      "step": 11518
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2869050772754128e-05,
+      "loss": 1.1549,
+      "step": 11519
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.286768348334066e-05,
+      "loss": 1.0154,
+      "step": 11520
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2866316177582778e-05,
+      "loss": 1.0136,
+      "step": 11521
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2864948855506227e-05,
+      "loss": 1.1918,
+      "step": 11522
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2863581517136775e-05,
+      "loss": 1.162,
+      "step": 11523
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2862214162500171e-05,
+      "loss": 1.0925,
+      "step": 11524
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.286084679162217e-05,
+      "loss": 1.0435,
+      "step": 11525
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2859479404528531e-05,
+      "loss": 1.0178,
+      "step": 11526
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2858112001245016e-05,
+      "loss": 1.0246,
+      "step": 11527
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2856744581797374e-05,
+      "loss": 1.0272,
+      "step": 11528
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2855377146211369e-05,
+      "loss": 0.9737,
+      "step": 11529
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2854009694512758e-05,
+      "loss": 1.0269,
+      "step": 11530
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2852642226727299e-05,
+      "loss": 1.0041,
+      "step": 11531
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.285127474288075e-05,
+      "loss": 1.0556,
+      "step": 11532
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.284990724299887e-05,
+      "loss": 1.0793,
+      "step": 11533
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2848539727107421e-05,
+      "loss": 1.0155,
+      "step": 11534
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.284717219523216e-05,
+      "loss": 0.9997,
+      "step": 11535
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2845804647398848e-05,
+      "loss": 1.03,
+      "step": 11536
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2844437083633245e-05,
+      "loss": 1.1101,
+      "step": 11537
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2843069503961111e-05,
+      "loss": 1.0514,
+      "step": 11538
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2841701908408208e-05,
+      "loss": 1.0512,
+      "step": 11539
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2840334297000296e-05,
+      "loss": 1.0006,
+      "step": 11540
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.283896666976314e-05,
+      "loss": 1.0982,
+      "step": 11541
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2837599026722498e-05,
+      "loss": 1.1495,
+      "step": 11542
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2836231367904133e-05,
+      "loss": 1.0417,
+      "step": 11543
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2834863693333805e-05,
+      "loss": 1.0053,
+      "step": 11544
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2833496003037281e-05,
+      "loss": 0.9844,
+      "step": 11545
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2832128297040323e-05,
+      "loss": 1.0349,
+      "step": 11546
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2830760575368693e-05,
+      "loss": 1.099,
+      "step": 11547
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2829392838048156e-05,
+      "loss": 1.0369,
+      "step": 11548
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2828025085104471e-05,
+      "loss": 1.0497,
+      "step": 11549
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2826657316563412e-05,
+      "loss": 1.0929,
+      "step": 11550
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2825289532450734e-05,
+      "loss": 1.0297,
+      "step": 11551
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2823921732792204e-05,
+      "loss": 0.9547,
+      "step": 11552
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2822553917613588e-05,
+      "loss": 1.1141,
+      "step": 11553
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2821186086940653e-05,
+      "loss": 1.0627,
+      "step": 11554
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2819818240799164e-05,
+      "loss": 1.0613,
+      "step": 11555
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2818450379214886e-05,
+      "loss": 1.0367,
+      "step": 11556
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2817082502213587e-05,
+      "loss": 1.0538,
+      "step": 11557
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2815714609821026e-05,
+      "loss": 0.9788,
+      "step": 11558
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.281434670206298e-05,
+      "loss": 1.0577,
+      "step": 11559
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2812978778965207e-05,
+      "loss": 1.0298,
+      "step": 11560
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2811610840553486e-05,
+      "loss": 1.0507,
+      "step": 11561
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2810242886853572e-05,
+      "loss": 1.0635,
+      "step": 11562
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.280887491789124e-05,
+      "loss": 1.0585,
+      "step": 11563
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2807506933692255e-05,
+      "loss": 1.0633,
+      "step": 11564
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2806138934282392e-05,
+      "loss": 1.1804,
+      "step": 11565
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2804770919687413e-05,
+      "loss": 1.0547,
+      "step": 11566
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2803402889933088e-05,
+      "loss": 1.0107,
+      "step": 11567
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2802034845045188e-05,
+      "loss": 1.0342,
+      "step": 11568
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.280066678504948e-05,
+      "loss": 1.1176,
+      "step": 11569
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.279929870997174e-05,
+      "loss": 1.086,
+      "step": 11570
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.279793061983773e-05,
+      "loss": 0.9975,
+      "step": 11571
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2796562514673227e-05,
+      "loss": 1.0212,
+      "step": 11572
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2795194394504e-05,
+      "loss": 1.059,
+      "step": 11573
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2793826259355819e-05,
+      "loss": 1.1023,
+      "step": 11574
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2792458109254455e-05,
+      "loss": 0.9789,
+      "step": 11575
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2791089944225684e-05,
+      "loss": 1.0861,
+      "step": 11576
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.278972176429527e-05,
+      "loss": 0.9693,
+      "step": 11577
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2788353569488991e-05,
+      "loss": 1.0281,
+      "step": 11578
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.278698535983262e-05,
+      "loss": 1.03,
+      "step": 11579
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2785617135351926e-05,
+      "loss": 1.0292,
+      "step": 11580
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2784248896072685e-05,
+      "loss": 1.0435,
+      "step": 11581
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.278288064202067e-05,
+      "loss": 1.1198,
+      "step": 11582
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2781512373221652e-05,
+      "loss": 1.0461,
+      "step": 11583
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.278014408970141e-05,
+      "loss": 1.0546,
+      "step": 11584
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2778775791485711e-05,
+      "loss": 1.0661,
+      "step": 11585
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2777407478600334e-05,
+      "loss": 1.0331,
+      "step": 11586
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2776039151071053e-05,
+      "loss": 0.9973,
+      "step": 11587
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2774670808923649e-05,
+      "loss": 1.0615,
+      "step": 11588
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2773302452183883e-05,
+      "loss": 1.0178,
+      "step": 11589
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2771934080877542e-05,
+      "loss": 1.0004,
+      "step": 11590
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2770565695030397e-05,
+      "loss": 0.9811,
+      "step": 11591
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2769197294668227e-05,
+      "loss": 1.0803,
+      "step": 11592
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2767828879816805e-05,
+      "loss": 0.9965,
+      "step": 11593
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2766460450501914e-05,
+      "loss": 1.1422,
+      "step": 11594
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2765092006749322e-05,
+      "loss": 1.0801,
+      "step": 11595
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2763723548584812e-05,
+      "loss": 1.083,
+      "step": 11596
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.276235507603416e-05,
+      "loss": 0.9982,
+      "step": 11597
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2760986589123145e-05,
+      "loss": 1.0132,
+      "step": 11598
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2759618087877545e-05,
+      "loss": 0.9925,
+      "step": 11599
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2758249572323134e-05,
+      "loss": 0.9844,
+      "step": 11600
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2756881042485695e-05,
+      "loss": 1.0665,
+      "step": 11601
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2755512498391003e-05,
+      "loss": 1.0472,
+      "step": 11602
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2754143940064844e-05,
+      "loss": 1.0819,
+      "step": 11603
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2752775367532987e-05,
+      "loss": 1.0724,
+      "step": 11604
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2751406780821223e-05,
+      "loss": 1.0723,
+      "step": 11605
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.275003817995532e-05,
+      "loss": 0.9708,
+      "step": 11606
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.274866956496107e-05,
+      "loss": 1.1082,
+      "step": 11607
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2747300935864244e-05,
+      "loss": 1.0675,
+      "step": 11608
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2745932292690628e-05,
+      "loss": 1.1096,
+      "step": 11609
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2744563635466e-05,
+      "loss": 1.0411,
+      "step": 11610
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2743194964216143e-05,
+      "loss": 1.0559,
+      "step": 11611
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2741826278966835e-05,
+      "loss": 1.0442,
+      "step": 11612
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2740457579743866e-05,
+      "loss": 1.0295,
+      "step": 11613
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2739088866573008e-05,
+      "loss": 1.0613,
+      "step": 11614
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.273772013948005e-05,
+      "loss": 1.0157,
+      "step": 11615
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2736351398490772e-05,
+      "loss": 1.0834,
+      "step": 11616
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2734982643630955e-05,
+      "loss": 1.0584,
+      "step": 11617
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2733613874926384e-05,
+      "loss": 1.0804,
+      "step": 11618
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2732245092402846e-05,
+      "loss": 1.0619,
+      "step": 11619
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2730876296086118e-05,
+      "loss": 1.0103,
+      "step": 11620
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2729507486001986e-05,
+      "loss": 1.0605,
+      "step": 11621
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2728138662176236e-05,
+      "loss": 1.1432,
+      "step": 11622
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2726769824634652e-05,
+      "loss": 1.0518,
+      "step": 11623
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2725400973403018e-05,
+      "loss": 1.0057,
+      "step": 11624
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2724032108507116e-05,
+      "loss": 1.0872,
+      "step": 11625
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2722663229972733e-05,
+      "loss": 1.0589,
+      "step": 11626
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2721294337825659e-05,
+      "loss": 0.9659,
+      "step": 11627
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2719925432091671e-05,
+      "loss": 1.0977,
+      "step": 11628
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2718556512796562e-05,
+      "loss": 1.0305,
+      "step": 11629
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2717187579966117e-05,
+      "loss": 1.034,
+      "step": 11630
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2715818633626115e-05,
+      "loss": 1.0765,
+      "step": 11631
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2714449673802355e-05,
+      "loss": 1.0135,
+      "step": 11632
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2713080700520612e-05,
+      "loss": 0.9815,
+      "step": 11633
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2711711713806683e-05,
+      "loss": 1.1046,
+      "step": 11634
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2710342713686347e-05,
+      "loss": 1.0122,
+      "step": 11635
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2708973700185399e-05,
+      "loss": 1.0912,
+      "step": 11636
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2707604673329623e-05,
+      "loss": 1.1048,
+      "step": 11637
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2706235633144805e-05,
+      "loss": 1.0806,
+      "step": 11638
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.270486657965674e-05,
+      "loss": 1.0428,
+      "step": 11639
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2703497512891209e-05,
+      "loss": 1.0864,
+      "step": 11640
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2702128432874007e-05,
+      "loss": 1.023,
+      "step": 11641
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.270075933963092e-05,
+      "loss": 1.0166,
+      "step": 11642
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2699390233187738e-05,
+      "loss": 1.0014,
+      "step": 11643
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2698021113570247e-05,
+      "loss": 1.0375,
+      "step": 11644
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2696651980804247e-05,
+      "loss": 1.0102,
+      "step": 11645
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2695282834915517e-05,
+      "loss": 1.0693,
+      "step": 11646
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2693913675929857e-05,
+      "loss": 1.0463,
+      "step": 11647
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2692544503873046e-05,
+      "loss": 1.0415,
+      "step": 11648
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2691175318770887e-05,
+      "loss": 1.0965,
+      "step": 11649
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2689806120649164e-05,
+      "loss": 1.0974,
+      "step": 11650
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.268843690953367e-05,
+      "loss": 1.0504,
+      "step": 11651
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2687067685450198e-05,
+      "loss": 1.0433,
+      "step": 11652
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2685698448424537e-05,
+      "loss": 1.0152,
+      "step": 11653
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2684329198482484e-05,
+      "loss": 1.1194,
+      "step": 11654
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2682959935649824e-05,
+      "loss": 1.0668,
+      "step": 11655
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2681590659952355e-05,
+      "loss": 1.0354,
+      "step": 11656
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2680221371415868e-05,
+      "loss": 1.0486,
+      "step": 11657
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2678852070066159e-05,
+      "loss": 1.0888,
+      "step": 11658
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2677482755929015e-05,
+      "loss": 0.9873,
+      "step": 11659
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2676113429030237e-05,
+      "loss": 1.0594,
+      "step": 11660
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2674744089395614e-05,
+      "loss": 1.0742,
+      "step": 11661
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.267337473705094e-05,
+      "loss": 1.1092,
+      "step": 11662
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2672005372022014e-05,
+      "loss": 1.0468,
+      "step": 11663
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2670635994334626e-05,
+      "loss": 1.0722,
+      "step": 11664
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.266926660401457e-05,
+      "loss": 1.066,
+      "step": 11665
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2667897201087644e-05,
+      "loss": 0.9835,
+      "step": 11666
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.266652778557964e-05,
+      "loss": 1.0517,
+      "step": 11667
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2665158357516358e-05,
+      "loss": 1.0981,
+      "step": 11668
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2663788916923592e-05,
+      "loss": 1.0862,
+      "step": 11669
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2662419463827135e-05,
+      "loss": 1.0362,
+      "step": 11670
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2661049998252788e-05,
+      "loss": 1.044,
+      "step": 11671
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2659680520226346e-05,
+      "loss": 1.0925,
+      "step": 11672
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2658311029773598e-05,
+      "loss": 0.96,
+      "step": 11673
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2656941526920351e-05,
+      "loss": 1.0747,
+      "step": 11674
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2655572011692397e-05,
+      "loss": 1.087,
+      "step": 11675
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2654202484115539e-05,
+      "loss": 1.0627,
+      "step": 11676
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2652832944215564e-05,
+      "loss": 1.0742,
+      "step": 11677
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2651463392018279e-05,
+      "loss": 1.0191,
+      "step": 11678
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.265009382754948e-05,
+      "loss": 0.9938,
+      "step": 11679
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2648724250834965e-05,
+      "loss": 0.9982,
+      "step": 11680
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2647354661900533e-05,
+      "loss": 1.0152,
+      "step": 11681
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2645985060771978e-05,
+      "loss": 1.1153,
+      "step": 11682
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2644615447475106e-05,
+      "loss": 1.0545,
+      "step": 11683
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.264324582203571e-05,
+      "loss": 1.0009,
+      "step": 11684
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2641876184479596e-05,
+      "loss": 1.0733,
+      "step": 11685
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2640506534832554e-05,
+      "loss": 1.0633,
+      "step": 11686
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2639136873120397e-05,
+      "loss": 1.0503,
+      "step": 11687
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.263776719936891e-05,
+      "loss": 0.9897,
+      "step": 11688
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2636397513603911e-05,
+      "loss": 0.9764,
+      "step": 11689
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2635027815851182e-05,
+      "loss": 1.057,
+      "step": 11690
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2633658106136538e-05,
+      "loss": 1.0062,
+      "step": 11691
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2632288384485773e-05,
+      "loss": 1.0262,
+      "step": 11692
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2630918650924688e-05,
+      "loss": 1.0586,
+      "step": 11693
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.262954890547909e-05,
+      "loss": 1.0898,
+      "step": 11694
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2628179148174773e-05,
+      "loss": 1.0255,
+      "step": 11695
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2626809379037546e-05,
+      "loss": 0.997,
+      "step": 11696
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2625439598093206e-05,
+      "loss": 1.0459,
+      "step": 11697
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2624069805367558e-05,
+      "loss": 1.026,
+      "step": 11698
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2622700000886403e-05,
+      "loss": 1.0913,
+      "step": 11699
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2621330184675547e-05,
+      "loss": 1.0371,
+      "step": 11700
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2619960356760789e-05,
+      "loss": 1.0787,
+      "step": 11701
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2618590517167934e-05,
+      "loss": 1.0807,
+      "step": 11702
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2617220665922786e-05,
+      "loss": 0.9666,
+      "step": 11703
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2615850803051148e-05,
+      "loss": 0.9925,
+      "step": 11704
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2614480928578823e-05,
+      "loss": 1.0594,
+      "step": 11705
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2613111042531618e-05,
+      "loss": 1.0093,
+      "step": 11706
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2611741144935334e-05,
+      "loss": 1.1701,
+      "step": 11707
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2610371235815776e-05,
+      "loss": 1.1461,
+      "step": 11708
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2609001315198753e-05,
+      "loss": 1.0569,
+      "step": 11709
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2607631383110065e-05,
+      "loss": 1.1184,
+      "step": 11710
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.260626143957552e-05,
+      "loss": 1.0867,
+      "step": 11711
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.260489148462092e-05,
+      "loss": 1.0362,
+      "step": 11712
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2603521518272075e-05,
+      "loss": 1.0897,
+      "step": 11713
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2602151540554789e-05,
+      "loss": 0.9566,
+      "step": 11714
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2600781551494864e-05,
+      "loss": 1.1072,
+      "step": 11715
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2599411551118116e-05,
+      "loss": 0.9935,
+      "step": 11716
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2598041539450343e-05,
+      "loss": 1.0507,
+      "step": 11717
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2596671516517355e-05,
+      "loss": 1.0903,
+      "step": 11718
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2595301482344955e-05,
+      "loss": 0.9966,
+      "step": 11719
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2593931436958959e-05,
+      "loss": 1.0528,
+      "step": 11720
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2592561380385162e-05,
+      "loss": 1.0982,
+      "step": 11721
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2591191312649383e-05,
+      "loss": 1.0524,
+      "step": 11722
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2589821233777426e-05,
+      "loss": 1.0219,
+      "step": 11723
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2588451143795092e-05,
+      "loss": 1.0958,
+      "step": 11724
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2587081042728202e-05,
+      "loss": 1.004,
+      "step": 11725
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2585710930602552e-05,
+      "loss": 1.074,
+      "step": 11726
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2584340807443962e-05,
+      "loss": 1.0083,
+      "step": 11727
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2582970673278227e-05,
+      "loss": 1.027,
+      "step": 11728
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2581600528131172e-05,
+      "loss": 1.1082,
+      "step": 11729
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2580230372028592e-05,
+      "loss": 1.0331,
+      "step": 11730
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2578860204996308e-05,
+      "loss": 1.0568,
+      "step": 11731
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2577490027060118e-05,
+      "loss": 1.1138,
+      "step": 11732
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2576119838245843e-05,
+      "loss": 1.0628,
+      "step": 11733
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2574749638579287e-05,
+      "loss": 1.1187,
+      "step": 11734
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.257337942808626e-05,
+      "loss": 1.0145,
+      "step": 11735
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2572009206792574e-05,
+      "loss": 1.0769,
+      "step": 11736
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.257063897472404e-05,
+      "loss": 1.0197,
+      "step": 11737
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2569268731906468e-05,
+      "loss": 1.03,
+      "step": 11738
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2567898478365667e-05,
+      "loss": 1.0706,
+      "step": 11739
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2566528214127453e-05,
+      "loss": 1.0564,
+      "step": 11740
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2565157939217635e-05,
+      "loss": 1.02,
+      "step": 11741
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2563787653662025e-05,
+      "loss": 1.0336,
+      "step": 11742
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2562417357486434e-05,
+      "loss": 1.147,
+      "step": 11743
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2561047050716673e-05,
+      "loss": 0.9759,
+      "step": 11744
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2559676733378556e-05,
+      "loss": 0.9818,
+      "step": 11745
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2558306405497897e-05,
+      "loss": 1.0391,
+      "step": 11746
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2556936067100505e-05,
+      "loss": 1.0565,
+      "step": 11747
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2555565718212197e-05,
+      "loss": 0.9984,
+      "step": 11748
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2554195358858783e-05,
+      "loss": 0.9646,
+      "step": 11749
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2552824989066075e-05,
+      "loss": 1.0315,
+      "step": 11750
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2551454608859889e-05,
+      "loss": 1.0463,
+      "step": 11751
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.255008421826604e-05,
+      "loss": 1.0537,
+      "step": 11752
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.254871381731034e-05,
+      "loss": 1.0166,
+      "step": 11753
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2547343406018601e-05,
+      "loss": 0.9855,
+      "step": 11754
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2545972984416641e-05,
+      "loss": 1.0973,
+      "step": 11755
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2544602552530273e-05,
+      "loss": 1.0227,
+      "step": 11756
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2543232110385308e-05,
+      "loss": 1.0328,
+      "step": 11757
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.254186165800757e-05,
+      "loss": 1.0551,
+      "step": 11758
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.254049119542286e-05,
+      "loss": 1.0635,
+      "step": 11759
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2539120722657009e-05,
+      "loss": 1.016,
+      "step": 11760
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.253775023973582e-05,
+      "loss": 1.0158,
+      "step": 11761
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2536379746685117e-05,
+      "loss": 1.0246,
+      "step": 11762
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2535009243530706e-05,
+      "loss": 1.0847,
+      "step": 11763
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2533638730298414e-05,
+      "loss": 1.0621,
+      "step": 11764
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.253226820701405e-05,
+      "loss": 0.9915,
+      "step": 11765
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.253089767370343e-05,
+      "loss": 1.0315,
+      "step": 11766
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2529527130392379e-05,
+      "loss": 0.9655,
+      "step": 11767
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2528156577106703e-05,
+      "loss": 1.1257,
+      "step": 11768
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2526786013872225e-05,
+      "loss": 1.0641,
+      "step": 11769
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2525415440714759e-05,
+      "loss": 0.9999,
+      "step": 11770
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2524044857660127e-05,
+      "loss": 0.9764,
+      "step": 11771
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.252267426473414e-05,
+      "loss": 0.9931,
+      "step": 11772
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2521303661962623e-05,
+      "loss": 1.0431,
+      "step": 11773
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2519933049371384e-05,
+      "loss": 0.9827,
+      "step": 11774
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2518562426986253e-05,
+      "loss": 1.073,
+      "step": 11775
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2517191794833039e-05,
+      "loss": 1.0384,
+      "step": 11776
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2515821152937564e-05,
+      "loss": 1.0372,
+      "step": 11777
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2514450501325645e-05,
+      "loss": 1.0538,
+      "step": 11778
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2513079840023104e-05,
+      "loss": 1.039,
+      "step": 11779
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2511709169055757e-05,
+      "loss": 1.1423,
+      "step": 11780
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2510338488449423e-05,
+      "loss": 1.0341,
+      "step": 11781
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.250896779822992e-05,
+      "loss": 0.9997,
+      "step": 11782
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2507597098423072e-05,
+      "loss": 1.0624,
+      "step": 11783
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2506226389054696e-05,
+      "loss": 1.0111,
+      "step": 11784
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2504855670150612e-05,
+      "loss": 1.0671,
+      "step": 11785
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2503484941736641e-05,
+      "loss": 1.0179,
+      "step": 11786
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2502114203838603e-05,
+      "loss": 1.0761,
+      "step": 11787
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2500743456482313e-05,
+      "loss": 1.0812,
+      "step": 11788
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2499372699693602e-05,
+      "loss": 1.0926,
+      "step": 11789
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.249800193349828e-05,
+      "loss": 0.9638,
+      "step": 11790
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2496631157922176e-05,
+      "loss": 1.0653,
+      "step": 11791
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2495260372991107e-05,
+      "loss": 1.017,
+      "step": 11792
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2493889578730895e-05,
+      "loss": 1.026,
+      "step": 11793
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2492518775167364e-05,
+      "loss": 1.045,
+      "step": 11794
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.249114796232633e-05,
+      "loss": 0.986,
+      "step": 11795
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2489777140233618e-05,
+      "loss": 0.9879,
+      "step": 11796
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2488406308915052e-05,
+      "loss": 1.0054,
+      "step": 11797
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2487035468396455e-05,
+      "loss": 1.0274,
+      "step": 11798
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.248566461870364e-05,
+      "loss": 0.9971,
+      "step": 11799
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2484293759862443e-05,
+      "loss": 1.0401,
+      "step": 11800
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2482922891898672e-05,
+      "loss": 1.0749,
+      "step": 11801
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2481552014838164e-05,
+      "loss": 1.1195,
+      "step": 11802
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.248018112870673e-05,
+      "loss": 1.074,
+      "step": 11803
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2478810233530203e-05,
+      "loss": 1.0722,
+      "step": 11804
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2477439329334398e-05,
+      "loss": 1.0798,
+      "step": 11805
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2476068416145146e-05,
+      "loss": 0.9968,
+      "step": 11806
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2474697493988266e-05,
+      "loss": 1.0744,
+      "step": 11807
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2473326562889583e-05,
+      "loss": 1.0172,
+      "step": 11808
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2471955622874921e-05,
+      "loss": 1.0699,
+      "step": 11809
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2470584673970101e-05,
+      "loss": 1.0149,
+      "step": 11810
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2469213716200956e-05,
+      "loss": 1.026,
+      "step": 11811
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.24678427495933e-05,
+      "loss": 1.0884,
+      "step": 11812
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2466471774172968e-05,
+      "loss": 1.0973,
+      "step": 11813
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2465100789965773e-05,
+      "loss": 1.0473,
+      "step": 11814
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2463729796997552e-05,
+      "loss": 1.0163,
+      "step": 11815
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2462358795294123e-05,
+      "loss": 1.0536,
+      "step": 11816
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2460987784881313e-05,
+      "loss": 1.0586,
+      "step": 11817
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.245961676578495e-05,
+      "loss": 1.0121,
+      "step": 11818
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2458245738030854e-05,
+      "loss": 1.0762,
+      "step": 11819
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2456874701644856e-05,
+      "loss": 1.0353,
+      "step": 11820
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2455503656652778e-05,
+      "loss": 1.0833,
+      "step": 11821
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2454132603080451e-05,
+      "loss": 0.9957,
+      "step": 11822
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2452761540953698e-05,
+      "loss": 1.0499,
+      "step": 11823
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2451390470298346e-05,
+      "loss": 1.0655,
+      "step": 11824
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.245001939114022e-05,
+      "loss": 1.0305,
+      "step": 11825
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.244864830350515e-05,
+      "loss": 1.0196,
+      "step": 11826
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2447277207418962e-05,
+      "loss": 1.0754,
+      "step": 11827
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2445906102907481e-05,
+      "loss": 1.033,
+      "step": 11828
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2444534989996537e-05,
+      "loss": 1.0582,
+      "step": 11829
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2443163868711958e-05,
+      "loss": 1.1167,
+      "step": 11830
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2441792739079567e-05,
+      "loss": 1.0485,
+      "step": 11831
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2440421601125196e-05,
+      "loss": 1.0704,
+      "step": 11832
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2439050454874673e-05,
+      "loss": 0.9617,
+      "step": 11833
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2437679300353825e-05,
+      "loss": 1.0235,
+      "step": 11834
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2436308137588478e-05,
+      "loss": 1.0859,
+      "step": 11835
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2434936966604462e-05,
+      "loss": 0.9879,
+      "step": 11836
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2433565787427606e-05,
+      "loss": 1.1044,
+      "step": 11837
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.243219460008374e-05,
+      "loss": 1.0437,
+      "step": 11838
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.243082340459869e-05,
+      "loss": 1.04,
+      "step": 11839
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2429452200998291e-05,
+      "loss": 1.0423,
+      "step": 11840
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2428080989308359e-05,
+      "loss": 1.141,
+      "step": 11841
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.242670976955474e-05,
+      "loss": 1.1011,
+      "step": 11842
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2425338541763249e-05,
+      "loss": 1.0683,
+      "step": 11843
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2423967305959727e-05,
+      "loss": 0.9951,
+      "step": 11844
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2422596062169995e-05,
+      "loss": 1.0029,
+      "step": 11845
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.242122481041989e-05,
+      "loss": 0.9942,
+      "step": 11846
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2419853550735234e-05,
+      "loss": 0.9695,
+      "step": 11847
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2418482283141866e-05,
+      "loss": 0.9988,
+      "step": 11848
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2417111007665611e-05,
+      "loss": 1.0343,
+      "step": 11849
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.24157397243323e-05,
+      "loss": 1.0108,
+      "step": 11850
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2414368433167765e-05,
+      "loss": 1.0406,
+      "step": 11851
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2412997134197837e-05,
+      "loss": 1.0364,
+      "step": 11852
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2411625827448343e-05,
+      "loss": 1.0369,
+      "step": 11853
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2410254512945117e-05,
+      "loss": 0.9949,
+      "step": 11854
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2408883190713996e-05,
+      "loss": 1.1493,
+      "step": 11855
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2407511860780798e-05,
+      "loss": 1.0242,
+      "step": 11856
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2406140523171371e-05,
+      "loss": 1.036,
+      "step": 11857
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.240476917791153e-05,
+      "loss": 0.9827,
+      "step": 11858
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2403397825027122e-05,
+      "loss": 1.0219,
+      "step": 11859
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2402026464543968e-05,
+      "loss": 1.0279,
+      "step": 11860
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2400655096487903e-05,
+      "loss": 1.0686,
+      "step": 11861
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2399283720884761e-05,
+      "loss": 0.9873,
+      "step": 11862
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2397912337760374e-05,
+      "loss": 1.05,
+      "step": 11863
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2396540947140573e-05,
+      "loss": 1.0733,
+      "step": 11864
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2395169549051193e-05,
+      "loss": 1.0533,
+      "step": 11865
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2393798143518065e-05,
+      "loss": 1.0942,
+      "step": 11866
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2392426730567021e-05,
+      "loss": 1.02,
+      "step": 11867
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2391055310223898e-05,
+      "loss": 0.9846,
+      "step": 11868
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2389683882514527e-05,
+      "loss": 1.1251,
+      "step": 11869
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2388312447464739e-05,
+      "loss": 0.9894,
+      "step": 11870
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2386941005100372e-05,
+      "loss": 1.0644,
+      "step": 11871
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2385569555447257e-05,
+      "loss": 1.0188,
+      "step": 11872
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2384198098531226e-05,
+      "loss": 1.0637,
+      "step": 11873
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2382826634378115e-05,
+      "loss": 1.0868,
+      "step": 11874
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.238145516301376e-05,
+      "loss": 1.0527,
+      "step": 11875
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2380083684463993e-05,
+      "loss": 1.0423,
+      "step": 11876
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2378712198754648e-05,
+      "loss": 1.0599,
+      "step": 11877
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.237734070591156e-05,
+      "loss": 1.0194,
+      "step": 11878
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2375969205960566e-05,
+      "loss": 1.0162,
+      "step": 11879
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2374597698927495e-05,
+      "loss": 1.0477,
+      "step": 11880
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2373226184838188e-05,
+      "loss": 1.0091,
+      "step": 11881
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2371854663718477e-05,
+      "loss": 1.148,
+      "step": 11882
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2370483135594196e-05,
+      "loss": 1.0637,
+      "step": 11883
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2369111600491184e-05,
+      "loss": 1.0203,
+      "step": 11884
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.236774005843527e-05,
+      "loss": 1.0611,
+      "step": 11885
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2366368509452298e-05,
+      "loss": 1.0623,
+      "step": 11886
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2364996953568093e-05,
+      "loss": 1.0804,
+      "step": 11887
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2363625390808504e-05,
+      "loss": 0.9988,
+      "step": 11888
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2362253821199354e-05,
+      "loss": 1.0002,
+      "step": 11889
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2360882244766492e-05,
+      "loss": 0.9759,
+      "step": 11890
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2359510661535742e-05,
+      "loss": 0.996,
+      "step": 11891
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2358139071532948e-05,
+      "loss": 1.0943,
+      "step": 11892
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2356767474783942e-05,
+      "loss": 1.064,
+      "step": 11893
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2355395871314563e-05,
+      "loss": 1.0136,
+      "step": 11894
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2354024261150647e-05,
+      "loss": 1.0636,
+      "step": 11895
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2352652644318032e-05,
+      "loss": 1.0343,
+      "step": 11896
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2351281020842553e-05,
+      "loss": 0.9997,
+      "step": 11897
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2349909390750046e-05,
+      "loss": 1.0239,
+      "step": 11898
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2348537754066355e-05,
+      "loss": 1.0266,
+      "step": 11899
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2347166110817305e-05,
+      "loss": 1.0133,
+      "step": 11900
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2345794461028749e-05,
+      "loss": 1.0546,
+      "step": 11901
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2344422804726513e-05,
+      "loss": 1.056,
+      "step": 11902
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2343051141936437e-05,
+      "loss": 0.9678,
+      "step": 11903
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.234167947268436e-05,
+      "loss": 1.0531,
+      "step": 11904
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2340307796996123e-05,
+      "loss": 0.9783,
+      "step": 11905
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2338936114897558e-05,
+      "loss": 1.0268,
+      "step": 11906
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2337564426414505e-05,
+      "loss": 1.0816,
+      "step": 11907
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2336192731572804e-05,
+      "loss": 1.0246,
+      "step": 11908
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2334821030398293e-05,
+      "loss": 1.0713,
+      "step": 11909
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2333449322916812e-05,
+      "loss": 1.0184,
+      "step": 11910
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2332077609154195e-05,
+      "loss": 1.0494,
+      "step": 11911
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2330705889136288e-05,
+      "loss": 1.0744,
+      "step": 11912
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2329334162888922e-05,
+      "loss": 1.0249,
+      "step": 11913
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2327962430437938e-05,
+      "loss": 1.028,
+      "step": 11914
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2326590691809178e-05,
+      "loss": 1.0182,
+      "step": 11915
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2325218947028483e-05,
+      "loss": 1.0269,
+      "step": 11916
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2323847196121686e-05,
+      "loss": 1.0161,
+      "step": 11917
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2322475439114629e-05,
+      "loss": 1.0785,
+      "step": 11918
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2321103676033154e-05,
+      "loss": 1.0187,
+      "step": 11919
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2319731906903098e-05,
+      "loss": 1.0181,
+      "step": 11920
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2318360131750303e-05,
+      "loss": 1.0871,
+      "step": 11921
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2316988350600608e-05,
+      "loss": 1.0982,
+      "step": 11922
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.231561656347985e-05,
+      "loss": 1.0565,
+      "step": 11923
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2314244770413874e-05,
+      "loss": 0.9953,
+      "step": 11924
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2312872971428515e-05,
+      "loss": 1.011,
+      "step": 11925
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2311501166549622e-05,
+      "loss": 1.0502,
+      "step": 11926
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2310129355803024e-05,
+      "loss": 1.0302,
+      "step": 11927
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2308757539214573e-05,
+      "loss": 1.0648,
+      "step": 11928
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2307385716810098e-05,
+      "loss": 1.1014,
+      "step": 11929
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.230601388861545e-05,
+      "loss": 1.013,
+      "step": 11930
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2304642054656463e-05,
+      "loss": 1.0237,
+      "step": 11931
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2303270214958987e-05,
+      "loss": 1.0875,
+      "step": 11932
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2301898369548853e-05,
+      "loss": 0.9905,
+      "step": 11933
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2300526518451906e-05,
+      "loss": 1.0966,
+      "step": 11934
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2299154661693986e-05,
+      "loss": 1.0467,
+      "step": 11935
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2297782799300938e-05,
+      "loss": 1.164,
+      "step": 11936
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.22964109312986e-05,
+      "loss": 1.0456,
+      "step": 11937
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2295039057712819e-05,
+      "loss": 1.0518,
+      "step": 11938
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.229366717856943e-05,
+      "loss": 1.1095,
+      "step": 11939
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2292295293894279e-05,
+      "loss": 1.0348,
+      "step": 11940
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2290923403713207e-05,
+      "loss": 1.0794,
+      "step": 11941
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2289551508052053e-05,
+      "loss": 0.9986,
+      "step": 11942
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2288179606936665e-05,
+      "loss": 1.0952,
+      "step": 11943
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2286807700392884e-05,
+      "loss": 1.0977,
+      "step": 11944
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2285435788446548e-05,
+      "loss": 1.0308,
+      "step": 11945
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2284063871123503e-05,
+      "loss": 1.0923,
+      "step": 11946
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2282691948449592e-05,
+      "loss": 1.0052,
+      "step": 11947
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2281320020450657e-05,
+      "loss": 1.0583,
+      "step": 11948
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.227994808715254e-05,
+      "loss": 0.9892,
+      "step": 11949
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2278576148581082e-05,
+      "loss": 1.0042,
+      "step": 11950
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2277204204762132e-05,
+      "loss": 1.156,
+      "step": 11951
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2275832255721526e-05,
+      "loss": 1.014,
+      "step": 11952
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2274460301485114e-05,
+      "loss": 1.0218,
+      "step": 11953
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2273088342078733e-05,
+      "loss": 0.9909,
+      "step": 11954
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2271716377528231e-05,
+      "loss": 1.092,
+      "step": 11955
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.227034440785945e-05,
+      "loss": 1.0259,
+      "step": 11956
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2268972433098231e-05,
+      "loss": 1.0819,
+      "step": 11957
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2267600453270421e-05,
+      "loss": 1.0559,
+      "step": 11958
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2266228468401864e-05,
+      "loss": 1.048,
+      "step": 11959
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.22648564785184e-05,
+      "loss": 1.0013,
+      "step": 11960
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.226348448364588e-05,
+      "loss": 1.106,
+      "step": 11961
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.226211248381014e-05,
+      "loss": 1.0173,
+      "step": 11962
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2260740479037029e-05,
+      "loss": 1.1132,
+      "step": 11963
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.225936846935239e-05,
+      "loss": 1.0614,
+      "step": 11964
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2257996454782068e-05,
+      "loss": 1.0513,
+      "step": 11965
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2256624435351909e-05,
+      "loss": 1.0297,
+      "step": 11966
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.225525241108775e-05,
+      "loss": 1.0035,
+      "step": 11967
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2253880382015446e-05,
+      "loss": 0.9643,
+      "step": 11968
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2252508348160831e-05,
+      "loss": 1.0653,
+      "step": 11969
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2251136309549763e-05,
+      "loss": 0.9676,
+      "step": 11970
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2249764266208073e-05,
+      "loss": 0.9942,
+      "step": 11971
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2248392218161615e-05,
+      "loss": 1.103,
+      "step": 11972
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2247020165436229e-05,
+      "loss": 1.106,
+      "step": 11973
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2245648108057767e-05,
+      "loss": 1.035,
+      "step": 11974
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2244276046052065e-05,
+      "loss": 1.0506,
+      "step": 11975
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2242903979444975e-05,
+      "loss": 1.0645,
+      "step": 11976
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2241531908262341e-05,
+      "loss": 1.0748,
+      "step": 11977
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2240159832530006e-05,
+      "loss": 0.9458,
+      "step": 11978
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2238787752273821e-05,
+      "loss": 1.0404,
+      "step": 11979
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2237415667519625e-05,
+      "loss": 0.9908,
+      "step": 11980
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2236043578293268e-05,
+      "loss": 1.0902,
+      "step": 11981
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2234671484620594e-05,
+      "loss": 1.0845,
+      "step": 11982
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.223329938652745e-05,
+      "loss": 1.0515,
+      "step": 11983
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2231927284039684e-05,
+      "loss": 0.9675,
+      "step": 11984
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2230555177183136e-05,
+      "loss": 1.0627,
+      "step": 11985
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2229183065983658e-05,
+      "loss": 1.067,
+      "step": 11986
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2227810950467094e-05,
+      "loss": 1.0785,
+      "step": 11987
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.222643883065929e-05,
+      "loss": 1.0504,
+      "step": 11988
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2225066706586095e-05,
+      "loss": 1.0445,
+      "step": 11989
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.222369457827335e-05,
+      "loss": 1.0206,
+      "step": 11990
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2222322445746908e-05,
+      "loss": 1.0868,
+      "step": 11991
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.222095030903261e-05,
+      "loss": 1.0166,
+      "step": 11992
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2219578168156307e-05,
+      "loss": 1.0366,
+      "step": 11993
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2218206023143843e-05,
+      "loss": 1.0272,
+      "step": 11994
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2216833874021067e-05,
+      "loss": 1.0396,
+      "step": 11995
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2215461720813824e-05,
+      "loss": 1.0564,
+      "step": 11996
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2214089563547966e-05,
+      "loss": 1.0829,
+      "step": 11997
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.221271740224933e-05,
+      "loss": 1.0463,
+      "step": 11998
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2211345236943771e-05,
+      "loss": 0.9847,
+      "step": 11999
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2209973067657137e-05,
+      "loss": 1.0129,
+      "step": 12000
+    },
+    {
+      "epoch": 0.51,
+      "eval_text_loss": 0.7898672819137573,
+      "eval_text_runtime": 1.3968,
+      "eval_text_samples_per_second": 52.979,
+      "eval_text_steps_per_second": 1.432,
+      "step": 12000
+    },
+    {
+      "epoch": 0.51,
+      "eval_svit_loss": 1.124644160270691,
+      "eval_svit_runtime": 5.1058,
+      "eval_svit_samples_per_second": 133.377,
+      "eval_svit_steps_per_second": 2.154,
+      "step": 12000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.220860089441527e-05,
+      "loss": 1.1423,
+      "step": 12001
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2207228717244022e-05,
+      "loss": 0.9417,
+      "step": 12002
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2205856536169237e-05,
+      "loss": 1.0536,
+      "step": 12003
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2204484351216768e-05,
+      "loss": 1.1344,
+      "step": 12004
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2203112162412455e-05,
+      "loss": 1.0367,
+      "step": 12005
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2201739969782154e-05,
+      "loss": 1.0461,
+      "step": 12006
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2200367773351707e-05,
+      "loss": 1.1208,
+      "step": 12007
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2198995573146964e-05,
+      "loss": 1.0766,
+      "step": 12008
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2197623369193767e-05,
+      "loss": 1.1112,
+      "step": 12009
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2196251161517975e-05,
+      "loss": 1.0591,
+      "step": 12010
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2194878950145428e-05,
+      "loss": 1.0512,
+      "step": 12011
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.219350673510198e-05,
+      "loss": 1.0432,
+      "step": 12012
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.219213451641347e-05,
+      "loss": 1.0099,
+      "step": 12013
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2190762294105756e-05,
+      "loss": 1.037,
+      "step": 12014
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2189390068204678e-05,
+      "loss": 1.0354,
+      "step": 12015
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2188017838736094e-05,
+      "loss": 0.9771,
+      "step": 12016
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2186645605725843e-05,
+      "loss": 1.0071,
+      "step": 12017
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.218527336919978e-05,
+      "loss": 1.1053,
+      "step": 12018
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2183901129183752e-05,
+      "loss": 1.0627,
+      "step": 12019
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2182528885703605e-05,
+      "loss": 1.0946,
+      "step": 12020
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.218115663878519e-05,
+      "loss": 1.0109,
+      "step": 12021
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2179784388454356e-05,
+      "loss": 1.0434,
+      "step": 12022
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2178412134736951e-05,
+      "loss": 1.0597,
+      "step": 12023
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2177039877658825e-05,
+      "loss": 0.9917,
+      "step": 12024
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2175667617245825e-05,
+      "loss": 1.0482,
+      "step": 12025
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2174295353523804e-05,
+      "loss": 1.0364,
+      "step": 12026
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2172923086518606e-05,
+      "loss": 1.1903,
+      "step": 12027
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2171550816256083e-05,
+      "loss": 1.0686,
+      "step": 12028
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2170178542762083e-05,
+      "loss": 1.0765,
+      "step": 12029
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2168806266062458e-05,
+      "loss": 1.0251,
+      "step": 12030
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2167433986183054e-05,
+      "loss": 1.0505,
+      "step": 12031
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2166061703149726e-05,
+      "loss": 1.0575,
+      "step": 12032
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2164689416988315e-05,
+      "loss": 1.0499,
+      "step": 12033
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2163317127724677e-05,
+      "loss": 1.088,
+      "step": 12034
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2161944835384659e-05,
+      "loss": 0.9924,
+      "step": 12035
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.216057253999411e-05,
+      "loss": 1.1077,
+      "step": 12036
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2159200241578883e-05,
+      "loss": 1.0373,
+      "step": 12037
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2157827940164826e-05,
+      "loss": 1.0991,
+      "step": 12038
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2156455635777789e-05,
+      "loss": 1.1508,
+      "step": 12039
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2155083328443617e-05,
+      "loss": 1.0902,
+      "step": 12040
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.215371101818817e-05,
+      "loss": 1.1079,
+      "step": 12041
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2152338705037287e-05,
+      "loss": 1.0533,
+      "step": 12042
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2150966389016826e-05,
+      "loss": 1.1512,
+      "step": 12043
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2149594070152637e-05,
+      "loss": 1.0491,
+      "step": 12044
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2148221748470565e-05,
+      "loss": 1.0779,
+      "step": 12045
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2146849423996463e-05,
+      "loss": 1.0567,
+      "step": 12046
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2145477096756182e-05,
+      "loss": 1.0895,
+      "step": 12047
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2144104766775573e-05,
+      "loss": 1.1121,
+      "step": 12048
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2142732434080481e-05,
+      "loss": 1.1971,
+      "step": 12049
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2141360098696766e-05,
+      "loss": 1.113,
+      "step": 12050
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2139987760650267e-05,
+      "loss": 1.1274,
+      "step": 12051
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2138615419966843e-05,
+      "loss": 0.9819,
+      "step": 12052
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.213724307667234e-05,
+      "loss": 1.027,
+      "step": 12053
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2135870730792613e-05,
+      "loss": 0.9581,
+      "step": 12054
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2134498382353508e-05,
+      "loss": 0.9948,
+      "step": 12055
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2133126031380882e-05,
+      "loss": 1.0421,
+      "step": 12056
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2131753677900576e-05,
+      "loss": 1.1,
+      "step": 12057
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2130381321938451e-05,
+      "loss": 1.0973,
+      "step": 12058
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2129008963520349e-05,
+      "loss": 1.038,
+      "step": 12059
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2127636602672128e-05,
+      "loss": 1.0008,
+      "step": 12060
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2126264239419635e-05,
+      "loss": 1.0362,
+      "step": 12061
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2124891873788724e-05,
+      "loss": 1.1159,
+      "step": 12062
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2123519505805244e-05,
+      "loss": 1.0434,
+      "step": 12063
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2122147135495043e-05,
+      "loss": 1.0607,
+      "step": 12064
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2120774762883978e-05,
+      "loss": 0.9802,
+      "step": 12065
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2119402387997895e-05,
+      "loss": 1.0633,
+      "step": 12066
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.211803001086265e-05,
+      "loss": 1.0299,
+      "step": 12067
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2116657631504094e-05,
+      "loss": 0.9874,
+      "step": 12068
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2115285249948072e-05,
+      "loss": 1.0565,
+      "step": 12069
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2113912866220443e-05,
+      "loss": 1.0416,
+      "step": 12070
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2112540480347052e-05,
+      "loss": 1.1167,
+      "step": 12071
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2111168092353755e-05,
+      "loss": 1.0663,
+      "step": 12072
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2109795702266402e-05,
+      "loss": 1.0429,
+      "step": 12073
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2108423310110846e-05,
+      "loss": 1.033,
+      "step": 12074
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2107050915912934e-05,
+      "loss": 1.0442,
+      "step": 12075
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2105678519698523e-05,
+      "loss": 1.0424,
+      "step": 12076
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2104306121493461e-05,
+      "loss": 1.0788,
+      "step": 12077
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.21029337213236e-05,
+      "loss": 1.0035,
+      "step": 12078
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2101561319214791e-05,
+      "loss": 1.0334,
+      "step": 12079
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2100188915192891e-05,
+      "loss": 1.0709,
+      "step": 12080
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2098816509283749e-05,
+      "loss": 0.9818,
+      "step": 12081
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.209744410151321e-05,
+      "loss": 1.0767,
+      "step": 12082
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2096071691907137e-05,
+      "loss": 1.094,
+      "step": 12083
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2094699280491371e-05,
+      "loss": 1.0409,
+      "step": 12084
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2093326867291774e-05,
+      "loss": 1.0251,
+      "step": 12085
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2091954452334191e-05,
+      "loss": 1.0655,
+      "step": 12086
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2090582035644478e-05,
+      "loss": 1.0413,
+      "step": 12087
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2089209617248483e-05,
+      "loss": 1.0353,
+      "step": 12088
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2087837197172061e-05,
+      "loss": 1.0991,
+      "step": 12089
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2086464775441064e-05,
+      "loss": 1.0682,
+      "step": 12090
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2085092352081344e-05,
+      "loss": 1.1122,
+      "step": 12091
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2083719927118753e-05,
+      "loss": 0.9664,
+      "step": 12092
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2082347500579141e-05,
+      "loss": 1.0414,
+      "step": 12093
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2080975072488363e-05,
+      "loss": 1.031,
+      "step": 12094
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2079602642872268e-05,
+      "loss": 0.9358,
+      "step": 12095
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2078230211756713e-05,
+      "loss": 1.0698,
+      "step": 12096
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2076857779167545e-05,
+      "loss": 1.0902,
+      "step": 12097
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2075485345130624e-05,
+      "loss": 1.1006,
+      "step": 12098
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.207411290967179e-05,
+      "loss": 1.0615,
+      "step": 12099
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.207274047281691e-05,
+      "loss": 1.0729,
+      "step": 12100
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2071368034591826e-05,
+      "loss": 1.0892,
+      "step": 12101
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2069995595022393e-05,
+      "loss": 1.0914,
+      "step": 12102
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2068623154134461e-05,
+      "loss": 1.1381,
+      "step": 12103
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.206725071195389e-05,
+      "loss": 1.0286,
+      "step": 12104
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2065878268506526e-05,
+      "loss": 0.9915,
+      "step": 12105
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2064505823818222e-05,
+      "loss": 1.0615,
+      "step": 12106
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2063133377914832e-05,
+      "loss": 1.1236,
+      "step": 12107
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.206176093082221e-05,
+      "loss": 1.0982,
+      "step": 12108
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2060388482566208e-05,
+      "loss": 1.1101,
+      "step": 12109
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2059016033172675e-05,
+      "loss": 1.0443,
+      "step": 12110
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2057643582667467e-05,
+      "loss": 0.9849,
+      "step": 12111
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2056271131076435e-05,
+      "loss": 1.0106,
+      "step": 12112
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2054898678425433e-05,
+      "loss": 1.0476,
+      "step": 12113
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2053526224740313e-05,
+      "loss": 1.011,
+      "step": 12114
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2052153770046928e-05,
+      "loss": 0.9959,
+      "step": 12115
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2050781314371131e-05,
+      "loss": 1.0156,
+      "step": 12116
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2049408857738773e-05,
+      "loss": 1.048,
+      "step": 12117
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2048036400175709e-05,
+      "loss": 1.0491,
+      "step": 12118
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.204666394170779e-05,
+      "loss": 1.011,
+      "step": 12119
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.204529148236087e-05,
+      "loss": 1.0768,
+      "step": 12120
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2043919022160803e-05,
+      "loss": 0.993,
+      "step": 12121
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2042546561133438e-05,
+      "loss": 1.0883,
+      "step": 12122
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2041174099304633e-05,
+      "loss": 1.0557,
+      "step": 12123
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2039801636700234e-05,
+      "loss": 1.0831,
+      "step": 12124
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2038429173346102e-05,
+      "loss": 0.9244,
+      "step": 12125
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.203705670926808e-05,
+      "loss": 0.9879,
+      "step": 12126
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2035684244492032e-05,
+      "loss": 1.0642,
+      "step": 12127
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.20343117790438e-05,
+      "loss": 1.0244,
+      "step": 12128
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2032939312949248e-05,
+      "loss": 1.053,
+      "step": 12129
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.203156684623422e-05,
+      "loss": 1.0449,
+      "step": 12130
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.203019437892457e-05,
+      "loss": 1.0501,
+      "step": 12131
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2028821911046158e-05,
+      "loss": 1.0836,
+      "step": 12132
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.202744944262483e-05,
+      "loss": 1.0305,
+      "step": 12133
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2026076973686442e-05,
+      "loss": 0.9564,
+      "step": 12134
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2024704504256843e-05,
+      "loss": 1.0325,
+      "step": 12135
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2023332034361894e-05,
+      "loss": 1.1203,
+      "step": 12136
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.202195956402744e-05,
+      "loss": 1.033,
+      "step": 12137
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2020587093279338e-05,
+      "loss": 1.0426,
+      "step": 12138
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2019214622143438e-05,
+      "loss": 1.0434,
+      "step": 12139
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.20178421506456e-05,
+      "loss": 1.0318,
+      "step": 12140
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2016469678811666e-05,
+      "loss": 1.0412,
+      "step": 12141
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.20150972066675e-05,
+      "loss": 1.0076,
+      "step": 12142
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2013724734238949e-05,
+      "loss": 1.0965,
+      "step": 12143
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2012352261551867e-05,
+      "loss": 1.0493,
+      "step": 12144
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2010979788632106e-05,
+      "loss": 1.0671,
+      "step": 12145
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2009607315505523e-05,
+      "loss": 0.9772,
+      "step": 12146
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2008234842197966e-05,
+      "loss": 1.0456,
+      "step": 12147
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2006862368735292e-05,
+      "loss": 0.9254,
+      "step": 12148
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2005489895143352e-05,
+      "loss": 1.0434,
+      "step": 12149
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2004117421448e-05,
+      "loss": 1.0989,
+      "step": 12150
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.200274494767509e-05,
+      "loss": 1.044,
+      "step": 12151
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2001372473850471e-05,
+      "loss": 1.0263,
+      "step": 12152
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.2e-05,
+      "loss": 1.0486,
+      "step": 12153
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1998627526149531e-05,
+      "loss": 0.9537,
+      "step": 12154
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1997255052324915e-05,
+      "loss": 0.9932,
+      "step": 12155
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1995882578552001e-05,
+      "loss": 1.0524,
+      "step": 12156
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.199451010485665e-05,
+      "loss": 1.0407,
+      "step": 12157
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1993137631264709e-05,
+      "loss": 1.0898,
+      "step": 12158
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1991765157802037e-05,
+      "loss": 1.1404,
+      "step": 12159
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.199039268449448e-05,
+      "loss": 1.0483,
+      "step": 12160
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1989020211367895e-05,
+      "loss": 1.0313,
+      "step": 12161
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1987647738448133e-05,
+      "loss": 1.1205,
+      "step": 12162
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1986275265761055e-05,
+      "loss": 1.0545,
+      "step": 12163
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1984902793332502e-05,
+      "loss": 0.9559,
+      "step": 12164
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1983530321188334e-05,
+      "loss": 1.0475,
+      "step": 12165
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1982157849354405e-05,
+      "loss": 0.9532,
+      "step": 12166
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1980785377856564e-05,
+      "loss": 1.0902,
+      "step": 12167
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1979412906720662e-05,
+      "loss": 1.0245,
+      "step": 12168
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1978040435972561e-05,
+      "loss": 1.0475,
+      "step": 12169
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.197666796563811e-05,
+      "loss": 1.0258,
+      "step": 12170
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.197529549574316e-05,
+      "loss": 1.1323,
+      "step": 12171
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.197392302631356e-05,
+      "loss": 1.0038,
+      "step": 12172
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.197255055737517e-05,
+      "loss": 1.0247,
+      "step": 12173
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1971178088953845e-05,
+      "loss": 1.055,
+      "step": 12174
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.196980562107543e-05,
+      "loss": 1.0753,
+      "step": 12175
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1968433153765781e-05,
+      "loss": 1.0986,
+      "step": 12176
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1967060687050756e-05,
+      "loss": 1.0451,
+      "step": 12177
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1965688220956202e-05,
+      "loss": 1.0508,
+      "step": 12178
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.196431575550797e-05,
+      "loss": 1.1075,
+      "step": 12179
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1962943290731919e-05,
+      "loss": 1.078,
+      "step": 12180
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1961570826653902e-05,
+      "loss": 1.0695,
+      "step": 12181
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1960198363299769e-05,
+      "loss": 1.1024,
+      "step": 12182
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.195882590069537e-05,
+      "loss": 1.0061,
+      "step": 12183
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1957453438866562e-05,
+      "loss": 0.9601,
+      "step": 12184
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.19560809778392e-05,
+      "loss": 0.9721,
+      "step": 12185
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.195470851763913e-05,
+      "loss": 1.0103,
+      "step": 12186
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1953336058292209e-05,
+      "loss": 1.0534,
+      "step": 12187
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1951963599824293e-05,
+      "loss": 1.0519,
+      "step": 12188
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1950591142261229e-05,
+      "loss": 1.0605,
+      "step": 12189
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1949218685628871e-05,
+      "loss": 1.1113,
+      "step": 12190
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1947846229953072e-05,
+      "loss": 1.04,
+      "step": 12191
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.194647377525969e-05,
+      "loss": 1.0994,
+      "step": 12192
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1945101321574569e-05,
+      "loss": 1.0692,
+      "step": 12193
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1943728868923567e-05,
+      "loss": 1.082,
+      "step": 12194
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1942356417332537e-05,
+      "loss": 1.0711,
+      "step": 12195
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1940983966827328e-05,
+      "loss": 1.0696,
+      "step": 12196
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1939611517433796e-05,
+      "loss": 1.0402,
+      "step": 12197
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1938239069177791e-05,
+      "loss": 1.1027,
+      "step": 12198
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.193686662208517e-05,
+      "loss": 1.0093,
+      "step": 12199
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.193549417618178e-05,
+      "loss": 1.0229,
+      "step": 12200
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1934121731493476e-05,
+      "loss": 0.9928,
+      "step": 12201
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1932749288046113e-05,
+      "loss": 1.0206,
+      "step": 12202
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1931376845865541e-05,
+      "loss": 1.049,
+      "step": 12203
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.193000440497761e-05,
+      "loss": 1.015,
+      "step": 12204
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1928631965408178e-05,
+      "loss": 1.0527,
+      "step": 12205
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1927259527183094e-05,
+      "loss": 1.0396,
+      "step": 12206
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.192588709032821e-05,
+      "loss": 1.0722,
+      "step": 12207
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1924514654869378e-05,
+      "loss": 1.085,
+      "step": 12208
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1923142220832454e-05,
+      "loss": 1.0504,
+      "step": 12209
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.192176978824329e-05,
+      "loss": 1.0428,
+      "step": 12210
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1920397357127734e-05,
+      "loss": 1.0705,
+      "step": 12211
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.191902492751164e-05,
+      "loss": 0.9346,
+      "step": 12212
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.191765249942086e-05,
+      "loss": 1.0428,
+      "step": 12213
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.191628007288125e-05,
+      "loss": 1.0791,
+      "step": 12214
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1914907647918658e-05,
+      "loss": 1.0399,
+      "step": 12215
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1913535224558935e-05,
+      "loss": 1.0542,
+      "step": 12216
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.191216280282794e-05,
+      "loss": 0.9861,
+      "step": 12217
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.191079038275152e-05,
+      "loss": 1.0582,
+      "step": 12218
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1909417964355525e-05,
+      "loss": 0.9872,
+      "step": 12219
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1908045547665808e-05,
+      "loss": 1.0171,
+      "step": 12220
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1906673132708228e-05,
+      "loss": 1.0261,
+      "step": 12221
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.190530071950863e-05,
+      "loss": 1.0478,
+      "step": 12222
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1903928308092864e-05,
+      "loss": 1.0627,
+      "step": 12223
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1902555898486789e-05,
+      "loss": 1.0385,
+      "step": 12224
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1901183490716255e-05,
+      "loss": 0.9858,
+      "step": 12225
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1899811084807111e-05,
+      "loss": 1.0586,
+      "step": 12226
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1898438680785206e-05,
+      "loss": 1.002,
+      "step": 12227
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1897066278676405e-05,
+      "loss": 1.1285,
+      "step": 12228
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1895693878506541e-05,
+      "loss": 1.0497,
+      "step": 12229
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.189432148030148e-05,
+      "loss": 1.0143,
+      "step": 12230
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1892949084087065e-05,
+      "loss": 0.9502,
+      "step": 12231
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.189157668988916e-05,
+      "loss": 1.0038,
+      "step": 12232
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.18902042977336e-05,
+      "loss": 1.0472,
+      "step": 12233
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1888831907646246e-05,
+      "loss": 0.9581,
+      "step": 12234
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1887459519652947e-05,
+      "loss": 0.9717,
+      "step": 12235
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.188608713377956e-05,
+      "loss": 1.1029,
+      "step": 12236
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1884714750051928e-05,
+      "loss": 1.0591,
+      "step": 12237
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1883342368495909e-05,
+      "loss": 1.0231,
+      "step": 12238
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1881969989137353e-05,
+      "loss": 1.0097,
+      "step": 12239
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1880597612002105e-05,
+      "loss": 1.0147,
+      "step": 12240
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1879225237116025e-05,
+      "loss": 1.0763,
+      "step": 12241
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1877852864504959e-05,
+      "loss": 1.1343,
+      "step": 12242
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1876480494194762e-05,
+      "loss": 1.0471,
+      "step": 12243
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1875108126211278e-05,
+      "loss": 0.9969,
+      "step": 12244
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1873735760580365e-05,
+      "loss": 1.0786,
+      "step": 12245
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1872363397327872e-05,
+      "loss": 1.0736,
+      "step": 12246
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1870991036479653e-05,
+      "loss": 1.1062,
+      "step": 12247
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1869618678061551e-05,
+      "loss": 1.0619,
+      "step": 12248
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1868246322099425e-05,
+      "loss": 0.984,
+      "step": 12249
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1866873968619122e-05,
+      "loss": 1.0617,
+      "step": 12250
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1865501617646492e-05,
+      "loss": 1.0574,
+      "step": 12251
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1864129269207387e-05,
+      "loss": 1.0162,
+      "step": 12252
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.186275692332766e-05,
+      "loss": 1.004,
+      "step": 12253
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1861384580033159e-05,
+      "loss": 1.0448,
+      "step": 12254
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1860012239349735e-05,
+      "loss": 0.9754,
+      "step": 12255
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1858639901303236e-05,
+      "loss": 1.0508,
+      "step": 12256
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1857267565919518e-05,
+      "loss": 1.0582,
+      "step": 12257
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1855895233224431e-05,
+      "loss": 1.0145,
+      "step": 12258
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1854522903243819e-05,
+      "loss": 0.9769,
+      "step": 12259
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1853150576003536e-05,
+      "loss": 1.0495,
+      "step": 12260
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1851778251529436e-05,
+      "loss": 0.96,
+      "step": 12261
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1850405929847367e-05,
+      "loss": 1.0119,
+      "step": 12262
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1849033610983174e-05,
+      "loss": 1.0618,
+      "step": 12263
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1847661294962712e-05,
+      "loss": 1.0621,
+      "step": 12264
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1846288981811835e-05,
+      "loss": 1.1366,
+      "step": 12265
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1844916671556384e-05,
+      "loss": 1.1065,
+      "step": 12266
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1843544364222214e-05,
+      "loss": 0.9952,
+      "step": 12267
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1842172059835175e-05,
+      "loss": 1.0411,
+      "step": 12268
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.184079975842112e-05,
+      "loss": 0.9864,
+      "step": 12269
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1839427460005892e-05,
+      "loss": 1.0193,
+      "step": 12270
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1838055164615342e-05,
+      "loss": 1.0591,
+      "step": 12271
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1836682872275327e-05,
+      "loss": 0.9937,
+      "step": 12272
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1835310583011687e-05,
+      "loss": 1.0391,
+      "step": 12273
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1833938296850278e-05,
+      "loss": 1.108,
+      "step": 12274
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1832566013816943e-05,
+      "loss": 1.0929,
+      "step": 12275
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1831193733937546e-05,
+      "loss": 1.0979,
+      "step": 12276
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1829821457237918e-05,
+      "loss": 1.0206,
+      "step": 12277
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1828449183743919e-05,
+      "loss": 1.107,
+      "step": 12278
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.18270769134814e-05,
+      "loss": 1.0742,
+      "step": 12279
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.18257046464762e-05,
+      "loss": 0.9428,
+      "step": 12280
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1824332382754176e-05,
+      "loss": 1.0088,
+      "step": 12281
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1822960122341177e-05,
+      "loss": 1.1043,
+      "step": 12282
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1821587865263053e-05,
+      "loss": 1.0959,
+      "step": 12283
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1820215611545645e-05,
+      "loss": 1.0568,
+      "step": 12284
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.181884336121481e-05,
+      "loss": 1.0449,
+      "step": 12285
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1817471114296397e-05,
+      "loss": 1.0859,
+      "step": 12286
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1816098870816253e-05,
+      "loss": 0.9951,
+      "step": 12287
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.181472663080022e-05,
+      "loss": 1.0417,
+      "step": 12288
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1813354394274157e-05,
+      "loss": 0.9779,
+      "step": 12289
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.181198216126391e-05,
+      "loss": 1.0213,
+      "step": 12290
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1810609931795324e-05,
+      "loss": 1.0662,
+      "step": 12291
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1809237705894246e-05,
+      "loss": 1.0659,
+      "step": 12292
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.180786548358653e-05,
+      "loss": 1.0976,
+      "step": 12293
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1806493264898025e-05,
+      "loss": 1.0605,
+      "step": 12294
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1805121049854574e-05,
+      "loss": 1.0114,
+      "step": 12295
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1803748838482026e-05,
+      "loss": 0.9636,
+      "step": 12296
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.180237663080623e-05,
+      "loss": 1.0635,
+      "step": 12297
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.180100442685304e-05,
+      "loss": 1.0894,
+      "step": 12298
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1799632226648296e-05,
+      "loss": 1.1105,
+      "step": 12299
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1798260030217845e-05,
+      "loss": 1.0215,
+      "step": 12300
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1796887837587546e-05,
+      "loss": 0.9789,
+      "step": 12301
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1795515648783234e-05,
+      "loss": 1.0154,
+      "step": 12302
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1794143463830763e-05,
+      "loss": 1.0051,
+      "step": 12303
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1792771282755978e-05,
+      "loss": 1.0682,
+      "step": 12304
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1791399105584731e-05,
+      "loss": 1.0942,
+      "step": 12305
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1790026932342867e-05,
+      "loss": 1.0555,
+      "step": 12306
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1788654763056231e-05,
+      "loss": 1.0267,
+      "step": 12307
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.178728259775067e-05,
+      "loss": 1.053,
+      "step": 12308
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.178591043645204e-05,
+      "loss": 1.0158,
+      "step": 12309
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1784538279186177e-05,
+      "loss": 1.0923,
+      "step": 12310
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1783166125978934e-05,
+      "loss": 1.0514,
+      "step": 12311
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1781793976856161e-05,
+      "loss": 1.0169,
+      "step": 12312
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1780421831843696e-05,
+      "loss": 1.029,
+      "step": 12313
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1779049690967392e-05,
+      "loss": 0.9924,
+      "step": 12314
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1777677554253093e-05,
+      "loss": 1.1028,
+      "step": 12315
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1776305421726654e-05,
+      "loss": 1.0841,
+      "step": 12316
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1774933293413909e-05,
+      "loss": 1.017,
+      "step": 12317
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.177356116934071e-05,
+      "loss": 1.021,
+      "step": 12318
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1772189049532905e-05,
+      "loss": 1.0136,
+      "step": 12319
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1770816934016346e-05,
+      "loss": 1.0017,
+      "step": 12320
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1769444822816865e-05,
+      "loss": 1.0306,
+      "step": 12321
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1768072715960318e-05,
+      "loss": 1.0327,
+      "step": 12322
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1766700613472552e-05,
+      "loss": 1.0322,
+      "step": 12323
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1765328515379406e-05,
+      "loss": 1.0463,
+      "step": 12324
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1763956421706733e-05,
+      "loss": 1.01,
+      "step": 12325
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1762584332480377e-05,
+      "loss": 1.0017,
+      "step": 12326
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1761212247726183e-05,
+      "loss": 1.0143,
+      "step": 12327
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1759840167469994e-05,
+      "loss": 0.9882,
+      "step": 12328
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.175846809173766e-05,
+      "loss": 1.0093,
+      "step": 12329
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1757096020555025e-05,
+      "loss": 1.0326,
+      "step": 12330
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1755723953947937e-05,
+      "loss": 1.0162,
+      "step": 12331
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1754351891942236e-05,
+      "loss": 0.9943,
+      "step": 12332
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1752979834563772e-05,
+      "loss": 1.1428,
+      "step": 12333
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1751607781838389e-05,
+      "loss": 1.0349,
+      "step": 12334
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.175023573379193e-05,
+      "loss": 1.0273,
+      "step": 12335
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.174886369045024e-05,
+      "loss": 1.02,
+      "step": 12336
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1747491651839167e-05,
+      "loss": 1.0507,
+      "step": 12337
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1746119617984558e-05,
+      "loss": 1.1538,
+      "step": 12338
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1744747588912252e-05,
+      "loss": 1.0274,
+      "step": 12339
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1743375564648092e-05,
+      "loss": 0.9666,
+      "step": 12340
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.174200354521793e-05,
+      "loss": 1.0178,
+      "step": 12341
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.174063153064761e-05,
+      "loss": 1.0047,
+      "step": 12342
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1739259520962972e-05,
+      "loss": 0.9792,
+      "step": 12343
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.173788751618986e-05,
+      "loss": 1.087,
+      "step": 12344
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1736515516354123e-05,
+      "loss": 0.9485,
+      "step": 12345
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.17351435214816e-05,
+      "loss": 1.0914,
+      "step": 12346
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1733771531598139e-05,
+      "loss": 1.0192,
+      "step": 12347
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1732399546729578e-05,
+      "loss": 0.9801,
+      "step": 12348
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1731027566901771e-05,
+      "loss": 1.0696,
+      "step": 12349
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1729655592140554e-05,
+      "loss": 1.0583,
+      "step": 12350
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1728283622471771e-05,
+      "loss": 1.0685,
+      "step": 12351
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1726911657921267e-05,
+      "loss": 1.0291,
+      "step": 12352
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.172553969851489e-05,
+      "loss": 1.0392,
+      "step": 12353
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1724167744278474e-05,
+      "loss": 0.957,
+      "step": 12354
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1722795795237869e-05,
+      "loss": 1.0541,
+      "step": 12355
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1721423851418922e-05,
+      "loss": 1.0247,
+      "step": 12356
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1720051912847462e-05,
+      "loss": 1.0729,
+      "step": 12357
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1718679979549345e-05,
+      "loss": 0.9598,
+      "step": 12358
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1717308051550407e-05,
+      "loss": 1.0905,
+      "step": 12359
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1715936128876501e-05,
+      "loss": 1.0382,
+      "step": 12360
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1714564211553453e-05,
+      "loss": 0.9983,
+      "step": 12361
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1713192299607119e-05,
+      "loss": 1.0052,
+      "step": 12362
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1711820393063337e-05,
+      "loss": 1.0387,
+      "step": 12363
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.171044849194795e-05,
+      "loss": 0.9757,
+      "step": 12364
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1709076596286796e-05,
+      "loss": 1.0472,
+      "step": 12365
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1707704706105724e-05,
+      "loss": 1.0686,
+      "step": 12366
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1706332821430574e-05,
+      "loss": 0.9955,
+      "step": 12367
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1704960942287184e-05,
+      "loss": 1.0021,
+      "step": 12368
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.17035890687014e-05,
+      "loss": 1.0066,
+      "step": 12369
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1702217200699064e-05,
+      "loss": 1.0053,
+      "step": 12370
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1700845338306018e-05,
+      "loss": 0.9987,
+      "step": 12371
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1699473481548097e-05,
+      "loss": 0.9798,
+      "step": 12372
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1698101630451151e-05,
+      "loss": 1.0416,
+      "step": 12373
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1696729785041019e-05,
+      "loss": 1.0932,
+      "step": 12374
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1695357945343539e-05,
+      "loss": 0.9848,
+      "step": 12375
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1693986111384552e-05,
+      "loss": 0.9959,
+      "step": 12376
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1692614283189901e-05,
+      "loss": 1.0055,
+      "step": 12377
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1691242460785433e-05,
+      "loss": 1.0031,
+      "step": 12378
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1689870644196977e-05,
+      "loss": 1.0341,
+      "step": 12379
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.168849883345038e-05,
+      "loss": 1.0942,
+      "step": 12380
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1687127028571483e-05,
+      "loss": 1.0156,
+      "step": 12381
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1685755229586128e-05,
+      "loss": 1.0638,
+      "step": 12382
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1684383436520152e-05,
+      "loss": 1.0835,
+      "step": 12383
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1683011649399393e-05,
+      "loss": 0.9993,
+      "step": 12384
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1681639868249698e-05,
+      "loss": 1.0069,
+      "step": 12385
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1680268093096903e-05,
+      "loss": 1.0664,
+      "step": 12386
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1678896323966848e-05,
+      "loss": 1.0438,
+      "step": 12387
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.167752456088537e-05,
+      "loss": 1.0312,
+      "step": 12388
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1676152803878317e-05,
+      "loss": 1.0711,
+      "step": 12389
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.167478105297152e-05,
+      "loss": 1.0608,
+      "step": 12390
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1673409308190822e-05,
+      "loss": 0.95,
+      "step": 12391
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1672037569562061e-05,
+      "loss": 1.0256,
+      "step": 12392
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1670665837111082e-05,
+      "loss": 1.066,
+      "step": 12393
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1669294110863716e-05,
+      "loss": 1.0974,
+      "step": 12394
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1667922390845805e-05,
+      "loss": 0.9976,
+      "step": 12395
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1666550677083192e-05,
+      "loss": 1.0757,
+      "step": 12396
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1665178969601708e-05,
+      "loss": 0.9908,
+      "step": 12397
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1663807268427197e-05,
+      "loss": 1.0436,
+      "step": 12398
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1662435573585493e-05,
+      "loss": 1.0407,
+      "step": 12399
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1661063885102446e-05,
+      "loss": 1.0267,
+      "step": 12400
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.165969220300388e-05,
+      "loss": 1.0831,
+      "step": 12401
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1658320527315641e-05,
+      "loss": 1.0169,
+      "step": 12402
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1656948858063562e-05,
+      "loss": 1.0642,
+      "step": 12403
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1655577195273491e-05,
+      "loss": 1.1733,
+      "step": 12404
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1654205538971253e-05,
+      "loss": 1.0589,
+      "step": 12405
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1652833889182694e-05,
+      "loss": 1.0479,
+      "step": 12406
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.165146224593365e-05,
+      "loss": 0.9549,
+      "step": 12407
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1650090609249957e-05,
+      "loss": 1.0284,
+      "step": 12408
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.164871897915745e-05,
+      "loss": 1.1118,
+      "step": 12409
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.164734735568197e-05,
+      "loss": 1.1033,
+      "step": 12410
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1645975738849357e-05,
+      "loss": 1.0553,
+      "step": 12411
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1644604128685438e-05,
+      "loss": 1.0392,
+      "step": 12412
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1643232525216059e-05,
+      "loss": 1.0503,
+      "step": 12413
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1641860928467054e-05,
+      "loss": 1.0911,
+      "step": 12414
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1640489338464262e-05,
+      "loss": 0.989,
+      "step": 12415
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.163911775523351e-05,
+      "loss": 1.0097,
+      "step": 12416
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1637746178800643e-05,
+      "loss": 1.1084,
+      "step": 12417
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.16363746091915e-05,
+      "loss": 1.1041,
+      "step": 12418
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1635003046431908e-05,
+      "loss": 1.1192,
+      "step": 12419
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1633631490547705e-05,
+      "loss": 1.0113,
+      "step": 12420
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1632259941564731e-05,
+      "loss": 1.0562,
+      "step": 12421
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.163088839950882e-05,
+      "loss": 1.0284,
+      "step": 12422
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1629516864405808e-05,
+      "loss": 1.0052,
+      "step": 12423
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1628145336281524e-05,
+      "loss": 1.0219,
+      "step": 12424
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1626773815161813e-05,
+      "loss": 1.0465,
+      "step": 12425
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1625402301072507e-05,
+      "loss": 1.0371,
+      "step": 12426
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1624030794039437e-05,
+      "loss": 0.9823,
+      "step": 12427
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.162265929408844e-05,
+      "loss": 1.092,
+      "step": 12428
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1621287801245353e-05,
+      "loss": 1.074,
+      "step": 12429
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.161991631553601e-05,
+      "loss": 1.0199,
+      "step": 12430
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1618544836986241e-05,
+      "loss": 1.0773,
+      "step": 12431
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1617173365621884e-05,
+      "loss": 1.042,
+      "step": 12432
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1615801901468776e-05,
+      "loss": 1.0286,
+      "step": 12433
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1614430444552746e-05,
+      "loss": 1.0337,
+      "step": 12434
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.161305899489963e-05,
+      "loss": 1.0394,
+      "step": 12435
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.161168755253526e-05,
+      "loss": 1.0862,
+      "step": 12436
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1610316117485476e-05,
+      "loss": 1.0317,
+      "step": 12437
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1608944689776104e-05,
+      "loss": 1.0742,
+      "step": 12438
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.160757326943298e-05,
+      "loss": 1.0685,
+      "step": 12439
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1606201856481939e-05,
+      "loss": 1.0024,
+      "step": 12440
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1604830450948808e-05,
+      "loss": 1.0886,
+      "step": 12441
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1603459052859428e-05,
+      "loss": 1.0755,
+      "step": 12442
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1602087662239626e-05,
+      "loss": 1.0474,
+      "step": 12443
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1600716279115243e-05,
+      "loss": 0.9599,
+      "step": 12444
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.15993449035121e-05,
+      "loss": 1.0594,
+      "step": 12445
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1597973535456036e-05,
+      "loss": 1.0316,
+      "step": 12446
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1596602174972882e-05,
+      "loss": 1.0555,
+      "step": 12447
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1595230822088471e-05,
+      "loss": 1.065,
+      "step": 12448
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1593859476828631e-05,
+      "loss": 1.0413,
+      "step": 12449
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.15924881392192e-05,
+      "loss": 1.1091,
+      "step": 12450
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1591116809286008e-05,
+      "loss": 1.0261,
+      "step": 12451
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1589745487054883e-05,
+      "loss": 1.0613,
+      "step": 12452
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1588374172551658e-05,
+      "loss": 1.0494,
+      "step": 12453
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1587002865802166e-05,
+      "loss": 1.0716,
+      "step": 12454
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1585631566832239e-05,
+      "loss": 1.0561,
+      "step": 12455
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1584260275667702e-05,
+      "loss": 1.0622,
+      "step": 12456
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.158288899233439e-05,
+      "loss": 0.9629,
+      "step": 12457
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1581517716858136e-05,
+      "loss": 1.0209,
+      "step": 12458
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1580146449264768e-05,
+      "loss": 1.1117,
+      "step": 12459
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1578775189580112e-05,
+      "loss": 1.105,
+      "step": 12460
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1577403937830006e-05,
+      "loss": 1.0305,
+      "step": 12461
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1576032694040277e-05,
+      "loss": 1.101,
+      "step": 12462
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1574661458236752e-05,
+      "loss": 1.1277,
+      "step": 12463
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1573290230445262e-05,
+      "loss": 1.0385,
+      "step": 12464
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1571919010691638e-05,
+      "loss": 1.1024,
+      "step": 12465
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1570547799001714e-05,
+      "loss": 0.9775,
+      "step": 12466
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.156917659540131e-05,
+      "loss": 1.0548,
+      "step": 12467
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.156780539991626e-05,
+      "loss": 1.0519,
+      "step": 12468
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1566434212572394e-05,
+      "loss": 1.0294,
+      "step": 12469
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.156506303339554e-05,
+      "loss": 1.0473,
+      "step": 12470
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1563691862411524e-05,
+      "loss": 1.0637,
+      "step": 12471
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1562320699646176e-05,
+      "loss": 1.0245,
+      "step": 12472
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.156094954512533e-05,
+      "loss": 1.0591,
+      "step": 12473
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1559578398874805e-05,
+      "loss": 1.0597,
+      "step": 12474
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1558207260920434e-05,
+      "loss": 1.0999,
+      "step": 12475
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1556836131288043e-05,
+      "loss": 1.0176,
+      "step": 12476
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1555465010003463e-05,
+      "loss": 1.0531,
+      "step": 12477
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1554093897092519e-05,
+      "loss": 0.9637,
+      "step": 12478
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.155272279258104e-05,
+      "loss": 1.0083,
+      "step": 12479
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1551351696494854e-05,
+      "loss": 1.107,
+      "step": 12480
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1549980608859781e-05,
+      "loss": 1.0059,
+      "step": 12481
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1548609529701657e-05,
+      "loss": 0.933,
+      "step": 12482
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1547238459046305e-05,
+      "loss": 1.0397,
+      "step": 12483
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1545867396919553e-05,
+      "loss": 1.0984,
+      "step": 12484
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1544496343347223e-05,
+      "loss": 1.0924,
+      "step": 12485
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1543125298355146e-05,
+      "loss": 1.0266,
+      "step": 12486
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1541754261969145e-05,
+      "loss": 1.054,
+      "step": 12487
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1540383234215056e-05,
+      "loss": 1.0215,
+      "step": 12488
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1539012215118688e-05,
+      "loss": 1.0513,
+      "step": 12489
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1537641204705878e-05,
+      "loss": 1.0137,
+      "step": 12490
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1536270203002452e-05,
+      "loss": 1.0772,
+      "step": 12491
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.153489921003423e-05,
+      "loss": 0.9972,
+      "step": 12492
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1533528225827034e-05,
+      "loss": 1.0275,
+      "step": 12493
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.15321572504067e-05,
+      "loss": 0.9821,
+      "step": 12494
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.153078628379905e-05,
+      "loss": 1.0717,
+      "step": 12495
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1529415326029901e-05,
+      "loss": 0.957,
+      "step": 12496
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1528044377125081e-05,
+      "loss": 1.0262,
+      "step": 12497
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.152667343711042e-05,
+      "loss": 1.0919,
+      "step": 12498
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1525302506011738e-05,
+      "loss": 1.0232,
+      "step": 12499
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1523931583854855e-05,
+      "loss": 1.0742,
+      "step": 12500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1522560670665601e-05,
+      "loss": 0.992,
+      "step": 12501
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1521189766469801e-05,
+      "loss": 1.0686,
+      "step": 12502
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1519818871293272e-05,
+      "loss": 1.0612,
+      "step": 12503
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1518447985161839e-05,
+      "loss": 1.1417,
+      "step": 12504
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1517077108101327e-05,
+      "loss": 1.1412,
+      "step": 12505
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1515706240137563e-05,
+      "loss": 0.9998,
+      "step": 12506
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1514335381296362e-05,
+      "loss": 1.0219,
+      "step": 12507
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1512964531603547e-05,
+      "loss": 0.9269,
+      "step": 12508
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1511593691084947e-05,
+      "loss": 1.074,
+      "step": 12509
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1510222859766382e-05,
+      "loss": 1.0215,
+      "step": 12510
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1508852037673672e-05,
+      "loss": 1.0881,
+      "step": 12511
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1507481224832637e-05,
+      "loss": 1.0894,
+      "step": 12512
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1506110421269106e-05,
+      "loss": 1.0251,
+      "step": 12513
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1504739627008894e-05,
+      "loss": 1.0551,
+      "step": 12514
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1503368842077825e-05,
+      "loss": 1.0288,
+      "step": 12515
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1501998066501718e-05,
+      "loss": 1.0557,
+      "step": 12516
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1500627300306402e-05,
+      "loss": 1.0507,
+      "step": 12517
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1499256543517688e-05,
+      "loss": 1.0747,
+      "step": 12518
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1497885796161401e-05,
+      "loss": 1.0482,
+      "step": 12519
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1496515058263358e-05,
+      "loss": 0.9807,
+      "step": 12520
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1495144329849388e-05,
+      "loss": 0.9999,
+      "step": 12521
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1493773610945305e-05,
+      "loss": 1.0501,
+      "step": 12522
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1492402901576928e-05,
+      "loss": 1.0816,
+      "step": 12523
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1491032201770082e-05,
+      "loss": 1.0676,
+      "step": 12524
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1489661511550581e-05,
+      "loss": 1.0144,
+      "step": 12525
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1488290830944247e-05,
+      "loss": 1.0413,
+      "step": 12526
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1486920159976898e-05,
+      "loss": 1.0215,
+      "step": 12527
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1485549498674357e-05,
+      "loss": 1.0449,
+      "step": 12528
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1484178847062437e-05,
+      "loss": 1.0043,
+      "step": 12529
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1482808205166962e-05,
+      "loss": 1.0522,
+      "step": 12530
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.148143757301375e-05,
+      "loss": 1.0215,
+      "step": 12531
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1480066950628618e-05,
+      "loss": 0.9748,
+      "step": 12532
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.147869633803738e-05,
+      "loss": 1.0564,
+      "step": 12533
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.147732573526586e-05,
+      "loss": 1.0604,
+      "step": 12534
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1475955142339877e-05,
+      "loss": 1.0799,
+      "step": 12535
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1474584559285243e-05,
+      "loss": 1.0739,
+      "step": 12536
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1473213986127776e-05,
+      "loss": 1.0193,
+      "step": 12537
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1471843422893298e-05,
+      "loss": 1.019,
+      "step": 12538
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1470472869607625e-05,
+      "loss": 1.0674,
+      "step": 12539
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.146910232629657e-05,
+      "loss": 1.1452,
+      "step": 12540
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1467731792985952e-05,
+      "loss": 1.0465,
+      "step": 12541
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.146636126970159e-05,
+      "loss": 1.038,
+      "step": 12542
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1464990756469296e-05,
+      "loss": 1.1068,
+      "step": 12543
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1463620253314886e-05,
+      "loss": 1.1018,
+      "step": 12544
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.146224976026418e-05,
+      "loss": 0.9896,
+      "step": 12545
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1460879277342995e-05,
+      "loss": 1.0992,
+      "step": 12546
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.145950880457714e-05,
+      "loss": 1.0642,
+      "step": 12547
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1458138341992433e-05,
+      "loss": 1.0713,
+      "step": 12548
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1456767889614691e-05,
+      "loss": 1.0493,
+      "step": 12549
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.145539744746973e-05,
+      "loss": 1.0724,
+      "step": 12550
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1454027015583361e-05,
+      "loss": 1.0825,
+      "step": 12551
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1452656593981398e-05,
+      "loss": 1.0731,
+      "step": 12552
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1451286182689665e-05,
+      "loss": 0.9383,
+      "step": 12553
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.144991578173396e-05,
+      "loss": 1.1305,
+      "step": 12554
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1448545391140112e-05,
+      "loss": 1.0336,
+      "step": 12555
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1447175010933924e-05,
+      "loss": 1.0039,
+      "step": 12556
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.144580464114122e-05,
+      "loss": 1.1015,
+      "step": 12557
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1444434281787805e-05,
+      "loss": 1.0348,
+      "step": 12558
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1443063932899495e-05,
+      "loss": 1.0027,
+      "step": 12559
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1441693594502104e-05,
+      "loss": 1.008,
+      "step": 12560
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1440323266621446e-05,
+      "loss": 1.0504,
+      "step": 12561
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.143895294928333e-05,
+      "loss": 1.1084,
+      "step": 12562
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1437582642513569e-05,
+      "loss": 1.0912,
+      "step": 12563
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1436212346337981e-05,
+      "loss": 1.0464,
+      "step": 12564
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1434842060782367e-05,
+      "loss": 1.1366,
+      "step": 12565
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1433471785872548e-05,
+      "loss": 0.9929,
+      "step": 12566
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1432101521634334e-05,
+      "loss": 1.0209,
+      "step": 12567
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1430731268093536e-05,
+      "loss": 1.0941,
+      "step": 12568
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1429361025275962e-05,
+      "loss": 0.933,
+      "step": 12569
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1427990793207428e-05,
+      "loss": 1.0245,
+      "step": 12570
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1426620571913742e-05,
+      "loss": 1.0145,
+      "step": 12571
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1425250361420717e-05,
+      "loss": 1.0255,
+      "step": 12572
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1423880161754158e-05,
+      "loss": 1.0322,
+      "step": 12573
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1422509972939881e-05,
+      "loss": 1.0648,
+      "step": 12574
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1421139795003698e-05,
+      "loss": 1.0439,
+      "step": 12575
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.141976962797141e-05,
+      "loss": 1.024,
+      "step": 12576
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.141839947186883e-05,
+      "loss": 1.0412,
+      "step": 12577
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.141702932672177e-05,
+      "loss": 1.0437,
+      "step": 12578
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1415659192556043e-05,
+      "loss": 1.0486,
+      "step": 12579
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.141428906939745e-05,
+      "loss": 1.0699,
+      "step": 12580
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.14129189572718e-05,
+      "loss": 1.1539,
+      "step": 12581
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1411548856204907e-05,
+      "loss": 0.9573,
+      "step": 12582
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1410178766222578e-05,
+      "loss": 1.069,
+      "step": 12583
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.140880868735062e-05,
+      "loss": 1.056,
+      "step": 12584
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1407438619614837e-05,
+      "loss": 1.0525,
+      "step": 12585
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1406068563041047e-05,
+      "loss": 0.9506,
+      "step": 12586
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1404698517655045e-05,
+      "loss": 1.0098,
+      "step": 12587
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1403328483482647e-05,
+      "loss": 1.0601,
+      "step": 12588
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1401958460549657e-05,
+      "loss": 1.0278,
+      "step": 12589
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1400588448881888e-05,
+      "loss": 1.0334,
+      "step": 12590
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1399218448505137e-05,
+      "loss": 1.0241,
+      "step": 12591
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1397848459445212e-05,
+      "loss": 1.0724,
+      "step": 12592
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1396478481727925e-05,
+      "loss": 1.0722,
+      "step": 12593
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1395108515379081e-05,
+      "loss": 1.0141,
+      "step": 12594
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1393738560424484e-05,
+      "loss": 1.0457,
+      "step": 12595
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1392368616889935e-05,
+      "loss": 1.043,
+      "step": 12596
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1390998684801251e-05,
+      "loss": 1.0864,
+      "step": 12597
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1389628764184225e-05,
+      "loss": 1.0801,
+      "step": 12598
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1388258855064667e-05,
+      "loss": 1.091,
+      "step": 12599
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1386888957468383e-05,
+      "loss": 1.0909,
+      "step": 12600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1385519071421178e-05,
+      "loss": 1.0895,
+      "step": 12601
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1384149196948854e-05,
+      "loss": 1.0658,
+      "step": 12602
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1382779334077216e-05,
+      "loss": 1.0905,
+      "step": 12603
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1381409482832067e-05,
+      "loss": 0.9924,
+      "step": 12604
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1380039643239213e-05,
+      "loss": 1.1503,
+      "step": 12605
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1378669815324455e-05,
+      "loss": 1.0661,
+      "step": 12606
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1377299999113597e-05,
+      "loss": 0.9134,
+      "step": 12607
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1375930194632446e-05,
+      "loss": 1.0402,
+      "step": 12608
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1374560401906796e-05,
+      "loss": 1.0413,
+      "step": 12609
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1373190620962458e-05,
+      "loss": 1.1348,
+      "step": 12610
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1371820851825227e-05,
+      "loss": 0.9587,
+      "step": 12611
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1370451094520915e-05,
+      "loss": 1.0364,
+      "step": 12612
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1369081349075312e-05,
+      "loss": 1.0672,
+      "step": 12613
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.136771161551423e-05,
+      "loss": 1.1482,
+      "step": 12614
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1366341893863466e-05,
+      "loss": 0.9995,
+      "step": 12615
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.136497218414882e-05,
+      "loss": 1.0348,
+      "step": 12616
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1363602486396093e-05,
+      "loss": 1.0554,
+      "step": 12617
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1362232800631087e-05,
+      "loss": 1.0559,
+      "step": 12618
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1360863126879607e-05,
+      "loss": 1.1094,
+      "step": 12619
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1359493465167446e-05,
+      "loss": 1.0755,
+      "step": 12620
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1358123815520406e-05,
+      "loss": 1.0486,
+      "step": 12621
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1356754177964289e-05,
+      "loss": 0.9893,
+      "step": 12622
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1355384552524897e-05,
+      "loss": 1.0401,
+      "step": 12623
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1354014939228023e-05,
+      "loss": 1.0639,
+      "step": 12624
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1352645338099467e-05,
+      "loss": 1.0274,
+      "step": 12625
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1351275749165037e-05,
+      "loss": 0.9871,
+      "step": 12626
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.134990617245052e-05,
+      "loss": 1.0189,
+      "step": 12627
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1348536607981722e-05,
+      "loss": 1.0509,
+      "step": 12628
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1347167055784435e-05,
+      "loss": 1.0699,
+      "step": 12629
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1345797515884465e-05,
+      "loss": 1.0496,
+      "step": 12630
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1344427988307604e-05,
+      "loss": 1.0341,
+      "step": 12631
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.134305847307965e-05,
+      "loss": 1.075,
+      "step": 12632
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1341688970226402e-05,
+      "loss": 1.0295,
+      "step": 12633
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.134031947977366e-05,
+      "loss": 1.1003,
+      "step": 12634
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1338950001747215e-05,
+      "loss": 0.9688,
+      "step": 12635
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1337580536172864e-05,
+      "loss": 1.1016,
+      "step": 12636
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1336211083076412e-05,
+      "loss": 0.9931,
+      "step": 12637
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1334841642483643e-05,
+      "loss": 1.0351,
+      "step": 12638
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.133347221442036e-05,
+      "loss": 1.0631,
+      "step": 12639
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1332102798912357e-05,
+      "loss": 1.0281,
+      "step": 12640
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1330733395985435e-05,
+      "loss": 1.0249,
+      "step": 12641
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1329364005665377e-05,
+      "loss": 1.1218,
+      "step": 12642
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1327994627977988e-05,
+      "loss": 1.094,
+      "step": 12643
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1326625262949059e-05,
+      "loss": 1.0202,
+      "step": 12644
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1325255910604388e-05,
+      "loss": 1.0604,
+      "step": 12645
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1323886570969763e-05,
+      "loss": 1.0462,
+      "step": 12646
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1322517244070985e-05,
+      "loss": 1.0378,
+      "step": 12647
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1321147929933847e-05,
+      "loss": 1.0866,
+      "step": 12648
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1319778628584132e-05,
+      "loss": 1.0894,
+      "step": 12649
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1318409340047646e-05,
+      "loss": 1.0798,
+      "step": 12650
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1317040064350179e-05,
+      "loss": 1.0756,
+      "step": 12651
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1315670801517522e-05,
+      "loss": 0.9985,
+      "step": 12652
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1314301551575464e-05,
+      "loss": 1.0375,
+      "step": 12653
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1312932314549804e-05,
+      "loss": 1.0881,
+      "step": 12654
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1311563090466331e-05,
+      "loss": 0.9889,
+      "step": 12655
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1310193879350839e-05,
+      "loss": 1.0408,
+      "step": 12656
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1308824681229114e-05,
+      "loss": 1.0517,
+      "step": 12657
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1307455496126953e-05,
+      "loss": 1.0443,
+      "step": 12658
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1306086324070149e-05,
+      "loss": 1.0547,
+      "step": 12659
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1304717165084485e-05,
+      "loss": 1.0565,
+      "step": 12660
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1303348019195754e-05,
+      "loss": 0.9523,
+      "step": 12661
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1301978886429752e-05,
+      "loss": 1.0567,
+      "step": 12662
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1300609766812266e-05,
+      "loss": 1.0154,
+      "step": 12663
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1299240660369084e-05,
+      "loss": 1.1122,
+      "step": 12664
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1297871567125993e-05,
+      "loss": 1.0266,
+      "step": 12665
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1296502487108792e-05,
+      "loss": 1.0312,
+      "step": 12666
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1295133420343263e-05,
+      "loss": 0.9871,
+      "step": 12667
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1293764366855196e-05,
+      "loss": 1.0013,
+      "step": 12668
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1292395326670377e-05,
+      "loss": 1.0642,
+      "step": 12669
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1291026299814602e-05,
+      "loss": 1.1487,
+      "step": 12670
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1289657286313654e-05,
+      "loss": 0.9598,
+      "step": 12671
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.128828828619332e-05,
+      "loss": 1.0437,
+      "step": 12672
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1286919299479387e-05,
+      "loss": 1.0823,
+      "step": 12673
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1285550326197649e-05,
+      "loss": 1.0191,
+      "step": 12674
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1284181366373885e-05,
+      "loss": 1.0721,
+      "step": 12675
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1282812420033885e-05,
+      "loss": 0.9025,
+      "step": 12676
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1281443487203439e-05,
+      "loss": 1.1136,
+      "step": 12677
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.128007456790833e-05,
+      "loss": 1.1313,
+      "step": 12678
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1278705662174345e-05,
+      "loss": 1.0222,
+      "step": 12679
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1277336770027266e-05,
+      "loss": 1.0714,
+      "step": 12680
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.127596789149289e-05,
+      "loss": 1.0896,
+      "step": 12681
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1274599026596986e-05,
+      "loss": 1.0887,
+      "step": 12682
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.127323017536535e-05,
+      "loss": 1.0892,
+      "step": 12683
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1271861337823763e-05,
+      "loss": 1.0671,
+      "step": 12684
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1270492513998017e-05,
+      "loss": 1.0118,
+      "step": 12685
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1269123703913885e-05,
+      "loss": 1.027,
+      "step": 12686
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1267754907597156e-05,
+      "loss": 1.1008,
+      "step": 12687
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1266386125073613e-05,
+      "loss": 1.045,
+      "step": 12688
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1265017356369046e-05,
+      "loss": 1.0329,
+      "step": 12689
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.126364860150923e-05,
+      "loss": 1.0264,
+      "step": 12690
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1262279860519953e-05,
+      "loss": 0.9985,
+      "step": 12691
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1260911133426996e-05,
+      "loss": 1.0282,
+      "step": 12692
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1259542420256138e-05,
+      "loss": 1.0727,
+      "step": 12693
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1258173721033165e-05,
+      "loss": 1.0629,
+      "step": 12694
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.125680503578386e-05,
+      "loss": 0.9942,
+      "step": 12695
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1255436364534004e-05,
+      "loss": 1.1162,
+      "step": 12696
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1254067707309375e-05,
+      "loss": 1.0806,
+      "step": 12697
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1252699064135758e-05,
+      "loss": 1.0804,
+      "step": 12698
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1251330435038934e-05,
+      "loss": 1.0355,
+      "step": 12699
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1249961820044683e-05,
+      "loss": 0.9608,
+      "step": 12700
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.124859321917878e-05,
+      "loss": 1.0523,
+      "step": 12701
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1247224632467013e-05,
+      "loss": 0.9977,
+      "step": 12702
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.124585605993516e-05,
+      "loss": 0.9708,
+      "step": 12703
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1244487501608998e-05,
+      "loss": 1.0432,
+      "step": 12704
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1243118957514306e-05,
+      "loss": 1.0411,
+      "step": 12705
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1241750427676865e-05,
+      "loss": 1.0645,
+      "step": 12706
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1240381912122459e-05,
+      "loss": 1.0668,
+      "step": 12707
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1239013410876856e-05,
+      "loss": 1.0291,
+      "step": 12708
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1237644923965837e-05,
+      "loss": 1.055,
+      "step": 12709
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1236276451415188e-05,
+      "loss": 1.0854,
+      "step": 12710
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.123490799325068e-05,
+      "loss": 1.0897,
+      "step": 12711
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.123353954949809e-05,
+      "loss": 1.0494,
+      "step": 12712
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1232171120183192e-05,
+      "loss": 1.0691,
+      "step": 12713
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1230802705331775e-05,
+      "loss": 0.9837,
+      "step": 12714
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1229434304969606e-05,
+      "loss": 1.0449,
+      "step": 12715
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1228065919122462e-05,
+      "loss": 1.025,
+      "step": 12716
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1226697547816118e-05,
+      "loss": 1.0722,
+      "step": 12717
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1225329191076357e-05,
+      "loss": 0.9892,
+      "step": 12718
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1223960848928948e-05,
+      "loss": 1.0893,
+      "step": 12719
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1222592521399665e-05,
+      "loss": 1.0893,
+      "step": 12720
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1221224208514293e-05,
+      "loss": 0.9929,
+      "step": 12721
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1219855910298593e-05,
+      "loss": 1.0178,
+      "step": 12722
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1218487626778349e-05,
+      "loss": 0.9904,
+      "step": 12723
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.121711935797933e-05,
+      "loss": 1.0932,
+      "step": 12724
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1215751103927319e-05,
+      "loss": 0.9845,
+      "step": 12725
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1214382864648074e-05,
+      "loss": 1.0214,
+      "step": 12726
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1213014640167382e-05,
+      "loss": 1.022,
+      "step": 12727
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1211646430511008e-05,
+      "loss": 1.0839,
+      "step": 12728
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1210278235704734e-05,
+      "loss": 1.1071,
+      "step": 12729
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.120891005577432e-05,
+      "loss": 1.069,
+      "step": 12730
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1207541890745546e-05,
+      "loss": 1.0221,
+      "step": 12731
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1206173740644185e-05,
+      "loss": 1.0486,
+      "step": 12732
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1204805605496002e-05,
+      "loss": 1.0044,
+      "step": 12733
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1203437485326775e-05,
+      "loss": 1.1097,
+      "step": 12734
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.120206938016227e-05,
+      "loss": 1.0124,
+      "step": 12735
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1200701290028264e-05,
+      "loss": 1.0352,
+      "step": 12736
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1199333214950521e-05,
+      "loss": 1.0474,
+      "step": 12737
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1197965154954814e-05,
+      "loss": 1.0369,
+      "step": 12738
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1196597110066914e-05,
+      "loss": 0.9801,
+      "step": 12739
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1195229080312591e-05,
+      "loss": 1.0539,
+      "step": 12740
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.119386106571761e-05,
+      "loss": 1.0399,
+      "step": 12741
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1192493066307744e-05,
+      "loss": 1.0253,
+      "step": 12742
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1191125082108762e-05,
+      "loss": 1.0099,
+      "step": 12743
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.118975711314643e-05,
+      "loss": 1.076,
+      "step": 12744
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1188389159446516e-05,
+      "loss": 1.0654,
+      "step": 12745
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1187021221034791e-05,
+      "loss": 1.0349,
+      "step": 12746
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1185653297937024e-05,
+      "loss": 1.0049,
+      "step": 12747
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1184285390178978e-05,
+      "loss": 1.0197,
+      "step": 12748
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1182917497786415e-05,
+      "loss": 0.9891,
+      "step": 12749
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1181549620785114e-05,
+      "loss": 1.0775,
+      "step": 12750
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1180181759200838e-05,
+      "loss": 1.1056,
+      "step": 12751
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1178813913059348e-05,
+      "loss": 0.9688,
+      "step": 12752
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.117744608238641e-05,
+      "loss": 1.0232,
+      "step": 12753
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1176078267207798e-05,
+      "loss": 1.0143,
+      "step": 12754
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.117471046754927e-05,
+      "loss": 0.9787,
+      "step": 12755
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1173342683436592e-05,
+      "loss": 1.0173,
+      "step": 12756
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1171974914895527e-05,
+      "loss": 1.0826,
+      "step": 12757
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1170607161951848e-05,
+      "loss": 0.9919,
+      "step": 12758
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1169239424631308e-05,
+      "loss": 0.9976,
+      "step": 12759
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1167871702959678e-05,
+      "loss": 1.0007,
+      "step": 12760
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1166503996962718e-05,
+      "loss": 1.0457,
+      "step": 12761
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1165136306666196e-05,
+      "loss": 1.0459,
+      "step": 12762
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1163768632095871e-05,
+      "loss": 0.993,
+      "step": 12763
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1162400973277502e-05,
+      "loss": 1.0595,
+      "step": 12764
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1161033330236864e-05,
+      "loss": 1.0743,
+      "step": 12765
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1159665702999704e-05,
+      "loss": 1.0269,
+      "step": 12766
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1158298091591794e-05,
+      "loss": 1.1426,
+      "step": 12767
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.115693049603889e-05,
+      "loss": 1.0504,
+      "step": 12768
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.115556291636676e-05,
+      "loss": 1.05,
+      "step": 12769
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1154195352601154e-05,
+      "loss": 1.0435,
+      "step": 12770
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1152827804767843e-05,
+      "loss": 1.0278,
+      "step": 12771
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1151460272892583e-05,
+      "loss": 1.0813,
+      "step": 12772
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1150092757001134e-05,
+      "loss": 1.0601,
+      "step": 12773
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1148725257119252e-05,
+      "loss": 1.057,
+      "step": 12774
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1147357773272703e-05,
+      "loss": 1.06,
+      "step": 12775
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1145990305487246e-05,
+      "loss": 1.0504,
+      "step": 12776
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1144622853788631e-05,
+      "loss": 0.9857,
+      "step": 12777
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1143255418202627e-05,
+      "loss": 1.0425,
+      "step": 12778
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1141887998754988e-05,
+      "loss": 1.1054,
+      "step": 12779
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1140520595471473e-05,
+      "loss": 1.0587,
+      "step": 12780
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1139153208377831e-05,
+      "loss": 1.0257,
+      "step": 12781
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1137785837499833e-05,
+      "loss": 1.0418,
+      "step": 12782
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1136418482863229e-05,
+      "loss": 1.0379,
+      "step": 12783
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1135051144493774e-05,
+      "loss": 1.016,
+      "step": 12784
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1133683822417224e-05,
+      "loss": 1.0662,
+      "step": 12785
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1132316516659341e-05,
+      "loss": 1.0905,
+      "step": 12786
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1130949227245878e-05,
+      "loss": 1.0757,
+      "step": 12787
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1129581954202587e-05,
+      "loss": 1.0691,
+      "step": 12788
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1128214697555223e-05,
+      "loss": 1.0023,
+      "step": 12789
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1126847457329546e-05,
+      "loss": 1.0288,
+      "step": 12790
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1125480233551313e-05,
+      "loss": 1.0736,
+      "step": 12791
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1124113026246268e-05,
+      "loss": 1.0509,
+      "step": 12792
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1122745835440169e-05,
+      "loss": 1.0297,
+      "step": 12793
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1121378661158773e-05,
+      "loss": 1.0884,
+      "step": 12794
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.112001150342783e-05,
+      "loss": 0.9685,
+      "step": 12795
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1118644362273094e-05,
+      "loss": 1.0114,
+      "step": 12796
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1117277237720315e-05,
+      "loss": 1.0326,
+      "step": 12797
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1115910129795252e-05,
+      "loss": 1.0101,
+      "step": 12798
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.111454303852365e-05,
+      "loss": 0.9777,
+      "step": 12799
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1113175963931264e-05,
+      "loss": 1.0749,
+      "step": 12800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1111808906043843e-05,
+      "loss": 1.0733,
+      "step": 12801
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1110441864887143e-05,
+      "loss": 1.0236,
+      "step": 12802
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.110907484048691e-05,
+      "loss": 1.1116,
+      "step": 12803
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1107707832868896e-05,
+      "loss": 1.0265,
+      "step": 12804
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1106340842058853e-05,
+      "loss": 1.053,
+      "step": 12805
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1104973868082525e-05,
+      "loss": 1.1513,
+      "step": 12806
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1103606910965666e-05,
+      "loss": 1.0084,
+      "step": 12807
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1102239970734024e-05,
+      "loss": 1.0646,
+      "step": 12808
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1100873047413354e-05,
+      "loss": 1.1262,
+      "step": 12809
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1099506141029392e-05,
+      "loss": 1.0298,
+      "step": 12810
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1098139251607896e-05,
+      "loss": 1.0391,
+      "step": 12811
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1096772379174607e-05,
+      "loss": 1.1064,
+      "step": 12812
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1095405523755285e-05,
+      "loss": 1.0846,
+      "step": 12813
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1094038685375662e-05,
+      "loss": 0.9742,
+      "step": 12814
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1092671864061493e-05,
+      "loss": 1.0214,
+      "step": 12815
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1091305059838528e-05,
+      "loss": 1.0053,
+      "step": 12816
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1089938272732502e-05,
+      "loss": 1.006,
+      "step": 12817
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1088571502769169e-05,
+      "loss": 1.0646,
+      "step": 12818
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1087204749974273e-05,
+      "loss": 1.1131,
+      "step": 12819
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1085838014373563e-05,
+      "loss": 1.0758,
+      "step": 12820
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1084471295992776e-05,
+      "loss": 1.169,
+      "step": 12821
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1083104594857663e-05,
+      "loss": 1.0399,
+      "step": 12822
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1081737910993968e-05,
+      "loss": 1.1171,
+      "step": 12823
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1080371244427434e-05,
+      "loss": 1.0594,
+      "step": 12824
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1079004595183799e-05,
+      "loss": 1.0644,
+      "step": 12825
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1077637963288814e-05,
+      "loss": 1.0423,
+      "step": 12826
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1076271348768223e-05,
+      "loss": 1.0315,
+      "step": 12827
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1074904751647761e-05,
+      "loss": 1.0385,
+      "step": 12828
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1073538171953176e-05,
+      "loss": 1.0756,
+      "step": 12829
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1072171609710207e-05,
+      "loss": 1.0506,
+      "step": 12830
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1070805064944602e-05,
+      "loss": 1.1285,
+      "step": 12831
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1069438537682095e-05,
+      "loss": 1.0334,
+      "step": 12832
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1068072027948426e-05,
+      "loss": 0.9861,
+      "step": 12833
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1066705535769344e-05,
+      "loss": 0.9614,
+      "step": 12834
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1065339061170586e-05,
+      "loss": 1.0586,
+      "step": 12835
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.106397260417789e-05,
+      "loss": 0.9988,
+      "step": 12836
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1062606164816992e-05,
+      "loss": 1.0283,
+      "step": 12837
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1061239743113643e-05,
+      "loss": 1.0099,
+      "step": 12838
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1059873339093572e-05,
+      "loss": 0.9668,
+      "step": 12839
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1058506952782521e-05,
+      "loss": 1.1079,
+      "step": 12840
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1057140584206227e-05,
+      "loss": 0.9997,
+      "step": 12841
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1055774233390434e-05,
+      "loss": 1.0413,
+      "step": 12842
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1054407900360872e-05,
+      "loss": 1.057,
+      "step": 12843
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1053041585143282e-05,
+      "loss": 1.075,
+      "step": 12844
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1051675287763397e-05,
+      "loss": 1.0327,
+      "step": 12845
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1050309008246965e-05,
+      "loss": 1.0362,
+      "step": 12846
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.104894274661971e-05,
+      "loss": 1.0519,
+      "step": 12847
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1047576502907374e-05,
+      "loss": 1.0227,
+      "step": 12848
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1046210277135694e-05,
+      "loss": 1.0638,
+      "step": 12849
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1044844069330397e-05,
+      "loss": 1.0886,
+      "step": 12850
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1043477879517229e-05,
+      "loss": 0.9746,
+      "step": 12851
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1042111707721915e-05,
+      "loss": 0.9995,
+      "step": 12852
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1040745553970202e-05,
+      "loss": 1.0489,
+      "step": 12853
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.103937941828781e-05,
+      "loss": 0.9903,
+      "step": 12854
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.103801330070048e-05,
+      "loss": 1.061,
+      "step": 12855
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1036647201233946e-05,
+      "loss": 1.0778,
+      "step": 12856
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1035281119913937e-05,
+      "loss": 1.0834,
+      "step": 12857
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1033915056766187e-05,
+      "loss": 1.0411,
+      "step": 12858
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.103254901181643e-05,
+      "loss": 1.0403,
+      "step": 12859
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.10311829850904e-05,
+      "loss": 1.0878,
+      "step": 12860
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1029816976613822e-05,
+      "loss": 1.0283,
+      "step": 12861
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1028450986412432e-05,
+      "loss": 0.9927,
+      "step": 12862
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1027085014511958e-05,
+      "loss": 1.0061,
+      "step": 12863
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1025719060938138e-05,
+      "loss": 1.0768,
+      "step": 12864
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1024353125716691e-05,
+      "loss": 1.0062,
+      "step": 12865
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1022987208873355e-05,
+      "loss": 1.0286,
+      "step": 12866
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.102162131043386e-05,
+      "loss": 1.0558,
+      "step": 12867
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1020255430423931e-05,
+      "loss": 1.0688,
+      "step": 12868
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1018889568869295e-05,
+      "loss": 1.0768,
+      "step": 12869
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1017523725795687e-05,
+      "loss": 0.9839,
+      "step": 12870
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1016157901228833e-05,
+      "loss": 1.0126,
+      "step": 12871
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1014792095194458e-05,
+      "loss": 0.9995,
+      "step": 12872
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.101342630771829e-05,
+      "loss": 1.0658,
+      "step": 12873
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.101206053882606e-05,
+      "loss": 1.0041,
+      "step": 12874
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1010694788543494e-05,
+      "loss": 1.0097,
+      "step": 12875
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1009329056896313e-05,
+      "loss": 1.0184,
+      "step": 12876
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1007963343910248e-05,
+      "loss": 1.0964,
+      "step": 12877
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1006597649611025e-05,
+      "loss": 1.1269,
+      "step": 12878
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1005231974024368e-05,
+      "loss": 1.07,
+      "step": 12879
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1003866317176e-05,
+      "loss": 1.0424,
+      "step": 12880
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1002500679091645e-05,
+      "loss": 1.0054,
+      "step": 12881
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1001135059797036e-05,
+      "loss": 0.9497,
+      "step": 12882
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0999769459317888e-05,
+      "loss": 1.1342,
+      "step": 12883
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0998403877679925e-05,
+      "loss": 0.9827,
+      "step": 12884
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0997038314908875e-05,
+      "loss": 1.1036,
+      "step": 12885
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.099567277103046e-05,
+      "loss": 1.0013,
+      "step": 12886
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.09943072460704e-05,
+      "loss": 1.0227,
+      "step": 12887
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0992941740054417e-05,
+      "loss": 1.1012,
+      "step": 12888
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0991576253008237e-05,
+      "loss": 0.9992,
+      "step": 12889
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0990210784957576e-05,
+      "loss": 0.9602,
+      "step": 12890
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0988845335928158e-05,
+      "loss": 1.0073,
+      "step": 12891
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0987479905945703e-05,
+      "loss": 1.1424,
+      "step": 12892
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0986114495035936e-05,
+      "loss": 1.0999,
+      "step": 12893
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0984749103224567e-05,
+      "loss": 0.9694,
+      "step": 12894
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0983383730537325e-05,
+      "loss": 1.1012,
+      "step": 12895
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0982018376999923e-05,
+      "loss": 1.0714,
+      "step": 12896
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.098065304263809e-05,
+      "loss": 1.0979,
+      "step": 12897
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.097928772747753e-05,
+      "loss": 1.0677,
+      "step": 12898
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.097792243154397e-05,
+      "loss": 1.0486,
+      "step": 12899
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.097655715486313e-05,
+      "loss": 1.0034,
+      "step": 12900
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0975191897460722e-05,
+      "loss": 1.0983,
+      "step": 12901
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0973826659362463e-05,
+      "loss": 1.0662,
+      "step": 12902
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0972461440594073e-05,
+      "loss": 1.0806,
+      "step": 12903
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0971096241181271e-05,
+      "loss": 0.9801,
+      "step": 12904
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0969731061149766e-05,
+      "loss": 1.0436,
+      "step": 12905
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0968365900525274e-05,
+      "loss": 1.0611,
+      "step": 12906
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0967000759333518e-05,
+      "loss": 1.0811,
+      "step": 12907
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.096563563760021e-05,
+      "loss": 1.0933,
+      "step": 12908
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0964270535351058e-05,
+      "loss": 1.0797,
+      "step": 12909
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0962905452611784e-05,
+      "loss": 1.009,
+      "step": 12910
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0961540389408101e-05,
+      "loss": 1.0176,
+      "step": 12911
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0960175345765717e-05,
+      "loss": 1.1038,
+      "step": 12912
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0958810321710349e-05,
+      "loss": 1.0518,
+      "step": 12913
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0957445317267708e-05,
+      "loss": 1.0223,
+      "step": 12914
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0956080332463512e-05,
+      "loss": 1.0393,
+      "step": 12915
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0954715367323467e-05,
+      "loss": 1.0225,
+      "step": 12916
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0953350421873283e-05,
+      "loss": 1.0008,
+      "step": 12917
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0951985496138678e-05,
+      "loss": 1.0121,
+      "step": 12918
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0950620590145362e-05,
+      "loss": 1.0521,
+      "step": 12919
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0949255703919039e-05,
+      "loss": 1.0273,
+      "step": 12920
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0947890837485422e-05,
+      "loss": 1.1146,
+      "step": 12921
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0946525990870227e-05,
+      "loss": 1.027,
+      "step": 12922
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0945161164099155e-05,
+      "loss": 1.1034,
+      "step": 12923
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.094379635719792e-05,
+      "loss": 1.1151,
+      "step": 12924
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0942431570192226e-05,
+      "loss": 1.0124,
+      "step": 12925
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0941066803107788e-05,
+      "loss": 1.0847,
+      "step": 12926
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0939702055970311e-05,
+      "loss": 1.1225,
+      "step": 12927
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.09383373288055e-05,
+      "loss": 1.0795,
+      "step": 12928
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.093697262163906e-05,
+      "loss": 1.0866,
+      "step": 12929
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0935607934496707e-05,
+      "loss": 1.0555,
+      "step": 12930
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0934243267404141e-05,
+      "loss": 1.0389,
+      "step": 12931
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.093287862038707e-05,
+      "loss": 1.0201,
+      "step": 12932
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0931513993471199e-05,
+      "loss": 0.9851,
+      "step": 12933
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0930149386682229e-05,
+      "loss": 1.1082,
+      "step": 12934
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0928784800045872e-05,
+      "loss": 1.0607,
+      "step": 12935
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0927420233587829e-05,
+      "loss": 1.008,
+      "step": 12936
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0926055687333807e-05,
+      "loss": 1.063,
+      "step": 12937
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0924691161309503e-05,
+      "loss": 0.9996,
+      "step": 12938
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0923326655540626e-05,
+      "loss": 1.0317,
+      "step": 12939
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0921962170052882e-05,
+      "loss": 1.0759,
+      "step": 12940
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0920597704871964e-05,
+      "loss": 1.1126,
+      "step": 12941
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.091923326002358e-05,
+      "loss": 1.0319,
+      "step": 12942
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0917868835533432e-05,
+      "loss": 1.0911,
+      "step": 12943
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0916504431427223e-05,
+      "loss": 1.0607,
+      "step": 12944
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.091514004773065e-05,
+      "loss": 1.1758,
+      "step": 12945
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0913775684469414e-05,
+      "loss": 1.0064,
+      "step": 12946
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0912411341669217e-05,
+      "loss": 1.068,
+      "step": 12947
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0911047019355763e-05,
+      "loss": 1.0056,
+      "step": 12948
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0909682717554745e-05,
+      "loss": 1.0444,
+      "step": 12949
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0908318436291861e-05,
+      "loss": 0.9725,
+      "step": 12950
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0906954175592818e-05,
+      "loss": 1.1066,
+      "step": 12951
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0905589935483308e-05,
+      "loss": 1.0608,
+      "step": 12952
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0904225715989029e-05,
+      "loss": 1.0473,
+      "step": 12953
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0902861517135681e-05,
+      "loss": 0.9946,
+      "step": 12954
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0901497338948962e-05,
+      "loss": 1.0548,
+      "step": 12955
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0900133181454565e-05,
+      "loss": 1.1389,
+      "step": 12956
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0898769044678188e-05,
+      "loss": 1.0857,
+      "step": 12957
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0897404928645528e-05,
+      "loss": 1.0457,
+      "step": 12958
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0896040833382284e-05,
+      "loss": 1.0632,
+      "step": 12959
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0894676758914145e-05,
+      "loss": 1.0083,
+      "step": 12960
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0893312705266805e-05,
+      "loss": 0.9976,
+      "step": 12961
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.089194867246597e-05,
+      "loss": 1.0368,
+      "step": 12962
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0890584660537317e-05,
+      "loss": 1.1468,
+      "step": 12963
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0889220669506553e-05,
+      "loss": 1.0761,
+      "step": 12964
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0887856699399364e-05,
+      "loss": 1.0161,
+      "step": 12965
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.088649275024145e-05,
+      "loss": 1.1818,
+      "step": 12966
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0885128822058498e-05,
+      "loss": 1.0556,
+      "step": 12967
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0883764914876199e-05,
+      "loss": 1.1235,
+      "step": 12968
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0882401028720247e-05,
+      "loss": 1.0244,
+      "step": 12969
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0881037163616336e-05,
+      "loss": 1.058,
+      "step": 12970
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0879673319590152e-05,
+      "loss": 1.0482,
+      "step": 12971
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.087830949666739e-05,
+      "loss": 1.0315,
+      "step": 12972
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0876945694873738e-05,
+      "loss": 1.0205,
+      "step": 12973
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0875581914234882e-05,
+      "loss": 1.0756,
+      "step": 12974
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0874218154776517e-05,
+      "loss": 1.0075,
+      "step": 12975
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0872854416524332e-05,
+      "loss": 1.0845,
+      "step": 12976
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0871490699504013e-05,
+      "loss": 1.0201,
+      "step": 12977
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0870127003741245e-05,
+      "loss": 1.042,
+      "step": 12978
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.086876332926172e-05,
+      "loss": 1.0662,
+      "step": 12979
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0867399676091127e-05,
+      "loss": 1.0011,
+      "step": 12980
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0866036044255151e-05,
+      "loss": 1.1,
+      "step": 12981
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0864672433779475e-05,
+      "loss": 1.0137,
+      "step": 12982
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.086330884468979e-05,
+      "loss": 0.9942,
+      "step": 12983
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0861945277011781e-05,
+      "loss": 1.0109,
+      "step": 12984
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0860581730771131e-05,
+      "loss": 0.9973,
+      "step": 12985
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0859218205993524e-05,
+      "loss": 1.0329,
+      "step": 12986
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0857854702704648e-05,
+      "loss": 1.0501,
+      "step": 12987
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0856491220930188e-05,
+      "loss": 1.0357,
+      "step": 12988
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0855127760695824e-05,
+      "loss": 1.0147,
+      "step": 12989
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0853764322027238e-05,
+      "loss": 1.0804,
+      "step": 12990
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0852400904950117e-05,
+      "loss": 1.0176,
+      "step": 12991
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0851037509490144e-05,
+      "loss": 1.0419,
+      "step": 12992
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0849674135672999e-05,
+      "loss": 1.0199,
+      "step": 12993
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.084831078352436e-05,
+      "loss": 0.9997,
+      "step": 12994
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0846947453069917e-05,
+      "loss": 1.0308,
+      "step": 12995
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0845584144335344e-05,
+      "loss": 1.069,
+      "step": 12996
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.084422085734632e-05,
+      "loss": 0.9993,
+      "step": 12997
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.084285759212853e-05,
+      "loss": 1.0757,
+      "step": 12998
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0841494348707657e-05,
+      "loss": 1.0106,
+      "step": 12999
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.084013112710937e-05,
+      "loss": 0.9855,
+      "step": 13000
+    },
+    {
+      "epoch": 0.55,
+      "eval_text_loss": 0.7909708619117737,
+      "eval_text_runtime": 1.4009,
+      "eval_text_samples_per_second": 52.825,
+      "eval_text_steps_per_second": 1.428,
+      "step": 13000
+    },
+    {
+      "epoch": 0.55,
+      "eval_svit_loss": 1.1249957084655762,
+      "eval_svit_runtime": 4.9721,
+      "eval_svit_samples_per_second": 136.963,
+      "eval_svit_steps_per_second": 2.212,
+      "step": 13000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0838767927359352e-05,
+      "loss": 1.0491,
+      "step": 13001
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0837404749483285e-05,
+      "loss": 1.0238,
+      "step": 13002
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0836041593506846e-05,
+      "loss": 1.0281,
+      "step": 13003
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0834678459455708e-05,
+      "loss": 1.0298,
+      "step": 13004
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0833315347355551e-05,
+      "loss": 1.0623,
+      "step": 13005
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0831952257232055e-05,
+      "loss": 1.0214,
+      "step": 13006
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0830589189110889e-05,
+      "loss": 1.0176,
+      "step": 13007
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0829226143017734e-05,
+      "loss": 1.015,
+      "step": 13008
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0827863118978263e-05,
+      "loss": 1.0629,
+      "step": 13009
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0826500117018156e-05,
+      "loss": 0.9858,
+      "step": 13010
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0825137137163083e-05,
+      "loss": 1.0346,
+      "step": 13011
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0823774179438719e-05,
+      "loss": 1.1362,
+      "step": 13012
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0822411243870735e-05,
+      "loss": 1.0422,
+      "step": 13013
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0821048330484814e-05,
+      "loss": 1.096,
+      "step": 13014
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0819685439306618e-05,
+      "loss": 1.0947,
+      "step": 13015
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0818322570361825e-05,
+      "loss": 1.0927,
+      "step": 13016
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.081695972367611e-05,
+      "loss": 1.0641,
+      "step": 13017
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0815596899275135e-05,
+      "loss": 1.0472,
+      "step": 13018
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.081423409718458e-05,
+      "loss": 1.1286,
+      "step": 13019
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0812871317430113e-05,
+      "loss": 1.0546,
+      "step": 13020
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0811508560037409e-05,
+      "loss": 1.0788,
+      "step": 13021
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0810145825032127e-05,
+      "loss": 0.993,
+      "step": 13022
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0808783112439947e-05,
+      "loss": 1.0258,
+      "step": 13023
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0807420422286535e-05,
+      "loss": 1.0262,
+      "step": 13024
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0806057754597558e-05,
+      "loss": 1.0893,
+      "step": 13025
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0804695109398686e-05,
+      "loss": 1.08,
+      "step": 13026
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0803332486715587e-05,
+      "loss": 1.0251,
+      "step": 13027
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0801969886573929e-05,
+      "loss": 1.0493,
+      "step": 13028
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0800607308999378e-05,
+      "loss": 1.0411,
+      "step": 13029
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0799244754017598e-05,
+      "loss": 1.1183,
+      "step": 13030
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0797882221654261e-05,
+      "loss": 1.0206,
+      "step": 13031
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0796519711935032e-05,
+      "loss": 1.0121,
+      "step": 13032
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0795157224885572e-05,
+      "loss": 1.0207,
+      "step": 13033
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0793794760531546e-05,
+      "loss": 1.007,
+      "step": 13034
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0792432318898624e-05,
+      "loss": 1.0064,
+      "step": 13035
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0791069900012467e-05,
+      "loss": 1.1278,
+      "step": 13036
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0789707503898738e-05,
+      "loss": 1.1181,
+      "step": 13037
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0788345130583098e-05,
+      "loss": 1.0667,
+      "step": 13038
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0786982780091218e-05,
+      "loss": 1.098,
+      "step": 13039
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0785620452448753e-05,
+      "loss": 1.0664,
+      "step": 13040
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0784258147681364e-05,
+      "loss": 1.0965,
+      "step": 13041
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0782895865814718e-05,
+      "loss": 1.0473,
+      "step": 13042
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0781533606874477e-05,
+      "loss": 1.0928,
+      "step": 13043
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0780171370886294e-05,
+      "loss": 0.9734,
+      "step": 13044
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0778809157875834e-05,
+      "loss": 0.9907,
+      "step": 13045
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.077744696786876e-05,
+      "loss": 1.0324,
+      "step": 13046
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0776084800890724e-05,
+      "loss": 1.0728,
+      "step": 13047
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0774722656967389e-05,
+      "loss": 1.0448,
+      "step": 13048
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0773360536124411e-05,
+      "loss": 1.0796,
+      "step": 13049
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0771998438387457e-05,
+      "loss": 1.0382,
+      "step": 13050
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0770636363782172e-05,
+      "loss": 1.0915,
+      "step": 13051
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0769274312334222e-05,
+      "loss": 1.0617,
+      "step": 13052
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0767912284069257e-05,
+      "loss": 1.0857,
+      "step": 13053
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.076655027901294e-05,
+      "loss": 0.9963,
+      "step": 13054
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0765188297190924e-05,
+      "loss": 1.0029,
+      "step": 13055
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0763826338628864e-05,
+      "loss": 1.0464,
+      "step": 13056
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0762464403352418e-05,
+      "loss": 1.0515,
+      "step": 13057
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0761102491387231e-05,
+      "loss": 0.9904,
+      "step": 13058
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.075974060275897e-05,
+      "loss": 1.0103,
+      "step": 13059
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.075837873749328e-05,
+      "loss": 1.0601,
+      "step": 13060
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.075701689561582e-05,
+      "loss": 1.0457,
+      "step": 13061
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0755655077152236e-05,
+      "loss": 1.0577,
+      "step": 13062
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0754293282128186e-05,
+      "loss": 1.0726,
+      "step": 13063
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0752931510569319e-05,
+      "loss": 0.9369,
+      "step": 13064
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0751569762501293e-05,
+      "loss": 1.0179,
+      "step": 13065
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0750208037949747e-05,
+      "loss": 1.009,
+      "step": 13066
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0748846336940339e-05,
+      "loss": 1.0481,
+      "step": 13067
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0747484659498724e-05,
+      "loss": 1.0512,
+      "step": 13068
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0746123005650541e-05,
+      "loss": 1.0504,
+      "step": 13069
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0744761375421442e-05,
+      "loss": 1.0796,
+      "step": 13070
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0743399768837083e-05,
+      "loss": 1.0145,
+      "step": 13071
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0742038185923107e-05,
+      "loss": 1.055,
+      "step": 13072
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0740676626705163e-05,
+      "loss": 1.0569,
+      "step": 13073
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0739315091208892e-05,
+      "loss": 1.0975,
+      "step": 13074
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0737953579459951e-05,
+      "loss": 1.0531,
+      "step": 13075
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0736592091483986e-05,
+      "loss": 1.0716,
+      "step": 13076
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0735230627306636e-05,
+      "loss": 1.0954,
+      "step": 13077
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0733869186953549e-05,
+      "loss": 1.116,
+      "step": 13078
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0732507770450376e-05,
+      "loss": 1.1084,
+      "step": 13079
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0731146377822755e-05,
+      "loss": 1.1145,
+      "step": 13080
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0729785009096333e-05,
+      "loss": 1.067,
+      "step": 13081
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0728423664296751e-05,
+      "loss": 1.0797,
+      "step": 13082
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0727062343449658e-05,
+      "loss": 1.0585,
+      "step": 13083
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0725701046580694e-05,
+      "loss": 1.1349,
+      "step": 13084
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.07243397737155e-05,
+      "loss": 1.0642,
+      "step": 13085
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.072297852487972e-05,
+      "loss": 1.0904,
+      "step": 13086
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0721617300098998e-05,
+      "loss": 0.9777,
+      "step": 13087
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0720256099398972e-05,
+      "loss": 1.0334,
+      "step": 13088
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0718894922805278e-05,
+      "loss": 1.0634,
+      "step": 13089
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0717533770343569e-05,
+      "loss": 1.0838,
+      "step": 13090
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0716172642039472e-05,
+      "loss": 1.0466,
+      "step": 13091
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0714811537918634e-05,
+      "loss": 1.0678,
+      "step": 13092
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.071345045800669e-05,
+      "loss": 1.0712,
+      "step": 13093
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0712089402329285e-05,
+      "loss": 1.0851,
+      "step": 13094
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0710728370912044e-05,
+      "loss": 1.1304,
+      "step": 13095
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0709367363780618e-05,
+      "loss": 1.0333,
+      "step": 13096
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0708006380960633e-05,
+      "loss": 1.1162,
+      "step": 13097
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0706645422477738e-05,
+      "loss": 1.0193,
+      "step": 13098
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.070528448835756e-05,
+      "loss": 1.1047,
+      "step": 13099
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0703923578625737e-05,
+      "loss": 1.0623,
+      "step": 13100
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0702562693307907e-05,
+      "loss": 0.9904,
+      "step": 13101
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0701201832429697e-05,
+      "loss": 1.0189,
+      "step": 13102
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0699840996016749e-05,
+      "loss": 1.0078,
+      "step": 13103
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0698480184094694e-05,
+      "loss": 1.0916,
+      "step": 13104
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0697119396689168e-05,
+      "loss": 1.0374,
+      "step": 13105
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0695758633825798e-05,
+      "loss": 1.0495,
+      "step": 13106
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0694397895530222e-05,
+      "loss": 0.9831,
+      "step": 13107
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0693037181828072e-05,
+      "loss": 1.0272,
+      "step": 13108
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0691676492744977e-05,
+      "loss": 0.9285,
+      "step": 13109
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0690315828306566e-05,
+      "loss": 1.0787,
+      "step": 13110
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0688955188538474e-05,
+      "loss": 0.9996,
+      "step": 13111
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0687594573466333e-05,
+      "loss": 1.058,
+      "step": 13112
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0686233983115766e-05,
+      "loss": 1.0679,
+      "step": 13113
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0684873417512402e-05,
+      "loss": 1.0841,
+      "step": 13114
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0683512876681878e-05,
+      "loss": 1.0163,
+      "step": 13115
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0682152360649818e-05,
+      "loss": 0.9567,
+      "step": 13116
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0680791869441849e-05,
+      "loss": 1.0983,
+      "step": 13117
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0679431403083596e-05,
+      "loss": 0.9319,
+      "step": 13118
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0678070961600692e-05,
+      "loss": 0.9985,
+      "step": 13119
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.067671054501876e-05,
+      "loss": 1.0432,
+      "step": 13120
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0675350153363424e-05,
+      "loss": 0.9959,
+      "step": 13121
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.067398978666031e-05,
+      "loss": 1.0873,
+      "step": 13122
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.067262944493505e-05,
+      "loss": 1.0161,
+      "step": 13123
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0671269128213259e-05,
+      "loss": 1.0553,
+      "step": 13124
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0669908836520567e-05,
+      "loss": 1.0804,
+      "step": 13125
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0668548569882592e-05,
+      "loss": 1.036,
+      "step": 13126
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0667188328324967e-05,
+      "loss": 1.0667,
+      "step": 13127
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0665828111873307e-05,
+      "loss": 0.9804,
+      "step": 13128
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0664467920553233e-05,
+      "loss": 1.1033,
+      "step": 13129
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0663107754390374e-05,
+      "loss": 1.1343,
+      "step": 13130
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.066174761341034e-05,
+      "loss": 1.0409,
+      "step": 13131
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0660387497638765e-05,
+      "loss": 1.1057,
+      "step": 13132
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0659027407101256e-05,
+      "loss": 1.0124,
+      "step": 13133
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0657667341823449e-05,
+      "loss": 1.0512,
+      "step": 13134
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0656307301830945e-05,
+      "loss": 1.0556,
+      "step": 13135
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0654947287149375e-05,
+      "loss": 0.9773,
+      "step": 13136
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0653587297804353e-05,
+      "loss": 1.0457,
+      "step": 13137
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0652227333821504e-05,
+      "loss": 1.1197,
+      "step": 13138
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0650867395226433e-05,
+      "loss": 1.0329,
+      "step": 13139
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0649507482044766e-05,
+      "loss": 1.0946,
+      "step": 13140
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0648147594302118e-05,
+      "loss": 1.0639,
+      "step": 13141
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.06467877320241e-05,
+      "loss": 1.1047,
+      "step": 13142
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0645427895236336e-05,
+      "loss": 1.0944,
+      "step": 13143
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0644068083964434e-05,
+      "loss": 1.0123,
+      "step": 13144
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0642708298234013e-05,
+      "loss": 1.0051,
+      "step": 13145
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0641348538070682e-05,
+      "loss": 1.0666,
+      "step": 13146
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0639988803500059e-05,
+      "loss": 1.1321,
+      "step": 13147
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0638629094547757e-05,
+      "loss": 1.0859,
+      "step": 13148
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0637269411239387e-05,
+      "loss": 1.0209,
+      "step": 13149
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.063590975360056e-05,
+      "loss": 1.0257,
+      "step": 13150
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.063455012165689e-05,
+      "loss": 1.0425,
+      "step": 13151
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0633190515433991e-05,
+      "loss": 1.1301,
+      "step": 13152
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0631830934957467e-05,
+      "loss": 1.0434,
+      "step": 13153
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.063047138025293e-05,
+      "loss": 0.9843,
+      "step": 13154
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0629111851345991e-05,
+      "loss": 1.1069,
+      "step": 13155
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0627752348262262e-05,
+      "loss": 1.0031,
+      "step": 13156
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.062639287102735e-05,
+      "loss": 1.1281,
+      "step": 13157
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0625033419666857e-05,
+      "loss": 1.0539,
+      "step": 13158
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0623673994206398e-05,
+      "loss": 1.0955,
+      "step": 13159
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0622314594671581e-05,
+      "loss": 1.0657,
+      "step": 13160
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0620955221088009e-05,
+      "loss": 1.0788,
+      "step": 13161
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0619595873481287e-05,
+      "loss": 1.1102,
+      "step": 13162
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0618236551877028e-05,
+      "loss": 1.1,
+      "step": 13163
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0616877256300828e-05,
+      "loss": 0.9967,
+      "step": 13164
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.06155179867783e-05,
+      "loss": 1.1216,
+      "step": 13165
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0614158743335041e-05,
+      "loss": 1.0625,
+      "step": 13166
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0612799525996662e-05,
+      "loss": 1.071,
+      "step": 13167
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0611440334788762e-05,
+      "loss": 1.074,
+      "step": 13168
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0610081169736946e-05,
+      "loss": 1.0632,
+      "step": 13169
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0608722030866811e-05,
+      "loss": 1.1226,
+      "step": 13170
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0607362918203968e-05,
+      "loss": 1.0611,
+      "step": 13171
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0606003831774013e-05,
+      "loss": 1.0507,
+      "step": 13172
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0604644771602543e-05,
+      "loss": 1.0688,
+      "step": 13173
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0603285737715171e-05,
+      "loss": 1.0507,
+      "step": 13174
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0601926730137481e-05,
+      "loss": 1.1223,
+      "step": 13175
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0600567748895083e-05,
+      "loss": 1.0609,
+      "step": 13176
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.059920879401357e-05,
+      "loss": 1.0855,
+      "step": 13177
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0597849865518551e-05,
+      "loss": 0.9514,
+      "step": 13178
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0596490963435612e-05,
+      "loss": 1.113,
+      "step": 13179
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0595132087790354e-05,
+      "loss": 1.1365,
+      "step": 13180
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0593773238608373e-05,
+      "loss": 1.0034,
+      "step": 13181
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0592414415915275e-05,
+      "loss": 1.0167,
+      "step": 13182
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0591055619736641e-05,
+      "loss": 1.0757,
+      "step": 13183
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0589696850098077e-05,
+      "loss": 1.0869,
+      "step": 13184
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0588338107025175e-05,
+      "loss": 1.0956,
+      "step": 13185
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0586979390543525e-05,
+      "loss": 1.0556,
+      "step": 13186
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0585620700678728e-05,
+      "loss": 1.1096,
+      "step": 13187
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0584262037456374e-05,
+      "loss": 1.0076,
+      "step": 13188
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0582903400902057e-05,
+      "loss": 1.0701,
+      "step": 13189
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0581544791041366e-05,
+      "loss": 1.1604,
+      "step": 13190
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0580186207899896e-05,
+      "loss": 0.9607,
+      "step": 13191
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0578827651503243e-05,
+      "loss": 1.0117,
+      "step": 13192
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.057746912187699e-05,
+      "loss": 1.0598,
+      "step": 13193
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0576110619046726e-05,
+      "loss": 1.101,
+      "step": 13194
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0574752143038048e-05,
+      "loss": 1.0211,
+      "step": 13195
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0573393693876546e-05,
+      "loss": 1.0797,
+      "step": 13196
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0572035271587803e-05,
+      "loss": 1.1063,
+      "step": 13197
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0570676876197408e-05,
+      "loss": 1.0758,
+      "step": 13198
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0569318507730952e-05,
+      "loss": 1.0749,
+      "step": 13199
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0567960166214026e-05,
+      "loss": 1.1067,
+      "step": 13200
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0566601851672207e-05,
+      "loss": 1.0731,
+      "step": 13201
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0565243564131086e-05,
+      "loss": 1.0583,
+      "step": 13202
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0563885303616252e-05,
+      "loss": 1.1064,
+      "step": 13203
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0562527070153288e-05,
+      "loss": 1.0644,
+      "step": 13204
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0561168863767776e-05,
+      "loss": 1.0079,
+      "step": 13205
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0559810684485301e-05,
+      "loss": 1.1183,
+      "step": 13206
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0558452532331455e-05,
+      "loss": 1.0669,
+      "step": 13207
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.055709440733181e-05,
+      "loss": 1.1138,
+      "step": 13208
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0555736309511956e-05,
+      "loss": 0.9709,
+      "step": 13209
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0554378238897469e-05,
+      "loss": 1.0269,
+      "step": 13210
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0553020195513939e-05,
+      "loss": 0.9531,
+      "step": 13211
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0551662179386944e-05,
+      "loss": 1.1158,
+      "step": 13212
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.055030419054206e-05,
+      "loss": 1.0579,
+      "step": 13213
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0548946229004875e-05,
+      "loss": 1.1192,
+      "step": 13214
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.054758829480096e-05,
+      "loss": 1.0963,
+      "step": 13215
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.05462303879559e-05,
+      "loss": 1.097,
+      "step": 13216
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0544872508495272e-05,
+      "loss": 1.1013,
+      "step": 13217
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0543514656444659e-05,
+      "loss": 1.0609,
+      "step": 13218
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0542156831829631e-05,
+      "loss": 1.076,
+      "step": 13219
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0540799034675768e-05,
+      "loss": 1.1533,
+      "step": 13220
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0539441265008647e-05,
+      "loss": 1.0865,
+      "step": 13221
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0538083522853848e-05,
+      "loss": 1.0798,
+      "step": 13222
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.053672580823694e-05,
+      "loss": 0.9414,
+      "step": 13223
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.05353681211835e-05,
+      "loss": 1.1763,
+      "step": 13224
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0534010461719106e-05,
+      "loss": 0.9798,
+      "step": 13225
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.053265282986933e-05,
+      "loss": 1.0677,
+      "step": 13226
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0531295225659738e-05,
+      "loss": 1.151,
+      "step": 13227
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0529937649115915e-05,
+      "loss": 1.081,
+      "step": 13228
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.052858010026343e-05,
+      "loss": 1.044,
+      "step": 13229
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0527222579127849e-05,
+      "loss": 1.1095,
+      "step": 13230
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.052586508573475e-05,
+      "loss": 1.0926,
+      "step": 13231
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.05245076201097e-05,
+      "loss": 1.1591,
+      "step": 13232
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0523150182278276e-05,
+      "loss": 1.0998,
+      "step": 13233
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0521792772266036e-05,
+      "loss": 1.099,
+      "step": 13234
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.052043539009856e-05,
+      "loss": 1.0863,
+      "step": 13235
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0519078035801415e-05,
+      "loss": 1.1245,
+      "step": 13236
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0517720709400164e-05,
+      "loss": 1.1052,
+      "step": 13237
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0516363410920376e-05,
+      "loss": 1.0903,
+      "step": 13238
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0515006140387622e-05,
+      "loss": 1.0255,
+      "step": 13239
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.051364889782747e-05,
+      "loss": 1.0966,
+      "step": 13240
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.051229168326548e-05,
+      "loss": 0.9799,
+      "step": 13241
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0510934496727218e-05,
+      "loss": 1.1257,
+      "step": 13242
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0509577338238254e-05,
+      "loss": 1.0853,
+      "step": 13243
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0508220207824155e-05,
+      "loss": 1.1075,
+      "step": 13244
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0506863105510476e-05,
+      "loss": 1.0033,
+      "step": 13245
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0505506031322783e-05,
+      "loss": 1.0527,
+      "step": 13246
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0504148985286645e-05,
+      "loss": 1.1086,
+      "step": 13247
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.050279196742762e-05,
+      "loss": 1.073,
+      "step": 13248
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0501434977771267e-05,
+      "loss": 1.1102,
+      "step": 13249
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0500078016343151e-05,
+      "loss": 1.1016,
+      "step": 13250
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0498721083168836e-05,
+      "loss": 1.0574,
+      "step": 13251
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0497364178273878e-05,
+      "loss": 1.1273,
+      "step": 13252
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0496007301683839e-05,
+      "loss": 1.0252,
+      "step": 13253
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0494650453424271e-05,
+      "loss": 1.1114,
+      "step": 13254
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0493293633520745e-05,
+      "loss": 1.0883,
+      "step": 13255
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.049193684199881e-05,
+      "loss": 1.1052,
+      "step": 13256
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0490580078884029e-05,
+      "loss": 1.0451,
+      "step": 13257
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0489223344201957e-05,
+      "loss": 1.114,
+      "step": 13258
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0487866637978147e-05,
+      "loss": 1.0455,
+      "step": 13259
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.048650996023816e-05,
+      "loss": 1.0142,
+      "step": 13260
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0485153311007548e-05,
+      "loss": 1.069,
+      "step": 13261
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0483796690311874e-05,
+      "loss": 1.0425,
+      "step": 13262
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0482440098176681e-05,
+      "loss": 1.079,
+      "step": 13263
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0481083534627529e-05,
+      "loss": 1.0719,
+      "step": 13264
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0479726999689974e-05,
+      "loss": 1.1079,
+      "step": 13265
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0478370493389562e-05,
+      "loss": 1.1083,
+      "step": 13266
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0477014015751848e-05,
+      "loss": 1.0393,
+      "step": 13267
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0475657566802387e-05,
+      "loss": 1.1056,
+      "step": 13268
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0474301146566729e-05,
+      "loss": 1.0195,
+      "step": 13269
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0472944755070422e-05,
+      "loss": 1.0523,
+      "step": 13270
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0471588392339016e-05,
+      "loss": 1.0883,
+      "step": 13271
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0470232058398062e-05,
+      "loss": 1.0789,
+      "step": 13272
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0468875753273114e-05,
+      "loss": 1.0678,
+      "step": 13273
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.046751947698971e-05,
+      "loss": 1.0555,
+      "step": 13274
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0466163229573407e-05,
+      "loss": 1.0943,
+      "step": 13275
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.046480701104975e-05,
+      "loss": 1.0407,
+      "step": 13276
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0463450821444286e-05,
+      "loss": 1.1461,
+      "step": 13277
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0462094660782555e-05,
+      "loss": 1.0588,
+      "step": 13278
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.046073852909011e-05,
+      "loss": 1.0733,
+      "step": 13279
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0459382426392499e-05,
+      "loss": 1.1143,
+      "step": 13280
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.045802635271526e-05,
+      "loss": 1.0845,
+      "step": 13281
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0456670308083936e-05,
+      "loss": 1.0554,
+      "step": 13282
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0455314292524075e-05,
+      "loss": 1.0811,
+      "step": 13283
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0453958306061222e-05,
+      "loss": 1.1093,
+      "step": 13284
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0452602348720914e-05,
+      "loss": 1.1118,
+      "step": 13285
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0451246420528694e-05,
+      "loss": 1.0229,
+      "step": 13286
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0449890521510109e-05,
+      "loss": 1.0197,
+      "step": 13287
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0448534651690693e-05,
+      "loss": 1.0252,
+      "step": 13288
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0447178811095987e-05,
+      "loss": 1.0965,
+      "step": 13289
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0445822999751535e-05,
+      "loss": 1.1331,
+      "step": 13290
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0444467217682876e-05,
+      "loss": 1.026,
+      "step": 13291
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0443111464915544e-05,
+      "loss": 1.1002,
+      "step": 13292
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0441755741475079e-05,
+      "loss": 1.0461,
+      "step": 13293
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0440400047387019e-05,
+      "loss": 1.105,
+      "step": 13294
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0439044382676905e-05,
+      "loss": 1.0911,
+      "step": 13295
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0437688747370267e-05,
+      "loss": 1.133,
+      "step": 13296
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0436333141492643e-05,
+      "loss": 1.0557,
+      "step": 13297
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0434977565069572e-05,
+      "loss": 0.9755,
+      "step": 13298
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0433622018126581e-05,
+      "loss": 1.1178,
+      "step": 13299
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0432266500689212e-05,
+      "loss": 1.0376,
+      "step": 13300
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0430911012782995e-05,
+      "loss": 1.0092,
+      "step": 13301
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0429555554433465e-05,
+      "loss": 1.0281,
+      "step": 13302
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.042820012566615e-05,
+      "loss": 1.0251,
+      "step": 13303
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0426844726506588e-05,
+      "loss": 1.0134,
+      "step": 13304
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0425489356980305e-05,
+      "loss": 1.0768,
+      "step": 13305
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.042413401711284e-05,
+      "loss": 1.1639,
+      "step": 13306
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0422778706929713e-05,
+      "loss": 1.0247,
+      "step": 13307
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.042142342645646e-05,
+      "loss": 1.0379,
+      "step": 13308
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.042006817571861e-05,
+      "loss": 1.0859,
+      "step": 13309
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0418712954741691e-05,
+      "loss": 1.001,
+      "step": 13310
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0417357763551227e-05,
+      "loss": 1.0016,
+      "step": 13311
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0416002602172751e-05,
+      "loss": 1.0845,
+      "step": 13312
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.041464747063179e-05,
+      "loss": 1.1575,
+      "step": 13313
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.041329236895387e-05,
+      "loss": 1.0976,
+      "step": 13314
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0411937297164508e-05,
+      "loss": 1.0467,
+      "step": 13315
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0410582255289243e-05,
+      "loss": 1.0172,
+      "step": 13316
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0409227243353591e-05,
+      "loss": 1.085,
+      "step": 13317
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0407872261383079e-05,
+      "loss": 1.1124,
+      "step": 13318
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.040651730940323e-05,
+      "loss": 0.9936,
+      "step": 13319
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.040516238743957e-05,
+      "loss": 0.9992,
+      "step": 13320
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0403807495517614e-05,
+      "loss": 1.0138,
+      "step": 13321
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.040245263366289e-05,
+      "loss": 1.0451,
+      "step": 13322
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0401097801900917e-05,
+      "loss": 1.0491,
+      "step": 13323
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.039974300025722e-05,
+      "loss": 1.0016,
+      "step": 13324
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0398388228757311e-05,
+      "loss": 1.1091,
+      "step": 13325
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0397033487426716e-05,
+      "loss": 1.0556,
+      "step": 13326
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0395678776290954e-05,
+      "loss": 1.0257,
+      "step": 13327
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0394324095375543e-05,
+      "loss": 1.02,
+      "step": 13328
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0392969444705998e-05,
+      "loss": 1.0623,
+      "step": 13329
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0391614824307835e-05,
+      "loss": 1.0445,
+      "step": 13330
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0390260234206578e-05,
+      "loss": 1.0683,
+      "step": 13331
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.038890567442774e-05,
+      "loss": 1.084,
+      "step": 13332
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0387551144996833e-05,
+      "loss": 1.0293,
+      "step": 13333
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0386196645939372e-05,
+      "loss": 1.0166,
+      "step": 13334
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0384842177280881e-05,
+      "loss": 1.0747,
+      "step": 13335
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0383487739046864e-05,
+      "loss": 1.0128,
+      "step": 13336
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0382133331262838e-05,
+      "loss": 1.0332,
+      "step": 13337
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0380778953954313e-05,
+      "loss": 1.0472,
+      "step": 13338
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0379424607146808e-05,
+      "loss": 1.04,
+      "step": 13339
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0378070290865826e-05,
+      "loss": 1.0212,
+      "step": 13340
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0376716005136884e-05,
+      "loss": 1.0032,
+      "step": 13341
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0375361749985493e-05,
+      "loss": 1.1092,
+      "step": 13342
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0374007525437156e-05,
+      "loss": 1.115,
+      "step": 13343
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0372653331517389e-05,
+      "loss": 0.991,
+      "step": 13344
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0371299168251698e-05,
+      "loss": 1.0633,
+      "step": 13345
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0369945035665595e-05,
+      "loss": 1.0315,
+      "step": 13346
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.036859093378458e-05,
+      "loss": 0.971,
+      "step": 13347
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0367236862634166e-05,
+      "loss": 1.0013,
+      "step": 13348
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.036588282223986e-05,
+      "loss": 1.0905,
+      "step": 13349
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0364528812627164e-05,
+      "loss": 1.0367,
+      "step": 13350
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0363174833821581e-05,
+      "loss": 1.0805,
+      "step": 13351
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0361820885848623e-05,
+      "loss": 1.0788,
+      "step": 13352
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0360466968733793e-05,
+      "loss": 0.981,
+      "step": 13353
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0359113082502589e-05,
+      "loss": 1.0595,
+      "step": 13354
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0357759227180517e-05,
+      "loss": 1.044,
+      "step": 13355
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.035640540279308e-05,
+      "loss": 1.0994,
+      "step": 13356
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0355051609365783e-05,
+      "loss": 1.1072,
+      "step": 13357
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.035369784692412e-05,
+      "loss": 1.0638,
+      "step": 13358
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0352344115493594e-05,
+      "loss": 1.0642,
+      "step": 13359
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.035099041509971e-05,
+      "loss": 1.0067,
+      "step": 13360
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0349636745767961e-05,
+      "loss": 1.0326,
+      "step": 13361
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0348283107523847e-05,
+      "loss": 1.0499,
+      "step": 13362
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.034692950039287e-05,
+      "loss": 0.9858,
+      "step": 13363
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0345575924400526e-05,
+      "loss": 1.0818,
+      "step": 13364
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0344222379572311e-05,
+      "loss": 1.0549,
+      "step": 13365
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.034286886593372e-05,
+      "loss": 1.0854,
+      "step": 13366
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0341515383510252e-05,
+      "loss": 1.1033,
+      "step": 13367
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0340161932327405e-05,
+      "loss": 1.0214,
+      "step": 13368
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0338808512410667e-05,
+      "loss": 1.0803,
+      "step": 13369
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0337455123785533e-05,
+      "loss": 1.04,
+      "step": 13370
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0336101766477505e-05,
+      "loss": 1.0417,
+      "step": 13371
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0334748440512063e-05,
+      "loss": 1.0577,
+      "step": 13372
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.033339514591471e-05,
+      "loss": 1.0405,
+      "step": 13373
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.033204188271093e-05,
+      "loss": 0.9762,
+      "step": 13374
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0330688650926224e-05,
+      "loss": 1.064,
+      "step": 13375
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0329335450586075e-05,
+      "loss": 1.0807,
+      "step": 13376
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0327982281715973e-05,
+      "loss": 1.0338,
+      "step": 13377
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0326629144341407e-05,
+      "loss": 1.0448,
+      "step": 13378
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0325276038487873e-05,
+      "loss": 1.003,
+      "step": 13379
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0323922964180853e-05,
+      "loss": 1.0606,
+      "step": 13380
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0322569921445835e-05,
+      "loss": 1.0667,
+      "step": 13381
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0321216910308308e-05,
+      "loss": 1.0616,
+      "step": 13382
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0319863930793757e-05,
+      "loss": 0.9606,
+      "step": 13383
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0318510982927668e-05,
+      "loss": 0.9796,
+      "step": 13384
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0317158066735526e-05,
+      "loss": 1.0452,
+      "step": 13385
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0315805182242819e-05,
+      "loss": 1.0164,
+      "step": 13386
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0314452329475025e-05,
+      "loss": 0.9814,
+      "step": 13387
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0313099508457632e-05,
+      "loss": 1.0603,
+      "step": 13388
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0311746719216122e-05,
+      "loss": 0.9696,
+      "step": 13389
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0310393961775979e-05,
+      "loss": 1.1054,
+      "step": 13390
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.030904123616268e-05,
+      "loss": 1.0371,
+      "step": 13391
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0307688542401708e-05,
+      "loss": 1.1192,
+      "step": 13392
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0306335880518549e-05,
+      "loss": 1.0445,
+      "step": 13393
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0304983250538675e-05,
+      "loss": 1.024,
+      "step": 13394
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0303630652487566e-05,
+      "loss": 1.0286,
+      "step": 13395
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0302278086390706e-05,
+      "loss": 1.1018,
+      "step": 13396
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0300925552273572e-05,
+      "loss": 1.007,
+      "step": 13397
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0299573050161637e-05,
+      "loss": 1.0747,
+      "step": 13398
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.029822058008038e-05,
+      "loss": 1.0074,
+      "step": 13399
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0296868142055278e-05,
+      "loss": 1.1215,
+      "step": 13400
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0295515736111809e-05,
+      "loss": 1.0667,
+      "step": 13401
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0294163362275444e-05,
+      "loss": 1.0735,
+      "step": 13402
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0292811020571656e-05,
+      "loss": 1.023,
+      "step": 13403
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0291458711025927e-05,
+      "loss": 1.1343,
+      "step": 13404
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0290106433663723e-05,
+      "loss": 1.0878,
+      "step": 13405
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0288754188510516e-05,
+      "loss": 1.0876,
+      "step": 13406
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0287401975591785e-05,
+      "loss": 1.0557,
+      "step": 13407
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0286049794932996e-05,
+      "loss": 1.0251,
+      "step": 13408
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.028469764655962e-05,
+      "loss": 1.0209,
+      "step": 13409
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0283345530497128e-05,
+      "loss": 0.9786,
+      "step": 13410
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0281993446770989e-05,
+      "loss": 1.0279,
+      "step": 13411
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0280641395406676e-05,
+      "loss": 1.093,
+      "step": 13412
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0279289376429653e-05,
+      "loss": 1.0463,
+      "step": 13413
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0277937389865386e-05,
+      "loss": 1.1012,
+      "step": 13414
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0276585435739352e-05,
+      "loss": 1.0349,
+      "step": 13415
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0275233514077003e-05,
+      "loss": 1.0418,
+      "step": 13416
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0273881624903817e-05,
+      "loss": 1.0062,
+      "step": 13417
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.027252976824525e-05,
+      "loss": 1.0488,
+      "step": 13418
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0271177944126777e-05,
+      "loss": 1.0374,
+      "step": 13419
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0269826152573857e-05,
+      "loss": 1.024,
+      "step": 13420
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0268474393611949e-05,
+      "loss": 1.0497,
+      "step": 13421
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.026712266726652e-05,
+      "loss": 1.0611,
+      "step": 13422
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0265770973563038e-05,
+      "loss": 1.1084,
+      "step": 13423
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0264419312526955e-05,
+      "loss": 1.1413,
+      "step": 13424
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0263067684183736e-05,
+      "loss": 1.013,
+      "step": 13425
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0261716088558846e-05,
+      "loss": 1.0925,
+      "step": 13426
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0260364525677733e-05,
+      "loss": 0.9744,
+      "step": 13427
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0259012995565867e-05,
+      "loss": 1.0973,
+      "step": 13428
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0257661498248702e-05,
+      "loss": 1.0045,
+      "step": 13429
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.02563100337517e-05,
+      "loss": 1.0329,
+      "step": 13430
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.025495860210031e-05,
+      "loss": 0.9665,
+      "step": 13431
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0253607203319997e-05,
+      "loss": 1.0858,
+      "step": 13432
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0252255837436217e-05,
+      "loss": 0.9777,
+      "step": 13433
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0250904504474419e-05,
+      "loss": 1.023,
+      "step": 13434
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.024955320446006e-05,
+      "loss": 1.0993,
+      "step": 13435
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0248201937418597e-05,
+      "loss": 0.9914,
+      "step": 13436
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0246850703375485e-05,
+      "loss": 1.0758,
+      "step": 13437
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0245499502356172e-05,
+      "loss": 1.0701,
+      "step": 13438
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.024414833438611e-05,
+      "loss": 1.0345,
+      "step": 13439
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0242797199490757e-05,
+      "loss": 1.054,
+      "step": 13440
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.024144609769556e-05,
+      "loss": 1.0513,
+      "step": 13441
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.024009502902597e-05,
+      "loss": 1.064,
+      "step": 13442
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0238743993507433e-05,
+      "loss": 1.0247,
+      "step": 13443
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0237392991165406e-05,
+      "loss": 1.0798,
+      "step": 13444
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0236042022025333e-05,
+      "loss": 0.9977,
+      "step": 13445
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0234691086112662e-05,
+      "loss": 1.036,
+      "step": 13446
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0233340183452839e-05,
+      "loss": 1.0126,
+      "step": 13447
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0231989314071317e-05,
+      "loss": 1.0325,
+      "step": 13448
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0230638477993535e-05,
+      "loss": 0.9991,
+      "step": 13449
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0229287675244942e-05,
+      "loss": 1.1078,
+      "step": 13450
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.022793690585098e-05,
+      "loss": 1.0524,
+      "step": 13451
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0226586169837102e-05,
+      "loss": 0.9985,
+      "step": 13452
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.022523546722874e-05,
+      "loss": 1.0199,
+      "step": 13453
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0223884798051341e-05,
+      "loss": 0.9911,
+      "step": 13454
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0222534162330355e-05,
+      "loss": 0.9996,
+      "step": 13455
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.022118356009121e-05,
+      "loss": 1.1506,
+      "step": 13456
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0219832991359356e-05,
+      "loss": 1.1296,
+      "step": 13457
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.021848245616023e-05,
+      "loss": 1.0284,
+      "step": 13458
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.021713195451928e-05,
+      "loss": 1.0901,
+      "step": 13459
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0215781486461932e-05,
+      "loss": 1.0229,
+      "step": 13460
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0214431052013635e-05,
+      "loss": 0.9555,
+      "step": 13461
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0213080651199818e-05,
+      "loss": 1.0876,
+      "step": 13462
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.021173028404593e-05,
+      "loss": 1.0217,
+      "step": 13463
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0210379950577398e-05,
+      "loss": 1.0557,
+      "step": 13464
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0209029650819662e-05,
+      "loss": 1.1054,
+      "step": 13465
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0207679384798159e-05,
+      "loss": 1.0273,
+      "step": 13466
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0206329152538315e-05,
+      "loss": 1.0762,
+      "step": 13467
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0204978954065573e-05,
+      "loss": 1.0429,
+      "step": 13468
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0203628789405364e-05,
+      "loss": 0.9306,
+      "step": 13469
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0202278658583125e-05,
+      "loss": 1.1259,
+      "step": 13470
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0200928561624277e-05,
+      "loss": 0.9183,
+      "step": 13471
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0199578498554261e-05,
+      "loss": 1.0585,
+      "step": 13472
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0198228469398505e-05,
+      "loss": 0.9993,
+      "step": 13473
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0196878474182443e-05,
+      "loss": 1.0468,
+      "step": 13474
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0195528512931497e-05,
+      "loss": 0.9716,
+      "step": 13475
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0194178585671102e-05,
+      "loss": 1.0415,
+      "step": 13476
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0192828692426686e-05,
+      "loss": 1.0513,
+      "step": 13477
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0191478833223675e-05,
+      "loss": 1.048,
+      "step": 13478
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0190129008087493e-05,
+      "loss": 1.0259,
+      "step": 13479
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0188779217043572e-05,
+      "loss": 1.0241,
+      "step": 13480
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.018742946011734e-05,
+      "loss": 1.1248,
+      "step": 13481
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0186079737334213e-05,
+      "loss": 1.0669,
+      "step": 13482
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.018473004871962e-05,
+      "loss": 1.114,
+      "step": 13483
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0183380394298987e-05,
+      "loss": 1.1183,
+      "step": 13484
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.018203077409774e-05,
+      "loss": 0.9722,
+      "step": 13485
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0180681188141292e-05,
+      "loss": 1.0216,
+      "step": 13486
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0179331636455067e-05,
+      "loss": 1.1495,
+      "step": 13487
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0177982119064496e-05,
+      "loss": 1.0088,
+      "step": 13488
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0176632635994991e-05,
+      "loss": 0.9962,
+      "step": 13489
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0175283187271974e-05,
+      "loss": 1.0418,
+      "step": 13490
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.017393377292086e-05,
+      "loss": 0.9655,
+      "step": 13491
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0172584392967078e-05,
+      "loss": 1.0433,
+      "step": 13492
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0171235047436038e-05,
+      "loss": 1.0484,
+      "step": 13493
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0169885736353157e-05,
+      "loss": 1.0671,
+      "step": 13494
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0168536459743859e-05,
+      "loss": 1.0412,
+      "step": 13495
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0167187217633555e-05,
+      "loss": 1.0164,
+      "step": 13496
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.016583801004766e-05,
+      "loss": 1.0696,
+      "step": 13497
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0164488837011585e-05,
+      "loss": 1.0077,
+      "step": 13498
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0163139698550758e-05,
+      "loss": 1.0189,
+      "step": 13499
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0161790594690576e-05,
+      "loss": 1.0806,
+      "step": 13500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0160441525456461e-05,
+      "loss": 1.0601,
+      "step": 13501
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0159092490873823e-05,
+      "loss": 1.0115,
+      "step": 13502
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.015774349096808e-05,
+      "loss": 1.0745,
+      "step": 13503
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.015639452576463e-05,
+      "loss": 1.0219,
+      "step": 13504
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.015504559528889e-05,
+      "loss": 1.0162,
+      "step": 13505
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0153696699566271e-05,
+      "loss": 0.9871,
+      "step": 13506
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0152347838622182e-05,
+      "loss": 1.0435,
+      "step": 13507
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.015099901248203e-05,
+      "loss": 1.0531,
+      "step": 13508
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0149650221171223e-05,
+      "loss": 1.0584,
+      "step": 13509
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0148301464715169e-05,
+      "loss": 1.0645,
+      "step": 13510
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0146952743139269e-05,
+      "loss": 0.9792,
+      "step": 13511
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0145604056468935e-05,
+      "loss": 1.0266,
+      "step": 13512
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0144255404729568e-05,
+      "loss": 1.0134,
+      "step": 13513
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0142906787946577e-05,
+      "loss": 1.0493,
+      "step": 13514
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0141558206145359e-05,
+      "loss": 1.0683,
+      "step": 13515
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0140209659351322e-05,
+      "loss": 1.0141,
+      "step": 13516
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0138861147589867e-05,
+      "loss": 1.0502,
+      "step": 13517
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0137512670886397e-05,
+      "loss": 1.0229,
+      "step": 13518
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0136164229266307e-05,
+      "loss": 1.1143,
+      "step": 13519
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0134815822755007e-05,
+      "loss": 1.0106,
+      "step": 13520
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0133467451377891e-05,
+      "loss": 1.0886,
+      "step": 13521
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.013211911516036e-05,
+      "loss": 1.1854,
+      "step": 13522
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0130770814127807e-05,
+      "loss": 1.109,
+      "step": 13523
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0129422548305637e-05,
+      "loss": 1.1399,
+      "step": 13524
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0128074317719246e-05,
+      "loss": 1.0431,
+      "step": 13525
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0126726122394027e-05,
+      "loss": 1.1144,
+      "step": 13526
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0125377962355373e-05,
+      "loss": 1.1066,
+      "step": 13527
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.012402983762869e-05,
+      "loss": 1.0863,
+      "step": 13528
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0122681748239363e-05,
+      "loss": 1.1357,
+      "step": 13529
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0121333694212789e-05,
+      "loss": 1.0343,
+      "step": 13530
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0119985675574356e-05,
+      "loss": 1.0186,
+      "step": 13531
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0118637692349467e-05,
+      "loss": 1.0984,
+      "step": 13532
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0117289744563506e-05,
+      "loss": 1.0073,
+      "step": 13533
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0115941832241866e-05,
+      "loss": 0.9835,
+      "step": 13534
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0114593955409934e-05,
+      "loss": 1.0477,
+      "step": 13535
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.011324611409311e-05,
+      "loss": 0.9319,
+      "step": 13536
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.011189830831677e-05,
+      "loss": 1.0006,
+      "step": 13537
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0110550538106309e-05,
+      "loss": 1.0412,
+      "step": 13538
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.010920280348712e-05,
+      "loss": 1.0161,
+      "step": 13539
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0107855104484579e-05,
+      "loss": 0.9598,
+      "step": 13540
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0106507441124078e-05,
+      "loss": 1.0274,
+      "step": 13541
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0105159813431002e-05,
+      "loss": 1.0613,
+      "step": 13542
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0103812221430742e-05,
+      "loss": 1.0589,
+      "step": 13543
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.010246466514867e-05,
+      "loss": 0.9834,
+      "step": 13544
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.010111714461018e-05,
+      "loss": 1.1042,
+      "step": 13545
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0099769659840648e-05,
+      "loss": 1.0337,
+      "step": 13546
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0098422210865467e-05,
+      "loss": 1.1408,
+      "step": 13547
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0097074797710006e-05,
+      "loss": 1.104,
+      "step": 13548
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0095727420399653e-05,
+      "loss": 1.0811,
+      "step": 13549
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.009438007895979e-05,
+      "loss": 1.0492,
+      "step": 13550
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0093032773415788e-05,
+      "loss": 1.0648,
+      "step": 13551
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0091685503793035e-05,
+      "loss": 1.0586,
+      "step": 13552
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0090338270116904e-05,
+      "loss": 1.016,
+      "step": 13553
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0088991072412779e-05,
+      "loss": 1.0993,
+      "step": 13554
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0087643910706029e-05,
+      "loss": 1.0898,
+      "step": 13555
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0086296785022034e-05,
+      "loss": 0.9662,
+      "step": 13556
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.008494969538617e-05,
+      "loss": 1.0564,
+      "step": 13557
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0083602641823813e-05,
+      "loss": 1.0811,
+      "step": 13558
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0082255624360332e-05,
+      "loss": 1.0834,
+      "step": 13559
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0080908643021107e-05,
+      "loss": 1.0514,
+      "step": 13560
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.007956169783151e-05,
+      "loss": 1.0433,
+      "step": 13561
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0078214788816906e-05,
+      "loss": 0.9883,
+      "step": 13562
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0076867916002673e-05,
+      "loss": 1.0423,
+      "step": 13563
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0075521079414182e-05,
+      "loss": 1.0667,
+      "step": 13564
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0074174279076803e-05,
+      "loss": 1.0617,
+      "step": 13565
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0072827515015902e-05,
+      "loss": 0.9181,
+      "step": 13566
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.007148078725685e-05,
+      "loss": 1.0554,
+      "step": 13567
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0070134095825015e-05,
+      "loss": 1.0777,
+      "step": 13568
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0068787440745768e-05,
+      "loss": 0.9749,
+      "step": 13569
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.006744082204447e-05,
+      "loss": 1.0626,
+      "step": 13570
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0066094239746486e-05,
+      "loss": 1.0506,
+      "step": 13571
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0064747693877189e-05,
+      "loss": 1.0687,
+      "step": 13572
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0063401184461938e-05,
+      "loss": 1.0931,
+      "step": 13573
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0062054711526098e-05,
+      "loss": 1.0025,
+      "step": 13574
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.006070827509503e-05,
+      "loss": 1.0693,
+      "step": 13575
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0059361875194105e-05,
+      "loss": 1.0674,
+      "step": 13576
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0058015511848675e-05,
+      "loss": 1.1381,
+      "step": 13577
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0056669185084107e-05,
+      "loss": 1.0708,
+      "step": 13578
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0055322894925756e-05,
+      "loss": 1.1165,
+      "step": 13579
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0053976641398992e-05,
+      "loss": 0.9972,
+      "step": 13580
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0052630424529166e-05,
+      "loss": 0.9553,
+      "step": 13581
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0051284244341635e-05,
+      "loss": 1.0515,
+      "step": 13582
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0049938100861767e-05,
+      "loss": 1.0567,
+      "step": 13583
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0048591994114905e-05,
+      "loss": 0.9966,
+      "step": 13584
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0047245924126416e-05,
+      "loss": 1.0334,
+      "step": 13585
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0045899890921652e-05,
+      "loss": 1.0411,
+      "step": 13586
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0044553894525973e-05,
+      "loss": 1.0194,
+      "step": 13587
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0043207934964721e-05,
+      "loss": 1.0221,
+      "step": 13588
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.004186201226326e-05,
+      "loss": 1.039,
+      "step": 13589
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0040516126446939e-05,
+      "loss": 1.124,
+      "step": 13590
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.003917027754112e-05,
+      "loss": 1.0265,
+      "step": 13591
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0037824465571137e-05,
+      "loss": 0.9843,
+      "step": 13592
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0036478690562353e-05,
+      "loss": 1.0325,
+      "step": 13593
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0035132952540117e-05,
+      "loss": 1.0842,
+      "step": 13594
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0033787251529773e-05,
+      "loss": 1.0314,
+      "step": 13595
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0032441587556675e-05,
+      "loss": 1.0626,
+      "step": 13596
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.003109596064617e-05,
+      "loss": 1.0724,
+      "step": 13597
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0029750370823606e-05,
+      "loss": 0.9915,
+      "step": 13598
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0028404818114326e-05,
+      "loss": 1.0698,
+      "step": 13599
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0027059302543677e-05,
+      "loss": 1.087,
+      "step": 13600
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0025713824137012e-05,
+      "loss": 1.0437,
+      "step": 13601
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0024368382919664e-05,
+      "loss": 1.0662,
+      "step": 13602
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0023022978916982e-05,
+      "loss": 1.0682,
+      "step": 13603
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.002167761215431e-05,
+      "loss": 1.0103,
+      "step": 13604
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0020332282656994e-05,
+      "loss": 1.1142,
+      "step": 13605
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0018986990450367e-05,
+      "loss": 1.0018,
+      "step": 13606
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0017641735559773e-05,
+      "loss": 1.0737,
+      "step": 13607
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0016296518010558e-05,
+      "loss": 1.1183,
+      "step": 13608
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0014951337828058e-05,
+      "loss": 0.9286,
+      "step": 13609
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.001360619503761e-05,
+      "loss": 1.0677,
+      "step": 13610
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.001226108966455e-05,
+      "loss": 1.0836,
+      "step": 13611
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0010916021734225e-05,
+      "loss": 1.014,
+      "step": 13612
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0009570991271965e-05,
+      "loss": 1.1318,
+      "step": 13613
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0008225998303106e-05,
+      "loss": 1.0564,
+      "step": 13614
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0006881042852983e-05,
+      "loss": 1.0076,
+      "step": 13615
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0005536124946936e-05,
+      "loss": 1.034,
+      "step": 13616
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0004191244610293e-05,
+      "loss": 1.0079,
+      "step": 13617
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0002846401868391e-05,
+      "loss": 1.0383,
+      "step": 13618
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.000150159674656e-05,
+      "loss": 0.9994,
+      "step": 13619
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0000156829270135e-05,
+      "loss": 1.0806,
+      "step": 13620
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.998812099464446e-06,
+      "loss": 1.0701,
+      "step": 13621
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.997467407354821e-06,
+      "loss": 1.1028,
+      "step": 13622
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.996122752966595e-06,
+      "loss": 1.0371,
+      "step": 13623
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.994778136325089e-06,
+      "loss": 0.9596,
+      "step": 13624
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.993433557455637e-06,
+      "loss": 1.029,
+      "step": 13625
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.992089016383564e-06,
+      "loss": 1.0021,
+      "step": 13626
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.990744513134206e-06,
+      "loss": 1.0424,
+      "step": 13627
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.989400047732874e-06,
+      "loss": 1.0905,
+      "step": 13628
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.988055620204905e-06,
+      "loss": 1.0422,
+      "step": 13629
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.986711230575616e-06,
+      "loss": 1.1268,
+      "step": 13630
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.985366878870344e-06,
+      "loss": 1.0518,
+      "step": 13631
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.984022565114395e-06,
+      "loss": 1.0546,
+      "step": 13632
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.982678289333102e-06,
+      "loss": 1.1434,
+      "step": 13633
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.981334051551787e-06,
+      "loss": 1.0124,
+      "step": 13634
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.979989851795769e-06,
+      "loss": 1.0093,
+      "step": 13635
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.978645690090364e-06,
+      "loss": 1.0437,
+      "step": 13636
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.977301566460901e-06,
+      "loss": 1.0631,
+      "step": 13637
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.975957480932695e-06,
+      "loss": 0.9287,
+      "step": 13638
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.97461343353106e-06,
+      "loss": 0.9952,
+      "step": 13639
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.973269424281319e-06,
+      "loss": 1.0098,
+      "step": 13640
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.971925453208789e-06,
+      "loss": 1.113,
+      "step": 13641
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.970581520338784e-06,
+      "loss": 1.0483,
+      "step": 13642
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.969237625696617e-06,
+      "loss": 1.0575,
+      "step": 13643
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.967893769307607e-06,
+      "loss": 1.0162,
+      "step": 13644
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.96654995119707e-06,
+      "loss": 0.9987,
+      "step": 13645
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.965206171390314e-06,
+      "loss": 1.0547,
+      "step": 13646
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.96386242991265e-06,
+      "loss": 1.1377,
+      "step": 13647
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.962518726789396e-06,
+      "loss": 1.028,
+      "step": 13648
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.961175062045863e-06,
+      "loss": 1.0369,
+      "step": 13649
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.959831435707357e-06,
+      "loss": 1.0744,
+      "step": 13650
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.958487847799188e-06,
+      "loss": 0.9859,
+      "step": 13651
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.957144298346666e-06,
+      "loss": 0.9722,
+      "step": 13652
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.955800787375104e-06,
+      "loss": 1.0075,
+      "step": 13653
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.954457314909804e-06,
+      "loss": 1.1212,
+      "step": 13654
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.95311388097607e-06,
+      "loss": 0.989,
+      "step": 13655
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.951770485599218e-06,
+      "loss": 1.0076,
+      "step": 13656
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.950427128804546e-06,
+      "loss": 0.9856,
+      "step": 13657
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.949083810617357e-06,
+      "loss": 0.975,
+      "step": 13658
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.947740531062957e-06,
+      "loss": 1.065,
+      "step": 13659
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.946397290166654e-06,
+      "loss": 0.9891,
+      "step": 13660
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.945054087953746e-06,
+      "loss": 1.0427,
+      "step": 13661
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.943710924449534e-06,
+      "loss": 1.0382,
+      "step": 13662
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.94236779967932e-06,
+      "loss": 0.9777,
+      "step": 13663
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.941024713668405e-06,
+      "loss": 0.9789,
+      "step": 13664
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.93968166644209e-06,
+      "loss": 1.0986,
+      "step": 13665
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.938338658025668e-06,
+      "loss": 1.0427,
+      "step": 13666
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.936995688444444e-06,
+      "loss": 0.9925,
+      "step": 13667
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.935652757723708e-06,
+      "loss": 0.9988,
+      "step": 13668
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.934309865888763e-06,
+      "loss": 1.0672,
+      "step": 13669
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.9329670129649e-06,
+      "loss": 1.048,
+      "step": 13670
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.931624198977423e-06,
+      "loss": 1.0154,
+      "step": 13671
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.930281423951615e-06,
+      "loss": 1.0767,
+      "step": 13672
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.928938687912775e-06,
+      "loss": 1.0318,
+      "step": 13673
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.927595990886194e-06,
+      "loss": 1.0825,
+      "step": 13674
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.92625333289717e-06,
+      "loss": 1.0409,
+      "step": 13675
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.924910713970986e-06,
+      "loss": 0.9629,
+      "step": 13676
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.92356813413294e-06,
+      "loss": 0.9778,
+      "step": 13677
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.922225593408318e-06,
+      "loss": 1.085,
+      "step": 13678
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.920883091822409e-06,
+      "loss": 1.0527,
+      "step": 13679
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.9195406294005e-06,
+      "loss": 1.1013,
+      "step": 13680
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.918198206167884e-06,
+      "loss": 1.0881,
+      "step": 13681
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.916855822149847e-06,
+      "loss": 0.9956,
+      "step": 13682
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.91551347737167e-06,
+      "loss": 1.0051,
+      "step": 13683
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.914171171858645e-06,
+      "loss": 0.9658,
+      "step": 13684
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.912828905636054e-06,
+      "loss": 1.1047,
+      "step": 13685
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.91148667872918e-06,
+      "loss": 1.1336,
+      "step": 13686
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.910144491163305e-06,
+      "loss": 1.0166,
+      "step": 13687
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.908802342963714e-06,
+      "loss": 1.0334,
+      "step": 13688
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.907460234155694e-06,
+      "loss": 1.0914,
+      "step": 13689
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.906118164764518e-06,
+      "loss": 1.0104,
+      "step": 13690
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.904776134815466e-06,
+      "loss": 0.9933,
+      "step": 13691
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.903434144333823e-06,
+      "loss": 1.0591,
+      "step": 13692
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.902092193344868e-06,
+      "loss": 1.0019,
+      "step": 13693
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.900750281873874e-06,
+      "loss": 0.9728,
+      "step": 13694
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.899408409946119e-06,
+      "loss": 1.014,
+      "step": 13695
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.898066577586889e-06,
+      "loss": 1.0773,
+      "step": 13696
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.896724784821446e-06,
+      "loss": 1.0096,
+      "step": 13697
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.895383031675074e-06,
+      "loss": 1.0528,
+      "step": 13698
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.894041318173044e-06,
+      "loss": 1.0442,
+      "step": 13699
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.892699644340633e-06,
+      "loss": 0.9496,
+      "step": 13700
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.89135801020311e-06,
+      "loss": 1.0298,
+      "step": 13701
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.89001641578575e-06,
+      "loss": 1.0548,
+      "step": 13702
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.888674861113821e-06,
+      "loss": 1.0466,
+      "step": 13703
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.887333346212599e-06,
+      "loss": 0.996,
+      "step": 13704
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.88599187110735e-06,
+      "loss": 0.9935,
+      "step": 13705
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.884650435823344e-06,
+      "loss": 1.0169,
+      "step": 13706
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.88330904038585e-06,
+      "loss": 1.0427,
+      "step": 13707
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.881967684820133e-06,
+      "loss": 0.9724,
+      "step": 13708
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.880626369151462e-06,
+      "loss": 1.0701,
+      "step": 13709
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.879285093405104e-06,
+      "loss": 1.0855,
+      "step": 13710
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.877943857606327e-06,
+      "loss": 1.024,
+      "step": 13711
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.876602661780386e-06,
+      "loss": 1.0661,
+      "step": 13712
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.875261505952554e-06,
+      "loss": 0.9864,
+      "step": 13713
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.873920390148089e-06,
+      "loss": 1.0926,
+      "step": 13714
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.872579314392264e-06,
+      "loss": 0.9707,
+      "step": 13715
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.871238278710322e-06,
+      "loss": 1.1052,
+      "step": 13716
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.86989728312754e-06,
+      "loss": 1.1083,
+      "step": 13717
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.868556327669172e-06,
+      "loss": 1.0198,
+      "step": 13718
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.867215412360476e-06,
+      "loss": 1.042,
+      "step": 13719
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.865874537226708e-06,
+      "loss": 1.0846,
+      "step": 13720
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.864533702293137e-06,
+      "loss": 0.9687,
+      "step": 13721
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.863192907585012e-06,
+      "loss": 0.9288,
+      "step": 13722
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.861852153127589e-06,
+      "loss": 0.9937,
+      "step": 13723
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.860511438946122e-06,
+      "loss": 1.1182,
+      "step": 13724
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.859170765065874e-06,
+      "loss": 1.0125,
+      "step": 13725
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.857830131512093e-06,
+      "loss": 1.0337,
+      "step": 13726
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.856489538310032e-06,
+      "loss": 1.0446,
+      "step": 13727
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.855148985484946e-06,
+      "loss": 0.9323,
+      "step": 13728
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.853808473062088e-06,
+      "loss": 1.0649,
+      "step": 13729
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.852468001066703e-06,
+      "loss": 1.045,
+      "step": 13730
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.851127569524044e-06,
+      "loss": 0.9751,
+      "step": 13731
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.849787178459362e-06,
+      "loss": 1.0102,
+      "step": 13732
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.848446827897908e-06,
+      "loss": 1.0067,
+      "step": 13733
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.847106517864926e-06,
+      "loss": 1.0958,
+      "step": 13734
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.845766248385663e-06,
+      "loss": 0.9989,
+      "step": 13735
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.844426019485367e-06,
+      "loss": 0.9867,
+      "step": 13736
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.843085831189287e-06,
+      "loss": 1.0081,
+      "step": 13737
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.841745683522662e-06,
+      "loss": 1.05,
+      "step": 13738
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.840405576510736e-06,
+      "loss": 1.0359,
+      "step": 13739
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.839065510178762e-06,
+      "loss": 0.9532,
+      "step": 13740
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.837725484551968e-06,
+      "loss": 1.0721,
+      "step": 13741
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.836385499655606e-06,
+      "loss": 0.9891,
+      "step": 13742
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.835045555514913e-06,
+      "loss": 1.0343,
+      "step": 13743
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.833705652155136e-06,
+      "loss": 1.0493,
+      "step": 13744
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.832365789601505e-06,
+      "loss": 1.007,
+      "step": 13745
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.831025967879265e-06,
+      "loss": 0.9623,
+      "step": 13746
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.82968618701365e-06,
+      "loss": 1.0948,
+      "step": 13747
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.828346447029903e-06,
+      "loss": 1.1216,
+      "step": 13748
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.827006747953254e-06,
+      "loss": 1.0431,
+      "step": 13749
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.825667089808943e-06,
+      "loss": 1.1135,
+      "step": 13750
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.824327472622207e-06,
+      "loss": 1.0104,
+      "step": 13751
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.822987896418268e-06,
+      "loss": 1.0644,
+      "step": 13752
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.821648361222373e-06,
+      "loss": 1.1345,
+      "step": 13753
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.82030886705975e-06,
+      "loss": 1.0997,
+      "step": 13754
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.818969413955632e-06,
+      "loss": 1.0256,
+      "step": 13755
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.817630001935246e-06,
+      "loss": 1.042,
+      "step": 13756
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.816290631023825e-06,
+      "loss": 1.0997,
+      "step": 13757
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.814951301246598e-06,
+      "loss": 1.0409,
+      "step": 13758
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.813612012628799e-06,
+      "loss": 0.9577,
+      "step": 13759
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.812272765195645e-06,
+      "loss": 0.9216,
+      "step": 13760
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.810933558972371e-06,
+      "loss": 0.9865,
+      "step": 13761
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.809594393984206e-06,
+      "loss": 1.0779,
+      "step": 13762
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.808255270256368e-06,
+      "loss": 0.9885,
+      "step": 13763
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.806916187814084e-06,
+      "loss": 1.042,
+      "step": 13764
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.80557714668258e-06,
+      "loss": 1.0427,
+      "step": 13765
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.80423814688708e-06,
+      "loss": 1.0753,
+      "step": 13766
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.802899188452806e-06,
+      "loss": 1.048,
+      "step": 13767
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.801560271404976e-06,
+      "loss": 1.0,
+      "step": 13768
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.800221395768818e-06,
+      "loss": 1.0445,
+      "step": 13769
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.798882561569545e-06,
+      "loss": 1.0828,
+      "step": 13770
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.797543768832379e-06,
+      "loss": 1.0166,
+      "step": 13771
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.79620501758254e-06,
+      "loss": 0.993,
+      "step": 13772
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.794866307845248e-06,
+      "loss": 1.0479,
+      "step": 13773
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.793527639645715e-06,
+      "loss": 1.0821,
+      "step": 13774
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.792189013009156e-06,
+      "loss": 1.0713,
+      "step": 13775
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.790850427960794e-06,
+      "loss": 0.9919,
+      "step": 13776
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.789511884525841e-06,
+      "loss": 1.0455,
+      "step": 13777
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.788173382729506e-06,
+      "loss": 1.0147,
+      "step": 13778
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.786834922597006e-06,
+      "loss": 0.9922,
+      "step": 13779
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.785496504153557e-06,
+      "loss": 1.0427,
+      "step": 13780
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.78415812742436e-06,
+      "loss": 1.0331,
+      "step": 13781
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.782819792434638e-06,
+      "loss": 1.0052,
+      "step": 13782
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.78148149920959e-06,
+      "loss": 1.0534,
+      "step": 13783
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.78014324777444e-06,
+      "loss": 1.0024,
+      "step": 13784
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.778805038154378e-06,
+      "loss": 1.0184,
+      "step": 13785
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.777466870374623e-06,
+      "loss": 1.0397,
+      "step": 13786
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.77612874446038e-06,
+      "loss": 0.9853,
+      "step": 13787
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.774790660436857e-06,
+      "loss": 0.9445,
+      "step": 13788
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.773452618329255e-06,
+      "loss": 0.9613,
+      "step": 13789
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.77211461816278e-06,
+      "loss": 1.0231,
+      "step": 13790
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.77077665996264e-06,
+      "loss": 1.0399,
+      "step": 13791
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.769438743754028e-06,
+      "loss": 1.0242,
+      "step": 13792
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.768100869562154e-06,
+      "loss": 1.0172,
+      "step": 13793
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.766763037412219e-06,
+      "loss": 1.0732,
+      "step": 13794
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.765425247329424e-06,
+      "loss": 1.0792,
+      "step": 13795
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.764087499338961e-06,
+      "loss": 1.0478,
+      "step": 13796
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.76274979346604e-06,
+      "loss": 1.054,
+      "step": 13797
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.761412129735853e-06,
+      "loss": 1.0752,
+      "step": 13798
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.760074508173597e-06,
+      "loss": 1.0471,
+      "step": 13799
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.758736928804468e-06,
+      "loss": 1.1453,
+      "step": 13800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.757399391653665e-06,
+      "loss": 0.9938,
+      "step": 13801
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.756061896746386e-06,
+      "loss": 0.9992,
+      "step": 13802
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.754724444107816e-06,
+      "loss": 1.0892,
+      "step": 13803
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.753387033763153e-06,
+      "loss": 1.0362,
+      "step": 13804
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.752049665737591e-06,
+      "loss": 1.0459,
+      "step": 13805
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.750712340056323e-06,
+      "loss": 1.089,
+      "step": 13806
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.749375056744535e-06,
+      "loss": 0.966,
+      "step": 13807
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.748037815827417e-06,
+      "loss": 1.034,
+      "step": 13808
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.746700617330163e-06,
+      "loss": 0.9234,
+      "step": 13809
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.745363461277963e-06,
+      "loss": 0.9835,
+      "step": 13810
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.744026347696e-06,
+      "loss": 1.0602,
+      "step": 13811
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.742689276609458e-06,
+      "loss": 1.0376,
+      "step": 13812
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.741352248043534e-06,
+      "loss": 1.0851,
+      "step": 13813
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.740015262023406e-06,
+      "loss": 1.0386,
+      "step": 13814
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.738678318574255e-06,
+      "loss": 1.122,
+      "step": 13815
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.737341417721274e-06,
+      "loss": 1.0877,
+      "step": 13816
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.736004559489642e-06,
+      "loss": 0.98,
+      "step": 13817
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.73466774390454e-06,
+      "loss": 1.0403,
+      "step": 13818
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.733330970991146e-06,
+      "loss": 1.0478,
+      "step": 13819
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.731994240774648e-06,
+      "loss": 1.0008,
+      "step": 13820
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.730657553280224e-06,
+      "loss": 1.0043,
+      "step": 13821
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.72932090853305e-06,
+      "loss": 1.0899,
+      "step": 13822
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.727984306558303e-06,
+      "loss": 1.0316,
+      "step": 13823
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.72664774738117e-06,
+      "loss": 1.0606,
+      "step": 13824
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.725311231026815e-06,
+      "loss": 1.0441,
+      "step": 13825
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.72397475752042e-06,
+      "loss": 0.9992,
+      "step": 13826
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.722638326887156e-06,
+      "loss": 1.0334,
+      "step": 13827
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.72130193915221e-06,
+      "loss": 1.045,
+      "step": 13828
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.719965594340736e-06,
+      "loss": 1.0489,
+      "step": 13829
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.71862929247792e-06,
+      "loss": 1.0368,
+      "step": 13830
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.717293033588927e-06,
+      "loss": 1.0379,
+      "step": 13831
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.715956817698937e-06,
+      "loss": 1.1242,
+      "step": 13832
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.71462064483311e-06,
+      "loss": 1.0734,
+      "step": 13833
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.71328451501662e-06,
+      "loss": 1.0648,
+      "step": 13834
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.711948428274636e-06,
+      "loss": 0.9358,
+      "step": 13835
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.710612384632321e-06,
+      "loss": 0.9923,
+      "step": 13836
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.709276384114847e-06,
+      "loss": 0.9913,
+      "step": 13837
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.707940426747378e-06,
+      "loss": 0.9729,
+      "step": 13838
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.706604512555082e-06,
+      "loss": 0.9673,
+      "step": 13839
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.705268641563114e-06,
+      "loss": 1.0518,
+      "step": 13840
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.70393281379665e-06,
+      "loss": 1.0436,
+      "step": 13841
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.702597029280848e-06,
+      "loss": 1.0791,
+      "step": 13842
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.701261288040867e-06,
+      "loss": 1.056,
+      "step": 13843
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.699925590101868e-06,
+      "loss": 1.0838,
+      "step": 13844
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.698589935489017e-06,
+      "loss": 1.0053,
+      "step": 13845
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.697254324227472e-06,
+      "loss": 1.0055,
+      "step": 13846
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.695918756342386e-06,
+      "loss": 1.0542,
+      "step": 13847
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.69458323185892e-06,
+      "loss": 0.9744,
+      "step": 13848
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.693247750802235e-06,
+      "loss": 1.0126,
+      "step": 13849
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.691912313197484e-06,
+      "loss": 1.0342,
+      "step": 13850
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.69057691906982e-06,
+      "loss": 1.0465,
+      "step": 13851
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.689241568444401e-06,
+      "loss": 1.0262,
+      "step": 13852
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.687906261346382e-06,
+      "loss": 0.9643,
+      "step": 13853
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.686570997800913e-06,
+      "loss": 1.0525,
+      "step": 13854
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.685235777833146e-06,
+      "loss": 1.0826,
+      "step": 13855
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.683900601468232e-06,
+      "loss": 1.0005,
+      "step": 13856
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.682565468731326e-06,
+      "loss": 0.9804,
+      "step": 13857
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.681230379647575e-06,
+      "loss": 1.0507,
+      "step": 13858
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.679895334242123e-06,
+      "loss": 1.0726,
+      "step": 13859
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.678560332540125e-06,
+      "loss": 1.0309,
+      "step": 13860
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.677225374566729e-06,
+      "loss": 0.9858,
+      "step": 13861
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.675890460347075e-06,
+      "loss": 0.9734,
+      "step": 13862
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.674555589906309e-06,
+      "loss": 1.085,
+      "step": 13863
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.673220763269585e-06,
+      "loss": 1.0543,
+      "step": 13864
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.671885980462035e-06,
+      "loss": 0.9971,
+      "step": 13865
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.670551241508808e-06,
+      "loss": 1.0782,
+      "step": 13866
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.669216546435045e-06,
+      "loss": 1.0277,
+      "step": 13867
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.667881895265894e-06,
+      "loss": 1.0991,
+      "step": 13868
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.666547288026484e-06,
+      "loss": 1.0126,
+      "step": 13869
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.66521272474196e-06,
+      "loss": 1.0165,
+      "step": 13870
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.663878205437464e-06,
+      "loss": 1.0693,
+      "step": 13871
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.662543730138135e-06,
+      "loss": 1.0584,
+      "step": 13872
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.6612092988691e-06,
+      "loss": 0.9868,
+      "step": 13873
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.659874911655506e-06,
+      "loss": 0.9227,
+      "step": 13874
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.658540568522487e-06,
+      "loss": 1.0469,
+      "step": 13875
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.657206269495171e-06,
+      "loss": 0.9849,
+      "step": 13876
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.655872014598701e-06,
+      "loss": 0.9667,
+      "step": 13877
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.654537803858204e-06,
+      "loss": 1.0672,
+      "step": 13878
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.653203637298819e-06,
+      "loss": 1.0901,
+      "step": 13879
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.651869514945666e-06,
+      "loss": 0.9842,
+      "step": 13880
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.650535436823884e-06,
+      "loss": 1.0405,
+      "step": 13881
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.649201402958603e-06,
+      "loss": 1.1336,
+      "step": 13882
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.64786741337495e-06,
+      "loss": 1.0404,
+      "step": 13883
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.646533468098054e-06,
+      "loss": 1.039,
+      "step": 13884
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.64519956715304e-06,
+      "loss": 1.0147,
+      "step": 13885
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.64386571056504e-06,
+      "loss": 1.0075,
+      "step": 13886
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.642531898359172e-06,
+      "loss": 1.0591,
+      "step": 13887
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.641198130560565e-06,
+      "loss": 1.0623,
+      "step": 13888
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.639864407194344e-06,
+      "loss": 1.005,
+      "step": 13889
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.638530728285633e-06,
+      "loss": 1.0661,
+      "step": 13890
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.63719709385955e-06,
+      "loss": 0.9989,
+      "step": 13891
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.635863503941216e-06,
+      "loss": 1.0309,
+      "step": 13892
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.634529958555756e-06,
+      "loss": 1.0464,
+      "step": 13893
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.63319645772829e-06,
+      "loss": 1.0477,
+      "step": 13894
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.631863001483934e-06,
+      "loss": 1.0694,
+      "step": 13895
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.630529589847806e-06,
+      "loss": 1.0391,
+      "step": 13896
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.629196222845027e-06,
+      "loss": 1.0533,
+      "step": 13897
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.627862900500709e-06,
+      "loss": 1.1069,
+      "step": 13898
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.626529622839968e-06,
+      "loss": 1.0509,
+      "step": 13899
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.62519638988792e-06,
+      "loss": 1.0248,
+      "step": 13900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.623863201669683e-06,
+      "loss": 1.0401,
+      "step": 13901
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.622530058210363e-06,
+      "loss": 1.0557,
+      "step": 13902
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.621196959535073e-06,
+      "loss": 1.0436,
+      "step": 13903
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.61986390566893e-06,
+      "loss": 0.9625,
+      "step": 13904
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.618530896637043e-06,
+      "loss": 0.9673,
+      "step": 13905
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.617197932464518e-06,
+      "loss": 1.0367,
+      "step": 13906
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.615865013176463e-06,
+      "loss": 0.9764,
+      "step": 13907
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.614532138797993e-06,
+      "loss": 0.9931,
+      "step": 13908
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.613199309354208e-06,
+      "loss": 1.0389,
+      "step": 13909
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.611866524870217e-06,
+      "loss": 1.0544,
+      "step": 13910
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.610533785371125e-06,
+      "loss": 1.0214,
+      "step": 13911
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.609201090882043e-06,
+      "loss": 1.0679,
+      "step": 13912
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.607868441428062e-06,
+      "loss": 1.0697,
+      "step": 13913
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.606535837034295e-06,
+      "loss": 1.0956,
+      "step": 13914
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.60520327772584e-06,
+      "loss": 1.0552,
+      "step": 13915
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.603870763527802e-06,
+      "loss": 1.0143,
+      "step": 13916
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.602538294465274e-06,
+      "loss": 1.0606,
+      "step": 13917
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.601205870563362e-06,
+      "loss": 1.0119,
+      "step": 13918
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.599873491847166e-06,
+      "loss": 1.0928,
+      "step": 13919
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.598541158341775e-06,
+      "loss": 1.0391,
+      "step": 13920
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.597208870072295e-06,
+      "loss": 1.0474,
+      "step": 13921
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.595876627063815e-06,
+      "loss": 1.1443,
+      "step": 13922
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.59454442934144e-06,
+      "loss": 1.0249,
+      "step": 13923
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.593212276930252e-06,
+      "loss": 1.0335,
+      "step": 13924
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.591880169855353e-06,
+      "loss": 1.0594,
+      "step": 13925
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.590548108141837e-06,
+      "loss": 1.0545,
+      "step": 13926
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.589216091814789e-06,
+      "loss": 1.0941,
+      "step": 13927
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.587884120899301e-06,
+      "loss": 1.1092,
+      "step": 13928
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.586552195420468e-06,
+      "loss": 1.0077,
+      "step": 13929
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.58522031540338e-06,
+      "loss": 1.1707,
+      "step": 13930
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.583888480873117e-06,
+      "loss": 0.9881,
+      "step": 13931
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.582556691854772e-06,
+      "loss": 0.966,
+      "step": 13932
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.581224948373432e-06,
+      "loss": 0.9893,
+      "step": 13933
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.579893250454185e-06,
+      "loss": 1.0529,
+      "step": 13934
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.578561598122112e-06,
+      "loss": 1.0373,
+      "step": 13935
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.577229991402295e-06,
+      "loss": 1.1158,
+      "step": 13936
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.575898430319826e-06,
+      "loss": 0.9711,
+      "step": 13937
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.574566914899778e-06,
+      "loss": 1.0955,
+      "step": 13938
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.573235445167238e-06,
+      "loss": 1.0356,
+      "step": 13939
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.571904021147284e-06,
+      "loss": 0.9371,
+      "step": 13940
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.570572642865e-06,
+      "loss": 1.0225,
+      "step": 13941
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.569241310345459e-06,
+      "loss": 1.0622,
+      "step": 13942
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.567910023613745e-06,
+      "loss": 1.0367,
+      "step": 13943
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.566578782694927e-06,
+      "loss": 1.0127,
+      "step": 13944
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.565247587614094e-06,
+      "loss": 1.0592,
+      "step": 13945
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.56391643839631e-06,
+      "loss": 1.0578,
+      "step": 13946
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.562585335066653e-06,
+      "loss": 1.1064,
+      "step": 13947
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.561254277650204e-06,
+      "loss": 1.0739,
+      "step": 13948
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.559923266172024e-06,
+      "loss": 1.073,
+      "step": 13949
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.55859230065719e-06,
+      "loss": 1.0159,
+      "step": 13950
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.557261381130775e-06,
+      "loss": 1.0314,
+      "step": 13951
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.555930507617853e-06,
+      "loss": 1.0964,
+      "step": 13952
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.554599680143483e-06,
+      "loss": 1.038,
+      "step": 13953
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.553268898732741e-06,
+      "loss": 1.1004,
+      "step": 13954
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.55193816341069e-06,
+      "loss": 1.0186,
+      "step": 13955
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.550607474202407e-06,
+      "loss": 1.1083,
+      "step": 13956
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.549276831132946e-06,
+      "loss": 1.0417,
+      "step": 13957
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.547946234227376e-06,
+      "loss": 1.0351,
+      "step": 13958
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.546615683510767e-06,
+      "loss": 1.121,
+      "step": 13959
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.545285179008174e-06,
+      "loss": 1.0911,
+      "step": 13960
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.543954720744661e-06,
+      "loss": 1.039,
+      "step": 13961
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.542624308745295e-06,
+      "loss": 1.0093,
+      "step": 13962
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.541293943035135e-06,
+      "loss": 1.0737,
+      "step": 13963
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.539963623639235e-06,
+      "loss": 0.9911,
+      "step": 13964
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.53863335058266e-06,
+      "loss": 1.0396,
+      "step": 13965
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.537303123890467e-06,
+      "loss": 1.067,
+      "step": 13966
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.535972943587715e-06,
+      "loss": 0.9887,
+      "step": 13967
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.534642809699455e-06,
+      "loss": 1.0395,
+      "step": 13968
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.533312722250746e-06,
+      "loss": 0.9843,
+      "step": 13969
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.531982681266645e-06,
+      "loss": 0.9575,
+      "step": 13970
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.530652686772202e-06,
+      "loss": 1.0249,
+      "step": 13971
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.529322738792468e-06,
+      "loss": 1.0415,
+      "step": 13972
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.527992837352501e-06,
+      "loss": 1.05,
+      "step": 13973
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.52666298247735e-06,
+      "loss": 1.0875,
+      "step": 13974
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.525333174192063e-06,
+      "loss": 1.099,
+      "step": 13975
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.524003412521692e-06,
+      "loss": 1.0345,
+      "step": 13976
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.522673697491283e-06,
+      "loss": 1.0119,
+      "step": 13977
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.521344029125887e-06,
+      "loss": 0.9489,
+      "step": 13978
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.520014407450548e-06,
+      "loss": 1.0783,
+      "step": 13979
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.51868483249031e-06,
+      "loss": 1.0803,
+      "step": 13980
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.517355304270225e-06,
+      "loss": 1.1136,
+      "step": 13981
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.51602582281533e-06,
+      "loss": 1.1147,
+      "step": 13982
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.514696388150673e-06,
+      "loss": 1.0947,
+      "step": 13983
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.51336700030129e-06,
+      "loss": 1.0693,
+      "step": 13984
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.512037659292232e-06,
+      "loss": 0.9931,
+      "step": 13985
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.510708365148533e-06,
+      "loss": 1.0833,
+      "step": 13986
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.509379117895235e-06,
+      "loss": 1.0238,
+      "step": 13987
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.508049917557371e-06,
+      "loss": 0.9896,
+      "step": 13988
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.50672076415999e-06,
+      "loss": 1.0255,
+      "step": 13989
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.50539165772812e-06,
+      "loss": 0.9567,
+      "step": 13990
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.5040625982868e-06,
+      "loss": 1.087,
+      "step": 13991
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.50273358586107e-06,
+      "loss": 1.0257,
+      "step": 13992
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.501404620475953e-06,
+      "loss": 0.9972,
+      "step": 13993
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.500075702156494e-06,
+      "loss": 1.0578,
+      "step": 13994
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.498746830927717e-06,
+      "loss": 1.0294,
+      "step": 13995
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.497418006814664e-06,
+      "loss": 1.1234,
+      "step": 13996
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.496089229842353e-06,
+      "loss": 1.0224,
+      "step": 13997
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.494760500035824e-06,
+      "loss": 1.0361,
+      "step": 13998
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.493431817420099e-06,
+      "loss": 1.0381,
+      "step": 13999
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.492103182020218e-06,
+      "loss": 0.9978,
+      "step": 14000
+    },
+    {
+      "epoch": 0.59,
+      "eval_text_loss": 0.781356155872345,
+      "eval_text_runtime": 1.3907,
+      "eval_text_samples_per_second": 53.209,
+      "eval_text_steps_per_second": 1.438,
+      "step": 14000
+    },
+    {
+      "epoch": 0.59,
+      "eval_svit_loss": 1.1167906522750854,
+      "eval_svit_runtime": 5.1921,
+      "eval_svit_samples_per_second": 131.16,
+      "eval_svit_steps_per_second": 2.119,
+      "step": 14000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.490774593861192e-06,
+      "loss": 1.0663,
+      "step": 14001
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.489446052968059e-06,
+      "loss": 1.0029,
+      "step": 14002
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.488117559365842e-06,
+      "loss": 1.07,
+      "step": 14003
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.486789113079563e-06,
+      "loss": 1.0647,
+      "step": 14004
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.485460714134245e-06,
+      "loss": 1.0733,
+      "step": 14005
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.484132362554915e-06,
+      "loss": 1.0139,
+      "step": 14006
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.482804058366597e-06,
+      "loss": 1.0223,
+      "step": 14007
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.481475801594302e-06,
+      "loss": 0.9768,
+      "step": 14008
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.480147592263061e-06,
+      "loss": 1.1018,
+      "step": 14009
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.478819430397888e-06,
+      "loss": 0.9246,
+      "step": 14010
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.477491316023802e-06,
+      "loss": 1.0813,
+      "step": 14011
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.476163249165817e-06,
+      "loss": 1.0106,
+      "step": 14012
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.474835229848957e-06,
+      "loss": 1.0382,
+      "step": 14013
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.473507258098235e-06,
+      "loss": 1.0393,
+      "step": 14014
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.472179333938663e-06,
+      "loss": 1.0543,
+      "step": 14015
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.470851457395254e-06,
+      "loss": 1.0087,
+      "step": 14016
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.469523628493025e-06,
+      "loss": 1.0635,
+      "step": 14017
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.468195847256989e-06,
+      "loss": 0.9792,
+      "step": 14018
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.466868113712154e-06,
+      "loss": 1.0684,
+      "step": 14019
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.465540427883529e-06,
+      "loss": 0.9873,
+      "step": 14020
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.464212789796128e-06,
+      "loss": 1.0203,
+      "step": 14021
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.462885199474956e-06,
+      "loss": 0.9958,
+      "step": 14022
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.461557656945022e-06,
+      "loss": 1.0405,
+      "step": 14023
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.46023016223133e-06,
+      "loss": 1.1013,
+      "step": 14024
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.458902715358893e-06,
+      "loss": 1.0575,
+      "step": 14025
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.457575316352707e-06,
+      "loss": 1.0629,
+      "step": 14026
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.456247965237782e-06,
+      "loss": 1.0806,
+      "step": 14027
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.454920662039118e-06,
+      "loss": 1.0103,
+      "step": 14028
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.453593406781721e-06,
+      "loss": 1.0233,
+      "step": 14029
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.452266199490587e-06,
+      "loss": 1.0614,
+      "step": 14030
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.45093904019072e-06,
+      "loss": 1.0062,
+      "step": 14031
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.449611928907123e-06,
+      "loss": 1.0367,
+      "step": 14032
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.448284865664784e-06,
+      "loss": 1.057,
+      "step": 14033
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.44695785048871e-06,
+      "loss": 0.9914,
+      "step": 14034
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.445630883403892e-06,
+      "loss": 0.9769,
+      "step": 14035
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.444303964435333e-06,
+      "loss": 1.0616,
+      "step": 14036
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.44297709360802e-06,
+      "loss": 1.0328,
+      "step": 14037
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.441650270946952e-06,
+      "loss": 1.0644,
+      "step": 14038
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.440323496477118e-06,
+      "loss": 1.0607,
+      "step": 14039
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.43899677022352e-06,
+      "loss": 0.9565,
+      "step": 14040
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.437670092211136e-06,
+      "loss": 1.1237,
+      "step": 14041
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.436343462464964e-06,
+      "loss": 1.0612,
+      "step": 14042
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.435016881009996e-06,
+      "loss": 1.0603,
+      "step": 14043
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.433690347871212e-06,
+      "loss": 1.021,
+      "step": 14044
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.432363863073605e-06,
+      "loss": 0.9971,
+      "step": 14045
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.431037426642162e-06,
+      "loss": 1.0241,
+      "step": 14046
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.429711038601872e-06,
+      "loss": 0.9658,
+      "step": 14047
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.428384698977713e-06,
+      "loss": 1.0446,
+      "step": 14048
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.427058407794671e-06,
+      "loss": 0.9844,
+      "step": 14049
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.425732165077732e-06,
+      "loss": 1.0645,
+      "step": 14050
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.424405970851879e-06,
+      "loss": 0.989,
+      "step": 14051
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.423079825142086e-06,
+      "loss": 0.9764,
+      "step": 14052
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.421753727973342e-06,
+      "loss": 1.0645,
+      "step": 14053
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.420427679370623e-06,
+      "loss": 0.9976,
+      "step": 14054
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.419101679358907e-06,
+      "loss": 1.0809,
+      "step": 14055
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.41777572796317e-06,
+      "loss": 1.0291,
+      "step": 14056
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.416449825208392e-06,
+      "loss": 1.089,
+      "step": 14057
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.415123971119549e-06,
+      "loss": 1.0935,
+      "step": 14058
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.413798165721614e-06,
+      "loss": 0.9876,
+      "step": 14059
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.41247240903956e-06,
+      "loss": 1.0314,
+      "step": 14060
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.411146701098362e-06,
+      "loss": 1.0853,
+      "step": 14061
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.409821041922993e-06,
+      "loss": 1.0464,
+      "step": 14062
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.408495431538422e-06,
+      "loss": 0.9467,
+      "step": 14063
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.407169869969619e-06,
+      "loss": 1.0887,
+      "step": 14064
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.405844357241559e-06,
+      "loss": 1.0437,
+      "step": 14065
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.404518893379204e-06,
+      "loss": 1.0922,
+      "step": 14066
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.403193478407525e-06,
+      "loss": 1.0642,
+      "step": 14067
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.401868112351481e-06,
+      "loss": 0.9652,
+      "step": 14068
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.400542795236052e-06,
+      "loss": 1.0154,
+      "step": 14069
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.399217527086192e-06,
+      "loss": 1.0266,
+      "step": 14070
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.397892307926868e-06,
+      "loss": 0.9657,
+      "step": 14071
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.39656713778304e-06,
+      "loss": 1.092,
+      "step": 14072
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.395242016679677e-06,
+      "loss": 1.0046,
+      "step": 14073
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.393916944641735e-06,
+      "loss": 1.0724,
+      "step": 14074
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.392591921694173e-06,
+      "loss": 1.1022,
+      "step": 14075
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.391266947861955e-06,
+      "loss": 1.0506,
+      "step": 14076
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.389942023170034e-06,
+      "loss": 0.998,
+      "step": 14077
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.38861714764337e-06,
+      "loss": 1.0023,
+      "step": 14078
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.387292321306918e-06,
+      "loss": 1.0152,
+      "step": 14079
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.385967544185642e-06,
+      "loss": 1.0602,
+      "step": 14080
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.384642816304484e-06,
+      "loss": 0.9752,
+      "step": 14081
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.383318137688404e-06,
+      "loss": 1.0597,
+      "step": 14082
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.381993508362351e-06,
+      "loss": 1.0425,
+      "step": 14083
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.380668928351289e-06,
+      "loss": 1.0375,
+      "step": 14084
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.379344397680152e-06,
+      "loss": 1.0586,
+      "step": 14085
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.3780199163739e-06,
+      "loss": 1.0423,
+      "step": 14086
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.376695484457483e-06,
+      "loss": 1.0173,
+      "step": 14087
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.375371101955842e-06,
+      "loss": 1.049,
+      "step": 14088
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.374046768893927e-06,
+      "loss": 1.0058,
+      "step": 14089
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.372722485296688e-06,
+      "loss": 0.9709,
+      "step": 14090
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.37139825118907e-06,
+      "loss": 1.0589,
+      "step": 14091
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.370074066596011e-06,
+      "loss": 1.0388,
+      "step": 14092
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.368749931542457e-06,
+      "loss": 1.0636,
+      "step": 14093
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.36742584605336e-06,
+      "loss": 1.0037,
+      "step": 14094
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.366101810153649e-06,
+      "loss": 1.0104,
+      "step": 14095
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.364777823868266e-06,
+      "loss": 1.0354,
+      "step": 14096
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.363453887222158e-06,
+      "loss": 1.0807,
+      "step": 14097
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.362130000240261e-06,
+      "loss": 1.0391,
+      "step": 14098
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.360806162947511e-06,
+      "loss": 1.0945,
+      "step": 14099
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.359482375368842e-06,
+      "loss": 0.9693,
+      "step": 14100
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.358158637529198e-06,
+      "loss": 1.0343,
+      "step": 14101
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.356834949453511e-06,
+      "loss": 1.0693,
+      "step": 14102
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.35551131116671e-06,
+      "loss": 0.9951,
+      "step": 14103
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.35418772269373e-06,
+      "loss": 1.0366,
+      "step": 14104
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.352864184059514e-06,
+      "loss": 1.0341,
+      "step": 14105
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.351540695288976e-06,
+      "loss": 1.0376,
+      "step": 14106
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.350217256407057e-06,
+      "loss": 1.0383,
+      "step": 14107
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.348893867438685e-06,
+      "loss": 1.0895,
+      "step": 14108
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.347570528408788e-06,
+      "loss": 1.0621,
+      "step": 14109
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.346247239342294e-06,
+      "loss": 0.9962,
+      "step": 14110
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.344924000264127e-06,
+      "loss": 1.0925,
+      "step": 14111
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.343600811199215e-06,
+      "loss": 1.0333,
+      "step": 14112
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.342277672172483e-06,
+      "loss": 1.0532,
+      "step": 14113
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.340954583208854e-06,
+      "loss": 0.9699,
+      "step": 14114
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.33963154433325e-06,
+      "loss": 1.0202,
+      "step": 14115
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.338308555570596e-06,
+      "loss": 0.9719,
+      "step": 14116
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.336985616945806e-06,
+      "loss": 1.1133,
+      "step": 14117
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.335662728483807e-06,
+      "loss": 1.0982,
+      "step": 14118
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.334339890209516e-06,
+      "loss": 0.9988,
+      "step": 14119
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.333017102147855e-06,
+      "loss": 1.0399,
+      "step": 14120
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.331694364323729e-06,
+      "loss": 0.998,
+      "step": 14121
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.330371676762066e-06,
+      "loss": 1.0437,
+      "step": 14122
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.329049039487774e-06,
+      "loss": 1.0124,
+      "step": 14123
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.327726452525778e-06,
+      "loss": 1.0174,
+      "step": 14124
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.326403915900979e-06,
+      "loss": 1.0279,
+      "step": 14125
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.325081429638294e-06,
+      "loss": 0.9593,
+      "step": 14126
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.323758993762639e-06,
+      "loss": 1.0114,
+      "step": 14127
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.322436608298918e-06,
+      "loss": 1.0803,
+      "step": 14128
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.32111427327204e-06,
+      "loss": 1.0053,
+      "step": 14129
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.31979198870692e-06,
+      "loss": 1.0877,
+      "step": 14130
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.318469754628464e-06,
+      "loss": 1.0808,
+      "step": 14131
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.317147571061574e-06,
+      "loss": 1.0649,
+      "step": 14132
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.315825438031157e-06,
+      "loss": 0.9712,
+      "step": 14133
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.314503355562124e-06,
+      "loss": 1.0077,
+      "step": 14134
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.313181323679374e-06,
+      "loss": 1.0648,
+      "step": 14135
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.311859342407808e-06,
+      "loss": 0.9846,
+      "step": 14136
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.31053741177233e-06,
+      "loss": 1.0099,
+      "step": 14137
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.309215531797843e-06,
+      "loss": 1.0296,
+      "step": 14138
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.307893702509247e-06,
+      "loss": 1.1206,
+      "step": 14139
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.306571923931435e-06,
+      "loss": 1.1046,
+      "step": 14140
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.305250196089314e-06,
+      "loss": 0.9947,
+      "step": 14141
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.303928519007776e-06,
+      "loss": 0.9543,
+      "step": 14142
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.302606892711716e-06,
+      "loss": 1.0389,
+      "step": 14143
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.301285317226032e-06,
+      "loss": 0.9762,
+      "step": 14144
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.299963792575618e-06,
+      "loss": 1.0222,
+      "step": 14145
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.298642318785367e-06,
+      "loss": 1.0291,
+      "step": 14146
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.297320895880171e-06,
+      "loss": 1.031,
+      "step": 14147
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.29599952388492e-06,
+      "loss": 0.9993,
+      "step": 14148
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.294678202824515e-06,
+      "loss": 1.0481,
+      "step": 14149
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.293356932723828e-06,
+      "loss": 1.092,
+      "step": 14150
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.292035713607758e-06,
+      "loss": 1.0452,
+      "step": 14151
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.290714545501189e-06,
+      "loss": 1.0802,
+      "step": 14152
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.289393428429014e-06,
+      "loss": 0.9851,
+      "step": 14153
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.288072362416111e-06,
+      "loss": 1.062,
+      "step": 14154
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.28675134748737e-06,
+      "loss": 1.0056,
+      "step": 14155
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.285430383667671e-06,
+      "loss": 1.0721,
+      "step": 14156
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.284109470981902e-06,
+      "loss": 1.0262,
+      "step": 14157
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.282788609454938e-06,
+      "loss": 1.0235,
+      "step": 14158
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.281467799111666e-06,
+      "loss": 1.0368,
+      "step": 14159
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.280147039976963e-06,
+      "loss": 1.0693,
+      "step": 14160
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.278826332075704e-06,
+      "loss": 0.9987,
+      "step": 14161
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.277505675432772e-06,
+      "loss": 1.0533,
+      "step": 14162
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.276185070073045e-06,
+      "loss": 1.0758,
+      "step": 14163
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.274864516021399e-06,
+      "loss": 1.0616,
+      "step": 14164
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.273544013302702e-06,
+      "loss": 1.03,
+      "step": 14165
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.272223561941837e-06,
+      "loss": 1.035,
+      "step": 14166
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.27090316196367e-06,
+      "loss": 1.0722,
+      "step": 14167
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.269582813393084e-06,
+      "loss": 1.0653,
+      "step": 14168
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.268262516254935e-06,
+      "loss": 0.9774,
+      "step": 14169
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.266942270574103e-06,
+      "loss": 1.0275,
+      "step": 14170
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.265622076375458e-06,
+      "loss": 1.0317,
+      "step": 14171
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.264301933683864e-06,
+      "loss": 1.0751,
+      "step": 14172
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.262981842524186e-06,
+      "loss": 1.0805,
+      "step": 14173
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.261661802921297e-06,
+      "loss": 1.0648,
+      "step": 14174
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.260341814900065e-06,
+      "loss": 0.9632,
+      "step": 14175
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.259021878485344e-06,
+      "loss": 1.0482,
+      "step": 14176
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.257701993702e-06,
+      "loss": 1.1372,
+      "step": 14177
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.256382160574901e-06,
+      "loss": 1.0511,
+      "step": 14178
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.255062379128906e-06,
+      "loss": 1.0393,
+      "step": 14179
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.253742649388874e-06,
+      "loss": 0.9832,
+      "step": 14180
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.252422971379664e-06,
+      "loss": 0.9852,
+      "step": 14181
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.251103345126138e-06,
+      "loss": 1.0912,
+      "step": 14182
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.249783770653151e-06,
+      "loss": 1.0068,
+      "step": 14183
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.248464247985558e-06,
+      "loss": 1.0109,
+      "step": 14184
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.24714477714822e-06,
+      "loss": 1.0105,
+      "step": 14185
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.24582535816599e-06,
+      "loss": 0.9704,
+      "step": 14186
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.244505991063718e-06,
+      "loss": 1.0811,
+      "step": 14187
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.243186675866255e-06,
+      "loss": 1.1602,
+      "step": 14188
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.241867412598466e-06,
+      "loss": 1.0099,
+      "step": 14189
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.240548201285186e-06,
+      "loss": 1.0009,
+      "step": 14190
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.239229041951272e-06,
+      "loss": 1.0316,
+      "step": 14191
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.23790993462157e-06,
+      "loss": 1.0142,
+      "step": 14192
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.236590879320937e-06,
+      "loss": 1.02,
+      "step": 14193
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.235271876074206e-06,
+      "loss": 0.9751,
+      "step": 14194
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.233952924906231e-06,
+      "loss": 1.0056,
+      "step": 14195
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.232634025841854e-06,
+      "loss": 0.9958,
+      "step": 14196
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.231315178905924e-06,
+      "loss": 1.0708,
+      "step": 14197
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.229996384123279e-06,
+      "loss": 1.0063,
+      "step": 14198
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.228677641518762e-06,
+      "loss": 1.0581,
+      "step": 14199
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.227358951117216e-06,
+      "loss": 1.0575,
+      "step": 14200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.226040312943474e-06,
+      "loss": 1.0215,
+      "step": 14201
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.224721727022384e-06,
+      "loss": 1.0293,
+      "step": 14202
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.22340319337878e-06,
+      "loss": 1.1335,
+      "step": 14203
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.222084712037499e-06,
+      "loss": 1.0637,
+      "step": 14204
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.220766283023374e-06,
+      "loss": 1.0755,
+      "step": 14205
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.219447906361246e-06,
+      "loss": 0.9444,
+      "step": 14206
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.218129582075946e-06,
+      "loss": 1.0441,
+      "step": 14207
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.216811310192308e-06,
+      "loss": 1.0115,
+      "step": 14208
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.215493090735161e-06,
+      "loss": 0.9834,
+      "step": 14209
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.21417492372934e-06,
+      "loss": 1.0539,
+      "step": 14210
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.212856809199676e-06,
+      "loss": 1.0378,
+      "step": 14211
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.211538747170992e-06,
+      "loss": 1.0637,
+      "step": 14212
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.210220737668119e-06,
+      "loss": 1.013,
+      "step": 14213
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.208902780715888e-06,
+      "loss": 1.0159,
+      "step": 14214
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.207584876339123e-06,
+      "loss": 1.01,
+      "step": 14215
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.206267024562646e-06,
+      "loss": 1.054,
+      "step": 14216
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.204949225411281e-06,
+      "loss": 0.9833,
+      "step": 14217
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.203631478909858e-06,
+      "loss": 1.0313,
+      "step": 14218
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.202313785083194e-06,
+      "loss": 1.0349,
+      "step": 14219
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.20099614395611e-06,
+      "loss": 1.0676,
+      "step": 14220
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.199678555553427e-06,
+      "loss": 1.0615,
+      "step": 14221
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.198361019899967e-06,
+      "loss": 0.987,
+      "step": 14222
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.197043537020543e-06,
+      "loss": 0.9819,
+      "step": 14223
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.195726106939975e-06,
+      "loss": 1.064,
+      "step": 14224
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.194408729683076e-06,
+      "loss": 1.0739,
+      "step": 14225
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.193091405274671e-06,
+      "loss": 1.0553,
+      "step": 14226
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.191774133739565e-06,
+      "loss": 0.9986,
+      "step": 14227
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.190456915102569e-06,
+      "loss": 1.0002,
+      "step": 14228
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.189139749388505e-06,
+      "loss": 0.9928,
+      "step": 14229
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.18782263662218e-06,
+      "loss": 0.9508,
+      "step": 14230
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.186505576828402e-06,
+      "loss": 1.0056,
+      "step": 14231
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.18518857003198e-06,
+      "loss": 0.9968,
+      "step": 14232
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.183871616257729e-06,
+      "loss": 1.028,
+      "step": 14233
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.182554715530445e-06,
+      "loss": 0.9816,
+      "step": 14234
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.181237867874944e-06,
+      "loss": 1.0565,
+      "step": 14235
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.179921073316025e-06,
+      "loss": 1.0191,
+      "step": 14236
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.178604331878503e-06,
+      "loss": 1.0403,
+      "step": 14237
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.177287643587164e-06,
+      "loss": 1.0218,
+      "step": 14238
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.175971008466823e-06,
+      "loss": 1.0645,
+      "step": 14239
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.174654426542275e-06,
+      "loss": 0.9751,
+      "step": 14240
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.173337897838328e-06,
+      "loss": 1.1004,
+      "step": 14241
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.172021422379776e-06,
+      "loss": 1.0257,
+      "step": 14242
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.170705000191416e-06,
+      "loss": 0.9359,
+      "step": 14243
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.169388631298048e-06,
+      "loss": 1.0238,
+      "step": 14244
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.168072315724463e-06,
+      "loss": 1.0367,
+      "step": 14245
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.166756053495463e-06,
+      "loss": 1.0031,
+      "step": 14246
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.16543984463584e-06,
+      "loss": 0.9867,
+      "step": 14247
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.164123689170387e-06,
+      "loss": 1.0342,
+      "step": 14248
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.162807587123893e-06,
+      "loss": 0.9384,
+      "step": 14249
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.161491538521155e-06,
+      "loss": 0.9902,
+      "step": 14250
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.160175543386958e-06,
+      "loss": 1.0358,
+      "step": 14251
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.158859601746098e-06,
+      "loss": 1.0508,
+      "step": 14252
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.157543713623353e-06,
+      "loss": 1.1032,
+      "step": 14253
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.156227879043518e-06,
+      "loss": 0.9571,
+      "step": 14254
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.15491209803138e-06,
+      "loss": 1.0878,
+      "step": 14255
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.15359637061172e-06,
+      "loss": 1.03,
+      "step": 14256
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.152280696809319e-06,
+      "loss": 1.0111,
+      "step": 14257
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.150965076648967e-06,
+      "loss": 1.0636,
+      "step": 14258
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.149649510155447e-06,
+      "loss": 1.0171,
+      "step": 14259
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.148333997353533e-06,
+      "loss": 1.0588,
+      "step": 14260
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.14701853826801e-06,
+      "loss": 0.9924,
+      "step": 14261
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.145703132923657e-06,
+      "loss": 0.9935,
+      "step": 14262
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.14438778134525e-06,
+      "loss": 0.9756,
+      "step": 14263
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.143072483557566e-06,
+      "loss": 0.9812,
+      "step": 14264
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.141757239585383e-06,
+      "loss": 0.9511,
+      "step": 14265
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.140442049453478e-06,
+      "loss": 1.0531,
+      "step": 14266
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.139126913186618e-06,
+      "loss": 1.0547,
+      "step": 14267
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.137811830809584e-06,
+      "loss": 1.027,
+      "step": 14268
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.136496802347138e-06,
+      "loss": 1.0688,
+      "step": 14269
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.135181827824064e-06,
+      "loss": 0.9493,
+      "step": 14270
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.133866907265124e-06,
+      "loss": 1.0357,
+      "step": 14271
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.132552040695086e-06,
+      "loss": 1.006,
+      "step": 14272
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.131237228138725e-06,
+      "loss": 1.0093,
+      "step": 14273
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.129922469620798e-06,
+      "loss": 1.024,
+      "step": 14274
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.128607765166078e-06,
+      "loss": 1.0498,
+      "step": 14275
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.127293114799325e-06,
+      "loss": 1.0005,
+      "step": 14276
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.125978518545312e-06,
+      "loss": 0.982,
+      "step": 14277
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.12466397642879e-06,
+      "loss": 1.0529,
+      "step": 14278
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.123349488474526e-06,
+      "loss": 1.0606,
+      "step": 14279
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.122035054707282e-06,
+      "loss": 1.0653,
+      "step": 14280
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.120720675151823e-06,
+      "loss": 1.0332,
+      "step": 14281
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.119406349832895e-06,
+      "loss": 1.0664,
+      "step": 14282
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.118092078775265e-06,
+      "loss": 0.9513,
+      "step": 14283
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.116777862003688e-06,
+      "loss": 1.0624,
+      "step": 14284
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.115463699542917e-06,
+      "loss": 1.047,
+      "step": 14285
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.114149591417708e-06,
+      "loss": 1.049,
+      "step": 14286
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.112835537652817e-06,
+      "loss": 1.0083,
+      "step": 14287
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.111521538272997e-06,
+      "loss": 1.0959,
+      "step": 14288
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.110207593302992e-06,
+      "loss": 0.9963,
+      "step": 14289
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.10889370276756e-06,
+      "loss": 1.0496,
+      "step": 14290
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.10757986669145e-06,
+      "loss": 1.0393,
+      "step": 14291
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.106266085099411e-06,
+      "loss": 1.0642,
+      "step": 14292
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.104952358016184e-06,
+      "loss": 0.9886,
+      "step": 14293
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.103638685466521e-06,
+      "loss": 1.0693,
+      "step": 14294
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.10232506747517e-06,
+      "loss": 1.0577,
+      "step": 14295
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.10101150406687e-06,
+      "loss": 0.9813,
+      "step": 14296
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.099697995266365e-06,
+      "loss": 1.0383,
+      "step": 14297
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.0983845410984e-06,
+      "loss": 1.0835,
+      "step": 14298
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.097071141587717e-06,
+      "loss": 1.0997,
+      "step": 14299
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.095757796759053e-06,
+      "loss": 1.0574,
+      "step": 14300
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.094444506637144e-06,
+      "loss": 1.0236,
+      "step": 14301
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.093131271246737e-06,
+      "loss": 0.9697,
+      "step": 14302
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.091818090612567e-06,
+      "loss": 1.0501,
+      "step": 14303
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.090504964759366e-06,
+      "loss": 1.0222,
+      "step": 14304
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.089191893711869e-06,
+      "loss": 1.0348,
+      "step": 14305
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.087878877494817e-06,
+      "loss": 1.0284,
+      "step": 14306
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.086565916132938e-06,
+      "loss": 1.0239,
+      "step": 14307
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.085253009650962e-06,
+      "loss": 1.0322,
+      "step": 14308
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.083940158073621e-06,
+      "loss": 0.9418,
+      "step": 14309
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.082627361425652e-06,
+      "loss": 0.9926,
+      "step": 14310
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.081314619731777e-06,
+      "loss": 1.0692,
+      "step": 14311
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.080001933016724e-06,
+      "loss": 1.0175,
+      "step": 14312
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.07868930130522e-06,
+      "loss": 1.0001,
+      "step": 14313
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.077376724621995e-06,
+      "loss": 1.0086,
+      "step": 14314
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.07606420299177e-06,
+      "loss": 1.0333,
+      "step": 14315
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.07475173643927e-06,
+      "loss": 1.0641,
+      "step": 14316
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.07343932498922e-06,
+      "loss": 1.0096,
+      "step": 14317
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.072126968666335e-06,
+      "loss": 0.9759,
+      "step": 14318
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.070814667495342e-06,
+      "loss": 0.9991,
+      "step": 14319
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.069502421500956e-06,
+      "loss": 1.0078,
+      "step": 14320
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.068190230707904e-06,
+      "loss": 0.93,
+      "step": 14321
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.066878095140891e-06,
+      "loss": 1.0319,
+      "step": 14322
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.065566014824642e-06,
+      "loss": 1.0696,
+      "step": 14323
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.06425398978387e-06,
+      "loss": 1.0424,
+      "step": 14324
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.062942020043296e-06,
+      "loss": 0.969,
+      "step": 14325
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.06163010562762e-06,
+      "loss": 1.03,
+      "step": 14326
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.060318246561563e-06,
+      "loss": 1.0608,
+      "step": 14327
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.05900644286984e-06,
+      "loss": 0.9954,
+      "step": 14328
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.057694694577149e-06,
+      "loss": 1.0566,
+      "step": 14329
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.05638300170821e-06,
+      "loss": 1.0273,
+      "step": 14330
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.055071364287727e-06,
+      "loss": 1.0761,
+      "step": 14331
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.05375978234041e-06,
+      "loss": 0.9865,
+      "step": 14332
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.052448255890959e-06,
+      "loss": 0.9786,
+      "step": 14333
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.051136784964083e-06,
+      "loss": 0.9674,
+      "step": 14334
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.04982536958449e-06,
+      "loss": 0.959,
+      "step": 14335
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.048514009776874e-06,
+      "loss": 1.0159,
+      "step": 14336
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.047202705565942e-06,
+      "loss": 1.0793,
+      "step": 14337
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.045891456976394e-06,
+      "loss": 1.0092,
+      "step": 14338
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.044580264032935e-06,
+      "loss": 1.0624,
+      "step": 14339
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.043269126760255e-06,
+      "loss": 1.0711,
+      "step": 14340
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.041958045183054e-06,
+      "loss": 1.0381,
+      "step": 14341
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.040647019326032e-06,
+      "loss": 1.045,
+      "step": 14342
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.039336049213885e-06,
+      "loss": 1.0268,
+      "step": 14343
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.038025134871302e-06,
+      "loss": 1.0214,
+      "step": 14344
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.03671427632298e-06,
+      "loss": 0.9738,
+      "step": 14345
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.035403473593614e-06,
+      "loss": 0.9986,
+      "step": 14346
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.03409272670789e-06,
+      "loss": 1.0239,
+      "step": 14347
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.032782035690502e-06,
+      "loss": 1.1109,
+      "step": 14348
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.031471400566137e-06,
+      "loss": 1.0287,
+      "step": 14349
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.030160821359486e-06,
+      "loss": 1.126,
+      "step": 14350
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.028850298095235e-06,
+      "loss": 0.9566,
+      "step": 14351
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.02753983079807e-06,
+      "loss": 0.9872,
+      "step": 14352
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.026229419492672e-06,
+      "loss": 0.9607,
+      "step": 14353
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.024919064203733e-06,
+      "loss": 1.055,
+      "step": 14354
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.023608764955932e-06,
+      "loss": 1.0002,
+      "step": 14355
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.022298521773948e-06,
+      "loss": 1.0817,
+      "step": 14356
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.020988334682468e-06,
+      "loss": 1.0009,
+      "step": 14357
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.019678203706165e-06,
+      "loss": 1.1197,
+      "step": 14358
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.018368128869721e-06,
+      "loss": 0.9839,
+      "step": 14359
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.017058110197815e-06,
+      "loss": 1.0054,
+      "step": 14360
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.015748147715126e-06,
+      "loss": 0.9503,
+      "step": 14361
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.01443824144632e-06,
+      "loss": 1.0144,
+      "step": 14362
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.01312839141608e-06,
+      "loss": 1.0505,
+      "step": 14363
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.011818597649073e-06,
+      "loss": 1.0937,
+      "step": 14364
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.010508860169984e-06,
+      "loss": 1.016,
+      "step": 14365
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.00919917900347e-06,
+      "loss": 1.0307,
+      "step": 14366
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.007889554174208e-06,
+      "loss": 1.0713,
+      "step": 14367
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.006579985706868e-06,
+      "loss": 1.0111,
+      "step": 14368
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.005270473626112e-06,
+      "loss": 1.055,
+      "step": 14369
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.003961017956611e-06,
+      "loss": 1.0199,
+      "step": 14370
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.002651618723033e-06,
+      "loss": 1.09,
+      "step": 14371
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.001342275950046e-06,
+      "loss": 1.0331,
+      "step": 14372
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.0000329896623e-06,
+      "loss": 1.0671,
+      "step": 14373
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.998723759884472e-06,
+      "loss": 1.0175,
+      "step": 14374
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.997414586641219e-06,
+      "loss": 0.9408,
+      "step": 14375
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.996105469957204e-06,
+      "loss": 1.0831,
+      "step": 14376
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.994796409857078e-06,
+      "loss": 1.0147,
+      "step": 14377
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.993487406365509e-06,
+      "loss": 1.0288,
+      "step": 14378
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.992178459507152e-06,
+      "loss": 0.9779,
+      "step": 14379
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.990869569306662e-06,
+      "loss": 0.9607,
+      "step": 14380
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.98956073578869e-06,
+      "loss": 1.0698,
+      "step": 14381
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.9882519589779e-06,
+      "loss": 0.9983,
+      "step": 14382
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.986943238898942e-06,
+      "loss": 1.0763,
+      "step": 14383
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.985634575576462e-06,
+      "loss": 1.0267,
+      "step": 14384
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.984325969035117e-06,
+      "loss": 0.9738,
+      "step": 14385
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.983017419299557e-06,
+      "loss": 1.0571,
+      "step": 14386
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.98170892639443e-06,
+      "loss": 0.9778,
+      "step": 14387
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.980400490344383e-06,
+      "loss": 1.0336,
+      "step": 14388
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.979092111174062e-06,
+      "loss": 1.0339,
+      "step": 14389
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.977783788908116e-06,
+      "loss": 1.078,
+      "step": 14390
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.976475523571186e-06,
+      "loss": 1.0414,
+      "step": 14391
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.975167315187919e-06,
+      "loss": 0.9701,
+      "step": 14392
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.973859163782951e-06,
+      "loss": 1.0301,
+      "step": 14393
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.972551069380935e-06,
+      "loss": 0.9571,
+      "step": 14394
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.971243032006501e-06,
+      "loss": 1.067,
+      "step": 14395
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.969935051684293e-06,
+      "loss": 1.0461,
+      "step": 14396
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.968627128438944e-06,
+      "loss": 0.9517,
+      "step": 14397
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.9673192622951e-06,
+      "loss": 1.0511,
+      "step": 14398
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.966011453277393e-06,
+      "loss": 0.9967,
+      "step": 14399
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.964703701410456e-06,
+      "loss": 1.0243,
+      "step": 14400
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.963396006718925e-06,
+      "loss": 1.0635,
+      "step": 14401
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.962088369227429e-06,
+      "loss": 1.0305,
+      "step": 14402
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.960780788960603e-06,
+      "loss": 1.0471,
+      "step": 14403
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.959473265943079e-06,
+      "loss": 1.0395,
+      "step": 14404
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.95816580019949e-06,
+      "loss": 0.9924,
+      "step": 14405
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.956858391754453e-06,
+      "loss": 1.0336,
+      "step": 14406
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.955551040632605e-06,
+      "loss": 0.9928,
+      "step": 14407
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.954243746858566e-06,
+      "loss": 0.9839,
+      "step": 14408
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.95293651045697e-06,
+      "loss": 0.989,
+      "step": 14409
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.951629331452432e-06,
+      "loss": 0.9879,
+      "step": 14410
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.950322209869582e-06,
+      "loss": 1.0394,
+      "step": 14411
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.94901514573304e-06,
+      "loss": 1.047,
+      "step": 14412
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.947708139067424e-06,
+      "loss": 1.0322,
+      "step": 14413
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.946401189897354e-06,
+      "loss": 0.9625,
+      "step": 14414
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.945094298247453e-06,
+      "loss": 1.0258,
+      "step": 14415
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.943787464142339e-06,
+      "loss": 1.0272,
+      "step": 14416
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.942480687606622e-06,
+      "loss": 1.0308,
+      "step": 14417
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.941173968664923e-06,
+      "loss": 0.9854,
+      "step": 14418
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.939867307341858e-06,
+      "loss": 0.9975,
+      "step": 14419
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.938560703662036e-06,
+      "loss": 1.0336,
+      "step": 14420
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.937254157650068e-06,
+      "loss": 1.014,
+      "step": 14421
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.93594766933057e-06,
+      "loss": 0.9588,
+      "step": 14422
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.934641238728152e-06,
+      "loss": 1.0169,
+      "step": 14423
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.93333486586742e-06,
+      "loss": 0.9649,
+      "step": 14424
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.93202855077298e-06,
+      "loss": 1.0261,
+      "step": 14425
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.930722293469444e-06,
+      "loss": 0.9742,
+      "step": 14426
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.929416093981421e-06,
+      "loss": 0.9466,
+      "step": 14427
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.928109952333506e-06,
+      "loss": 1.0902,
+      "step": 14428
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.926803868550305e-06,
+      "loss": 1.0269,
+      "step": 14429
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.925497842656426e-06,
+      "loss": 1.0088,
+      "step": 14430
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.924191874676466e-06,
+      "loss": 0.9301,
+      "step": 14431
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.922885964635025e-06,
+      "loss": 1.0415,
+      "step": 14432
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.921580112556702e-06,
+      "loss": 1.0262,
+      "step": 14433
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.920274318466103e-06,
+      "loss": 1.0293,
+      "step": 14434
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.918968582387813e-06,
+      "loss": 0.9285,
+      "step": 14435
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.917662904346436e-06,
+      "loss": 1.027,
+      "step": 14436
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.916357284366562e-06,
+      "loss": 1.0868,
+      "step": 14437
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.91505172247279e-06,
+      "loss": 0.9954,
+      "step": 14438
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.913746218689709e-06,
+      "loss": 1.0137,
+      "step": 14439
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.912440773041911e-06,
+      "loss": 1.063,
+      "step": 14440
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.91113538555399e-06,
+      "loss": 0.9801,
+      "step": 14441
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.909830056250527e-06,
+      "loss": 0.9765,
+      "step": 14442
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.908524785156118e-06,
+      "loss": 0.9747,
+      "step": 14443
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.907219572295348e-06,
+      "loss": 1.0665,
+      "step": 14444
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.905914417692804e-06,
+      "loss": 0.9269,
+      "step": 14445
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.904609321373068e-06,
+      "loss": 1.0942,
+      "step": 14446
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.903304283360726e-06,
+      "loss": 0.9952,
+      "step": 14447
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.901999303680359e-06,
+      "loss": 1.0557,
+      "step": 14448
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.900694382356557e-06,
+      "loss": 1.0553,
+      "step": 14449
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.899389519413886e-06,
+      "loss": 1.0181,
+      "step": 14450
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.898084714876937e-06,
+      "loss": 1.0449,
+      "step": 14451
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.896779968770286e-06,
+      "loss": 1.0012,
+      "step": 14452
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.895475281118509e-06,
+      "loss": 1.087,
+      "step": 14453
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.894170651946179e-06,
+      "loss": 1.0016,
+      "step": 14454
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.892866081277876e-06,
+      "loss": 1.037,
+      "step": 14455
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.891561569138174e-06,
+      "loss": 1.0766,
+      "step": 14456
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.890257115551641e-06,
+      "loss": 0.9815,
+      "step": 14457
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.888952720542854e-06,
+      "loss": 0.9938,
+      "step": 14458
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.887648384136381e-06,
+      "loss": 0.9915,
+      "step": 14459
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.886344106356794e-06,
+      "loss": 1.0241,
+      "step": 14460
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.885039887228657e-06,
+      "loss": 1.0081,
+      "step": 14461
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.88373572677654e-06,
+      "loss": 1.0528,
+      "step": 14462
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.882431625025015e-06,
+      "loss": 0.9775,
+      "step": 14463
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.88112758199864e-06,
+      "loss": 1.0549,
+      "step": 14464
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.879823597721974e-06,
+      "loss": 1.0068,
+      "step": 14465
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.878519672219592e-06,
+      "loss": 1.0068,
+      "step": 14466
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.87721580551605e-06,
+      "loss": 0.9352,
+      "step": 14467
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.875911997635911e-06,
+      "loss": 1.0577,
+      "step": 14468
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.874608248603728e-06,
+      "loss": 0.9719,
+      "step": 14469
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.873304558444066e-06,
+      "loss": 1.0453,
+      "step": 14470
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.872000927181484e-06,
+      "loss": 1.0651,
+      "step": 14471
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.870697354840534e-06,
+      "loss": 1.0595,
+      "step": 14472
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.869393841445768e-06,
+      "loss": 1.0101,
+      "step": 14473
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.868090387021749e-06,
+      "loss": 1.058,
+      "step": 14474
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.866786991593024e-06,
+      "loss": 1.0105,
+      "step": 14475
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.865483655184146e-06,
+      "loss": 0.9133,
+      "step": 14476
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.864180377819664e-06,
+      "loss": 0.9762,
+      "step": 14477
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.862877159524132e-06,
+      "loss": 1.0145,
+      "step": 14478
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.861574000322095e-06,
+      "loss": 1.0225,
+      "step": 14479
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.860270900238102e-06,
+      "loss": 0.9689,
+      "step": 14480
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.858967859296696e-06,
+      "loss": 0.9914,
+      "step": 14481
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.85766487752243e-06,
+      "loss": 1.026,
+      "step": 14482
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.856361954939841e-06,
+      "loss": 1.0542,
+      "step": 14483
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.855059091573473e-06,
+      "loss": 0.9817,
+      "step": 14484
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.85375628744787e-06,
+      "loss": 0.9557,
+      "step": 14485
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.85245354258757e-06,
+      "loss": 1.0422,
+      "step": 14486
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.851150857017112e-06,
+      "loss": 1.0521,
+      "step": 14487
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.84984823076104e-06,
+      "loss": 0.9807,
+      "step": 14488
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.848545663843888e-06,
+      "loss": 1.0546,
+      "step": 14489
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.847243156290191e-06,
+      "loss": 1.0777,
+      "step": 14490
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.845940708124486e-06,
+      "loss": 1.0141,
+      "step": 14491
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.844638319371303e-06,
+      "loss": 1.0469,
+      "step": 14492
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.843335990055185e-06,
+      "loss": 1.0641,
+      "step": 14493
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.84203372020065e-06,
+      "loss": 1.0314,
+      "step": 14494
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.84073150983224e-06,
+      "loss": 1.0496,
+      "step": 14495
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.83942935897448e-06,
+      "loss": 1.0744,
+      "step": 14496
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.838127267651898e-06,
+      "loss": 0.9573,
+      "step": 14497
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.83682523588902e-06,
+      "loss": 0.9901,
+      "step": 14498
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.835523263710376e-06,
+      "loss": 0.994,
+      "step": 14499
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.834221351140492e-06,
+      "loss": 0.9927,
+      "step": 14500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.832919498203885e-06,
+      "loss": 1.0957,
+      "step": 14501
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.83161770492508e-06,
+      "loss": 1.0243,
+      "step": 14502
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.830315971328605e-06,
+      "loss": 1.091,
+      "step": 14503
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.829014297438973e-06,
+      "loss": 1.0355,
+      "step": 14504
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.827712683280707e-06,
+      "loss": 0.9179,
+      "step": 14505
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.826411128878325e-06,
+      "loss": 0.9773,
+      "step": 14506
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.825109634256347e-06,
+      "loss": 1.0107,
+      "step": 14507
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.823808199439283e-06,
+      "loss": 1.0172,
+      "step": 14508
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.82250682445165e-06,
+      "loss": 0.9395,
+      "step": 14509
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.821205509317965e-06,
+      "loss": 1.0245,
+      "step": 14510
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.81990425406274e-06,
+      "loss": 1.0762,
+      "step": 14511
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.818603058710482e-06,
+      "loss": 0.993,
+      "step": 14512
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.817301923285705e-06,
+      "loss": 1.0208,
+      "step": 14513
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.81600084781292e-06,
+      "loss": 1.0752,
+      "step": 14514
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.814699832316628e-06,
+      "loss": 1.0449,
+      "step": 14515
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.813398876821345e-06,
+      "loss": 0.9907,
+      "step": 14516
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.812097981351568e-06,
+      "loss": 0.9843,
+      "step": 14517
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.810797145931814e-06,
+      "loss": 1.0999,
+      "step": 14518
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.809496370586574e-06,
+      "loss": 1.0842,
+      "step": 14519
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.808195655340356e-06,
+      "loss": 1.1434,
+      "step": 14520
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.80689500021766e-06,
+      "loss": 0.9877,
+      "step": 14521
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.805594405242993e-06,
+      "loss": 0.956,
+      "step": 14522
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.804293870440844e-06,
+      "loss": 1.0479,
+      "step": 14523
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.802993395835716e-06,
+      "loss": 1.0075,
+      "step": 14524
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.801692981452106e-06,
+      "loss": 0.9973,
+      "step": 14525
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.800392627314506e-06,
+      "loss": 1.1314,
+      "step": 14526
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.799092333447415e-06,
+      "loss": 0.9801,
+      "step": 14527
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.797792099875326e-06,
+      "loss": 0.9801,
+      "step": 14528
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.796491926622731e-06,
+      "loss": 1.0187,
+      "step": 14529
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.795191813714117e-06,
+      "loss": 1.0828,
+      "step": 14530
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.793891761173979e-06,
+      "loss": 0.9852,
+      "step": 14531
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.792591769026803e-06,
+      "loss": 1.0514,
+      "step": 14532
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.791291837297083e-06,
+      "loss": 0.9894,
+      "step": 14533
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.789991966009295e-06,
+      "loss": 0.977,
+      "step": 14534
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.788692155187933e-06,
+      "loss": 1.0608,
+      "step": 14535
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.78739240485748e-06,
+      "loss": 0.957,
+      "step": 14536
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.786092715042416e-06,
+      "loss": 1.0296,
+      "step": 14537
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.78479308576722e-06,
+      "loss": 0.9943,
+      "step": 14538
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.783493517056383e-06,
+      "loss": 0.9937,
+      "step": 14539
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.782194008934379e-06,
+      "loss": 1.0014,
+      "step": 14540
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.780894561425687e-06,
+      "loss": 0.9848,
+      "step": 14541
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.779595174554782e-06,
+      "loss": 1.0319,
+      "step": 14542
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.778295848346143e-06,
+      "loss": 1.0006,
+      "step": 14543
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.77699658282425e-06,
+      "loss": 0.9868,
+      "step": 14544
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.775697378013569e-06,
+      "loss": 0.9649,
+      "step": 14545
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.774398233938572e-06,
+      "loss": 1.0476,
+      "step": 14546
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.773099150623742e-06,
+      "loss": 1.0371,
+      "step": 14547
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.771800128093538e-06,
+      "loss": 1.0403,
+      "step": 14548
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.770501166372434e-06,
+      "loss": 1.0691,
+      "step": 14549
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.769202265484899e-06,
+      "loss": 0.9899,
+      "step": 14550
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.767903425455403e-06,
+      "loss": 1.0073,
+      "step": 14551
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.766604646308403e-06,
+      "loss": 0.9504,
+      "step": 14552
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.76530592806837e-06,
+      "loss": 1.0068,
+      "step": 14553
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.764007270759769e-06,
+      "loss": 1.0558,
+      "step": 14554
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.762708674407064e-06,
+      "loss": 1.0211,
+      "step": 14555
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.76141013903471e-06,
+      "loss": 0.972,
+      "step": 14556
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.760111664667168e-06,
+      "loss": 1.0073,
+      "step": 14557
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.758813251328906e-06,
+      "loss": 1.0653,
+      "step": 14558
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.757514899044369e-06,
+      "loss": 1.0128,
+      "step": 14559
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.756216607838024e-06,
+      "loss": 0.9828,
+      "step": 14560
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.754918377734322e-06,
+      "loss": 1.0502,
+      "step": 14561
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.753620208757721e-06,
+      "loss": 0.9741,
+      "step": 14562
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.75232210093267e-06,
+      "loss": 1.0069,
+      "step": 14563
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.751024054283625e-06,
+      "loss": 0.9992,
+      "step": 14564
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.749726068835032e-06,
+      "loss": 1.0696,
+      "step": 14565
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.748428144611349e-06,
+      "loss": 0.9755,
+      "step": 14566
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.74713028163702e-06,
+      "loss": 0.9854,
+      "step": 14567
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.745832479936492e-06,
+      "loss": 1.0725,
+      "step": 14568
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.744534739534214e-06,
+      "loss": 0.9975,
+      "step": 14569
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.743237060454626e-06,
+      "loss": 0.9706,
+      "step": 14570
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.741939442722178e-06,
+      "loss": 0.9648,
+      "step": 14571
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.740641886361312e-06,
+      "loss": 1.0457,
+      "step": 14572
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.73934439139647e-06,
+      "loss": 0.9821,
+      "step": 14573
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.73804695785209e-06,
+      "loss": 0.9995,
+      "step": 14574
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.73674958575261e-06,
+      "loss": 1.0078,
+      "step": 14575
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.735452275122476e-06,
+      "loss": 1.0593,
+      "step": 14576
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.734155025986122e-06,
+      "loss": 1.0864,
+      "step": 14577
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.732857838367978e-06,
+      "loss": 0.9584,
+      "step": 14578
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.731560712292486e-06,
+      "loss": 1.0129,
+      "step": 14579
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.730263647784078e-06,
+      "loss": 0.9388,
+      "step": 14580
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.728966644867187e-06,
+      "loss": 1.0226,
+      "step": 14581
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.72766970356624e-06,
+      "loss": 1.0858,
+      "step": 14582
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.726372823905673e-06,
+      "loss": 0.9908,
+      "step": 14583
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.725076005909916e-06,
+      "loss": 1.0624,
+      "step": 14584
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.723779249603393e-06,
+      "loss": 1.0767,
+      "step": 14585
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.722482555010528e-06,
+      "loss": 0.939,
+      "step": 14586
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.721185922155753e-06,
+      "loss": 1.0716,
+      "step": 14587
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.71988935106349e-06,
+      "loss": 0.9639,
+      "step": 14588
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.718592841758161e-06,
+      "loss": 1.0222,
+      "step": 14589
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.717296394264189e-06,
+      "loss": 0.9081,
+      "step": 14590
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.716000008605998e-06,
+      "loss": 1.0346,
+      "step": 14591
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.714703684808004e-06,
+      "loss": 1.0015,
+      "step": 14592
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.713407422894624e-06,
+      "loss": 1.0494,
+      "step": 14593
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.71211122289028e-06,
+      "loss": 1.0347,
+      "step": 14594
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.710815084819388e-06,
+      "loss": 1.0198,
+      "step": 14595
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.709519008706363e-06,
+      "loss": 1.076,
+      "step": 14596
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.708222994575613e-06,
+      "loss": 1.0808,
+      "step": 14597
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.70692704245156e-06,
+      "loss": 1.0295,
+      "step": 14598
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.705631152358608e-06,
+      "loss": 0.9772,
+      "step": 14599
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.704335324321172e-06,
+      "loss": 1.0261,
+      "step": 14600
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.703039558363658e-06,
+      "loss": 0.9499,
+      "step": 14601
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.701743854510482e-06,
+      "loss": 1.047,
+      "step": 14602
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.700448212786039e-06,
+      "loss": 0.9886,
+      "step": 14603
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.699152633214743e-06,
+      "loss": 1.1245,
+      "step": 14604
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.697857115820994e-06,
+      "loss": 0.9937,
+      "step": 14605
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.696561660629202e-06,
+      "loss": 0.993,
+      "step": 14606
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.695266267663765e-06,
+      "loss": 0.9754,
+      "step": 14607
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.693970936949084e-06,
+      "loss": 0.9843,
+      "step": 14608
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.69267566850956e-06,
+      "loss": 1.0202,
+      "step": 14609
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.691380462369588e-06,
+      "loss": 1.0233,
+      "step": 14610
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.69008531855357e-06,
+      "loss": 1.107,
+      "step": 14611
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.6887902370859e-06,
+      "loss": 0.9688,
+      "step": 14612
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.687495217990979e-06,
+      "loss": 1.0252,
+      "step": 14613
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.68620026129319e-06,
+      "loss": 1.037,
+      "step": 14614
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.684905367016936e-06,
+      "loss": 0.9941,
+      "step": 14615
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.683610535186604e-06,
+      "loss": 1.0505,
+      "step": 14616
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.682315765826588e-06,
+      "loss": 1.0428,
+      "step": 14617
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.681021058961272e-06,
+      "loss": 1.0921,
+      "step": 14618
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.679726414615049e-06,
+      "loss": 0.9688,
+      "step": 14619
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.678431832812306e-06,
+      "loss": 0.9669,
+      "step": 14620
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.677137313577425e-06,
+      "loss": 1.0211,
+      "step": 14621
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.675842856934789e-06,
+      "loss": 1.0198,
+      "step": 14622
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.674548462908789e-06,
+      "loss": 0.9997,
+      "step": 14623
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.673254131523805e-06,
+      "loss": 1.0314,
+      "step": 14624
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.671959862804213e-06,
+      "loss": 1.0058,
+      "step": 14625
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.670665656774398e-06,
+      "loss": 0.9694,
+      "step": 14626
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.669371513458736e-06,
+      "loss": 1.0695,
+      "step": 14627
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.66807743288161e-06,
+      "loss": 1.0498,
+      "step": 14628
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.66678341506739e-06,
+      "loss": 1.0193,
+      "step": 14629
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.66548946004045e-06,
+      "loss": 1.1377,
+      "step": 14630
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.664195567825174e-06,
+      "loss": 1.0131,
+      "step": 14631
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.662901738445923e-06,
+      "loss": 0.9887,
+      "step": 14632
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.661607971927077e-06,
+      "loss": 0.9967,
+      "step": 14633
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.660314268293e-06,
+      "loss": 1.0058,
+      "step": 14634
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.659020627568069e-06,
+      "loss": 1.0227,
+      "step": 14635
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.657727049776646e-06,
+      "loss": 1.0132,
+      "step": 14636
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.656433534943096e-06,
+      "loss": 0.9746,
+      "step": 14637
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.655140083091793e-06,
+      "loss": 1.0619,
+      "step": 14638
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.653846694247098e-06,
+      "loss": 1.0137,
+      "step": 14639
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.652553368433373e-06,
+      "loss": 1.0333,
+      "step": 14640
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.651260105674977e-06,
+      "loss": 1.0574,
+      "step": 14641
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.649966905996282e-06,
+      "loss": 1.0078,
+      "step": 14642
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.648673769421633e-06,
+      "loss": 0.9849,
+      "step": 14643
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.647380695975399e-06,
+      "loss": 1.036,
+      "step": 14644
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.64608768568193e-06,
+      "loss": 1.0574,
+      "step": 14645
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.644794738565597e-06,
+      "loss": 1.0604,
+      "step": 14646
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.643501854650735e-06,
+      "loss": 1.0455,
+      "step": 14647
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.642209033961711e-06,
+      "loss": 1.005,
+      "step": 14648
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.640916276522873e-06,
+      "loss": 1.0307,
+      "step": 14649
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.639623582358577e-06,
+      "loss": 1.1205,
+      "step": 14650
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.638330951493169e-06,
+      "loss": 0.9887,
+      "step": 14651
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.637038383950998e-06,
+      "loss": 0.9725,
+      "step": 14652
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.635745879756416e-06,
+      "loss": 0.9994,
+      "step": 14653
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.63445343893376e-06,
+      "loss": 0.994,
+      "step": 14654
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.633161061507388e-06,
+      "loss": 1.043,
+      "step": 14655
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.631868747501636e-06,
+      "loss": 0.9449,
+      "step": 14656
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.630576496940853e-06,
+      "loss": 0.9885,
+      "step": 14657
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.629284309849374e-06,
+      "loss": 1.0329,
+      "step": 14658
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.627992186251543e-06,
+      "loss": 0.9314,
+      "step": 14659
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.626700126171702e-06,
+      "loss": 1.1384,
+      "step": 14660
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.625408129634187e-06,
+      "loss": 0.9633,
+      "step": 14661
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.624116196663332e-06,
+      "loss": 0.9546,
+      "step": 14662
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.622824327283479e-06,
+      "loss": 1.0054,
+      "step": 14663
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.621532521518961e-06,
+      "loss": 1.0703,
+      "step": 14664
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.620240779394109e-06,
+      "loss": 1.0403,
+      "step": 14665
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.618949100933256e-06,
+      "loss": 1.0254,
+      "step": 14666
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.617657486160736e-06,
+      "loss": 0.9984,
+      "step": 14667
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.616365935100877e-06,
+      "loss": 1.0562,
+      "step": 14668
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.615074447778007e-06,
+      "loss": 1.0279,
+      "step": 14669
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.61378302421645e-06,
+      "loss": 1.13,
+      "step": 14670
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.612491664440545e-06,
+      "loss": 0.9513,
+      "step": 14671
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.611200368474604e-06,
+      "loss": 1.0139,
+      "step": 14672
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.609909136342956e-06,
+      "loss": 0.9591,
+      "step": 14673
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.608617968069921e-06,
+      "loss": 1.0448,
+      "step": 14674
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.607326863679829e-06,
+      "loss": 1.0295,
+      "step": 14675
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.60603582319699e-06,
+      "loss": 1.0485,
+      "step": 14676
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.604744846645727e-06,
+      "loss": 1.0282,
+      "step": 14677
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.60345393405036e-06,
+      "loss": 0.933,
+      "step": 14678
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.602163085435203e-06,
+      "loss": 1.0369,
+      "step": 14679
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.600872300824574e-06,
+      "loss": 1.0079,
+      "step": 14680
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.599581580242783e-06,
+      "loss": 0.937,
+      "step": 14681
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.598290923714152e-06,
+      "loss": 1.0399,
+      "step": 14682
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.597000331262982e-06,
+      "loss": 1.111,
+      "step": 14683
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.595709802913587e-06,
+      "loss": 1.084,
+      "step": 14684
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.59441933869028e-06,
+      "loss": 1.0315,
+      "step": 14685
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.593128938617372e-06,
+      "loss": 1.0432,
+      "step": 14686
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.591838602719159e-06,
+      "loss": 1.0827,
+      "step": 14687
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.590548331019957e-06,
+      "loss": 0.9841,
+      "step": 14688
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.589258123544063e-06,
+      "loss": 0.9309,
+      "step": 14689
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.58796798031579e-06,
+      "loss": 1.0307,
+      "step": 14690
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.58667790135943e-06,
+      "loss": 1.0714,
+      "step": 14691
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.58538788669929e-06,
+      "loss": 1.0227,
+      "step": 14692
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.58409793635967e-06,
+      "loss": 1.0686,
+      "step": 14693
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.582808050364864e-06,
+      "loss": 0.9645,
+      "step": 14694
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.581518228739174e-06,
+      "loss": 0.9482,
+      "step": 14695
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.580228471506893e-06,
+      "loss": 1.0006,
+      "step": 14696
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.578938778692322e-06,
+      "loss": 0.9826,
+      "step": 14697
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.577649150319745e-06,
+      "loss": 0.9561,
+      "step": 14698
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.576359586413462e-06,
+      "loss": 1.0,
+      "step": 14699
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.575070086997761e-06,
+      "loss": 0.9948,
+      "step": 14700
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.573780652096937e-06,
+      "loss": 1.0352,
+      "step": 14701
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.572491281735271e-06,
+      "loss": 1.0284,
+      "step": 14702
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.571201975937057e-06,
+      "loss": 0.9853,
+      "step": 14703
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.569912734726577e-06,
+      "loss": 1.0583,
+      "step": 14704
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.568623558128121e-06,
+      "loss": 1.0239,
+      "step": 14705
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.567334446165967e-06,
+      "loss": 0.9891,
+      "step": 14706
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.566045398864403e-06,
+      "loss": 0.9755,
+      "step": 14707
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.564756416247712e-06,
+      "loss": 0.9661,
+      "step": 14708
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.563467498340167e-06,
+      "loss": 1.0749,
+      "step": 14709
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.562178645166052e-06,
+      "loss": 1.0275,
+      "step": 14710
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.560889856749645e-06,
+      "loss": 1.0247,
+      "step": 14711
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.559601133115224e-06,
+      "loss": 1.0267,
+      "step": 14712
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.558312474287061e-06,
+      "loss": 1.0018,
+      "step": 14713
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.557023880289428e-06,
+      "loss": 1.0618,
+      "step": 14714
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.555735351146608e-06,
+      "loss": 1.0498,
+      "step": 14715
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.554446886882864e-06,
+      "loss": 1.0842,
+      "step": 14716
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.553158487522469e-06,
+      "loss": 0.9907,
+      "step": 14717
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.55187015308969e-06,
+      "loss": 1.0378,
+      "step": 14718
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.550581883608801e-06,
+      "loss": 1.0129,
+      "step": 14719
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.549293679104065e-06,
+      "loss": 1.0081,
+      "step": 14720
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.54800553959975e-06,
+      "loss": 1.0215,
+      "step": 14721
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.546717465120113e-06,
+      "loss": 1.0579,
+      "step": 14722
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.54542945568943e-06,
+      "loss": 0.9933,
+      "step": 14723
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.544141511331954e-06,
+      "loss": 0.9954,
+      "step": 14724
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.542853632071944e-06,
+      "loss": 1.0421,
+      "step": 14725
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.541565817933671e-06,
+      "loss": 1.045,
+      "step": 14726
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.54027806894138e-06,
+      "loss": 0.9575,
+      "step": 14727
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.538990385119335e-06,
+      "loss": 0.9808,
+      "step": 14728
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.537702766491791e-06,
+      "loss": 1.0021,
+      "step": 14729
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.536415213083008e-06,
+      "loss": 1.1118,
+      "step": 14730
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.535127724917227e-06,
+      "loss": 1.0579,
+      "step": 14731
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.533840302018709e-06,
+      "loss": 0.9562,
+      "step": 14732
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.532552944411703e-06,
+      "loss": 1.029,
+      "step": 14733
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.531265652120463e-06,
+      "loss": 0.971,
+      "step": 14734
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.52997842516923e-06,
+      "loss": 0.9784,
+      "step": 14735
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.528691263582254e-06,
+      "loss": 1.0473,
+      "step": 14736
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.527404167383785e-06,
+      "loss": 0.9402,
+      "step": 14737
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.526117136598062e-06,
+      "loss": 0.9375,
+      "step": 14738
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.524830171249332e-06,
+      "loss": 1.0534,
+      "step": 14739
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.523543271361836e-06,
+      "loss": 0.996,
+      "step": 14740
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.52225643695982e-06,
+      "loss": 1.0463,
+      "step": 14741
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.520969668067514e-06,
+      "loss": 1.1132,
+      "step": 14742
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.519682964709164e-06,
+      "loss": 1.0502,
+      "step": 14743
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.518396326909007e-06,
+      "loss": 0.9718,
+      "step": 14744
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.51710975469128e-06,
+      "loss": 0.9457,
+      "step": 14745
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.515823248080211e-06,
+      "loss": 0.9705,
+      "step": 14746
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.514536807100043e-06,
+      "loss": 1.0586,
+      "step": 14747
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.513250431775003e-06,
+      "loss": 1.0351,
+      "step": 14748
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.511964122129324e-06,
+      "loss": 1.0091,
+      "step": 14749
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.510677878187233e-06,
+      "loss": 0.9814,
+      "step": 14750
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.509391699972963e-06,
+      "loss": 1.0043,
+      "step": 14751
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.50810558751074e-06,
+      "loss": 1.0059,
+      "step": 14752
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.506819540824789e-06,
+      "loss": 0.9855,
+      "step": 14753
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.505533559939335e-06,
+      "loss": 0.9924,
+      "step": 14754
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.504247644878608e-06,
+      "loss": 1.0067,
+      "step": 14755
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.502961795666821e-06,
+      "loss": 1.0763,
+      "step": 14756
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.501676012328202e-06,
+      "loss": 1.0287,
+      "step": 14757
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.500390294886965e-06,
+      "loss": 0.9709,
+      "step": 14758
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.499104643367338e-06,
+      "loss": 0.993,
+      "step": 14759
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.497819057793531e-06,
+      "loss": 0.932,
+      "step": 14760
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.496533538189764e-06,
+      "loss": 1.0628,
+      "step": 14761
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.495248084580247e-06,
+      "loss": 1.0657,
+      "step": 14762
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.493962696989204e-06,
+      "loss": 1.0851,
+      "step": 14763
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.492677375440838e-06,
+      "loss": 1.037,
+      "step": 14764
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.491392119959365e-06,
+      "loss": 0.981,
+      "step": 14765
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.490106930568995e-06,
+      "loss": 1.0373,
+      "step": 14766
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.488821807293932e-06,
+      "loss": 1.0106,
+      "step": 14767
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.48753675015839e-06,
+      "loss": 1.0151,
+      "step": 14768
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.486251759186572e-06,
+      "loss": 1.0692,
+      "step": 14769
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.48496683440269e-06,
+      "loss": 1.0849,
+      "step": 14770
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.483681975830935e-06,
+      "loss": 1.1684,
+      "step": 14771
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.48239718349552e-06,
+      "loss": 1.0025,
+      "step": 14772
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.48111245742064e-06,
+      "loss": 1.0634,
+      "step": 14773
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.479827797630506e-06,
+      "loss": 0.9927,
+      "step": 14774
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.478543204149302e-06,
+      "loss": 0.9722,
+      "step": 14775
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.477258677001237e-06,
+      "loss": 0.9676,
+      "step": 14776
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.475974216210503e-06,
+      "loss": 0.9883,
+      "step": 14777
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.474689821801295e-06,
+      "loss": 1.0476,
+      "step": 14778
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.473405493797805e-06,
+      "loss": 0.9541,
+      "step": 14779
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.472121232224232e-06,
+      "loss": 1.0117,
+      "step": 14780
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.470837037104764e-06,
+      "loss": 1.0876,
+      "step": 14781
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.469552908463588e-06,
+      "loss": 0.9961,
+      "step": 14782
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.4682688463249e-06,
+      "loss": 1.0823,
+      "step": 14783
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.46698485071288e-06,
+      "loss": 1.0896,
+      "step": 14784
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.465700921651721e-06,
+      "loss": 1.0112,
+      "step": 14785
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.464417059165602e-06,
+      "loss": 1.0018,
+      "step": 14786
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.463133263278712e-06,
+      "loss": 1.1014,
+      "step": 14787
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.461849534015233e-06,
+      "loss": 0.9831,
+      "step": 14788
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.460565871399343e-06,
+      "loss": 0.8992,
+      "step": 14789
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.459282275455222e-06,
+      "loss": 1.0301,
+      "step": 14790
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.457998746207055e-06,
+      "loss": 1.1404,
+      "step": 14791
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.456715283679018e-06,
+      "loss": 1.0548,
+      "step": 14792
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.455431887895282e-06,
+      "loss": 0.9746,
+      "step": 14793
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.454148558880023e-06,
+      "loss": 1.0023,
+      "step": 14794
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.45286529665742e-06,
+      "loss": 1.1529,
+      "step": 14795
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.451582101251645e-06,
+      "loss": 1.004,
+      "step": 14796
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.450298972686865e-06,
+      "loss": 1.0296,
+      "step": 14797
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.44901591098725e-06,
+      "loss": 0.9387,
+      "step": 14798
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.447732916176975e-06,
+      "loss": 0.9237,
+      "step": 14799
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.446449988280203e-06,
+      "loss": 1.0441,
+      "step": 14800
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.445167127321098e-06,
+      "loss": 0.9353,
+      "step": 14801
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.44388433332383e-06,
+      "loss": 0.9648,
+      "step": 14802
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.442601606312562e-06,
+      "loss": 1.0265,
+      "step": 14803
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.441318946311454e-06,
+      "loss": 1.0484,
+      "step": 14804
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.44003635334467e-06,
+      "loss": 1.0863,
+      "step": 14805
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.438753827436365e-06,
+      "loss": 1.0811,
+      "step": 14806
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.437471368610708e-06,
+      "loss": 0.9656,
+      "step": 14807
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.436188976891846e-06,
+      "loss": 1.022,
+      "step": 14808
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.434906652303938e-06,
+      "loss": 1.0515,
+      "step": 14809
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.433624394871145e-06,
+      "loss": 0.9631,
+      "step": 14810
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.432342204617609e-06,
+      "loss": 0.9326,
+      "step": 14811
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.431060081567493e-06,
+      "loss": 0.9937,
+      "step": 14812
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.42977802574494e-06,
+      "loss": 1.0266,
+      "step": 14813
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.428496037174112e-06,
+      "loss": 0.9319,
+      "step": 14814
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.427214115879143e-06,
+      "loss": 1.0413,
+      "step": 14815
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.42593226188419e-06,
+      "loss": 1.0291,
+      "step": 14816
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.424650475213394e-06,
+      "loss": 1.0442,
+      "step": 14817
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.423368755890908e-06,
+      "loss": 1.0663,
+      "step": 14818
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.422087103940864e-06,
+      "loss": 0.921,
+      "step": 14819
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.420805519387413e-06,
+      "loss": 1.0857,
+      "step": 14820
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.419524002254692e-06,
+      "loss": 1.0653,
+      "step": 14821
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.41824255256684e-06,
+      "loss": 1.1237,
+      "step": 14822
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.416961170347998e-06,
+      "loss": 0.9603,
+      "step": 14823
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.415679855622302e-06,
+      "loss": 1.0179,
+      "step": 14824
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.414398608413891e-06,
+      "loss": 1.0036,
+      "step": 14825
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.413117428746893e-06,
+      "loss": 1.0154,
+      "step": 14826
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.411836316645446e-06,
+      "loss": 1.0199,
+      "step": 14827
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.410555272133682e-06,
+      "loss": 1.0207,
+      "step": 14828
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.409274295235734e-06,
+      "loss": 1.0413,
+      "step": 14829
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.407993385975723e-06,
+      "loss": 0.9853,
+      "step": 14830
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.406712544377787e-06,
+      "loss": 1.0684,
+      "step": 14831
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.40543177046605e-06,
+      "loss": 0.9858,
+      "step": 14832
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.404151064264636e-06,
+      "loss": 1.0849,
+      "step": 14833
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.402870425797667e-06,
+      "loss": 0.983,
+      "step": 14834
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.401589855089273e-06,
+      "loss": 1.0141,
+      "step": 14835
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.400309352163575e-06,
+      "loss": 1.0743,
+      "step": 14836
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.399028917044686e-06,
+      "loss": 0.9999,
+      "step": 14837
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.39774854975673e-06,
+      "loss": 0.9608,
+      "step": 14838
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.396468250323829e-06,
+      "loss": 1.0592,
+      "step": 14839
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.395188018770095e-06,
+      "loss": 1.0855,
+      "step": 14840
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.393907855119646e-06,
+      "loss": 1.1027,
+      "step": 14841
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.39262775939659e-06,
+      "loss": 1.0634,
+      "step": 14842
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.391347731625051e-06,
+      "loss": 0.9961,
+      "step": 14843
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.390067771829131e-06,
+      "loss": 0.9779,
+      "step": 14844
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.388787880032945e-06,
+      "loss": 1.0281,
+      "step": 14845
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.3875080562606e-06,
+      "loss": 0.9486,
+      "step": 14846
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.386228300536208e-06,
+      "loss": 0.9835,
+      "step": 14847
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.384948612883872e-06,
+      "loss": 0.9498,
+      "step": 14848
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.383668993327695e-06,
+      "loss": 1.0159,
+      "step": 14849
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.382389441891789e-06,
+      "loss": 0.9775,
+      "step": 14850
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.381109958600247e-06,
+      "loss": 0.9815,
+      "step": 14851
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.379830543477177e-06,
+      "loss": 1.0473,
+      "step": 14852
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.378551196546676e-06,
+      "loss": 1.0128,
+      "step": 14853
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.377271917832848e-06,
+      "loss": 0.9307,
+      "step": 14854
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.37599270735978e-06,
+      "loss": 1.0025,
+      "step": 14855
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.374713565151579e-06,
+      "loss": 0.9971,
+      "step": 14856
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.373434491232333e-06,
+      "loss": 1.0065,
+      "step": 14857
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.372155485626145e-06,
+      "loss": 0.9925,
+      "step": 14858
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.370876548357095e-06,
+      "loss": 1.0205,
+      "step": 14859
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.369597679449282e-06,
+      "loss": 1.0383,
+      "step": 14860
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.368318878926796e-06,
+      "loss": 1.0893,
+      "step": 14861
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.367040146813721e-06,
+      "loss": 1.158,
+      "step": 14862
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.365761483134144e-06,
+      "loss": 1.0114,
+      "step": 14863
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.364482887912158e-06,
+      "loss": 1.0335,
+      "step": 14864
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.363204361171846e-06,
+      "loss": 1.0931,
+      "step": 14865
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.361925902937285e-06,
+      "loss": 1.1343,
+      "step": 14866
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.360647513232561e-06,
+      "loss": 1.0595,
+      "step": 14867
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.359369192081756e-06,
+      "loss": 1.0048,
+      "step": 14868
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.35809093950895e-06,
+      "loss": 1.01,
+      "step": 14869
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.356812755538216e-06,
+      "loss": 1.0633,
+      "step": 14870
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.355534640193637e-06,
+      "loss": 0.9945,
+      "step": 14871
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.354256593499288e-06,
+      "loss": 0.9997,
+      "step": 14872
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.35297861547924e-06,
+      "loss": 1.0346,
+      "step": 14873
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.351700706157565e-06,
+      "loss": 0.9866,
+      "step": 14874
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.350422865558341e-06,
+      "loss": 1.0508,
+      "step": 14875
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.349145093705635e-06,
+      "loss": 1.0071,
+      "step": 14876
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.347867390623516e-06,
+      "loss": 0.9951,
+      "step": 14877
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.34658975633605e-06,
+      "loss": 0.8801,
+      "step": 14878
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.345312190867308e-06,
+      "loss": 0.9969,
+      "step": 14879
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.344034694241353e-06,
+      "loss": 1.0748,
+      "step": 14880
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.342757266482249e-06,
+      "loss": 1.0593,
+      "step": 14881
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.341479907614055e-06,
+      "loss": 1.0365,
+      "step": 14882
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.340202617660842e-06,
+      "loss": 0.9667,
+      "step": 14883
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.338925396646661e-06,
+      "loss": 1.0569,
+      "step": 14884
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.337648244595574e-06,
+      "loss": 1.0471,
+      "step": 14885
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.336371161531639e-06,
+      "loss": 1.0281,
+      "step": 14886
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.335094147478913e-06,
+      "loss": 0.9904,
+      "step": 14887
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.333817202461447e-06,
+      "loss": 1.017,
+      "step": 14888
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.3325403265033e-06,
+      "loss": 1.0885,
+      "step": 14889
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.33126351962852e-06,
+      "loss": 1.0615,
+      "step": 14890
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.329986781861162e-06,
+      "loss": 1.0415,
+      "step": 14891
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.32871011322527e-06,
+      "loss": 1.0253,
+      "step": 14892
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.327433513744898e-06,
+      "loss": 1.0406,
+      "step": 14893
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.326156983444091e-06,
+      "loss": 1.0565,
+      "step": 14894
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.324880522346893e-06,
+      "loss": 0.9783,
+      "step": 14895
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.323604130477352e-06,
+      "loss": 0.9526,
+      "step": 14896
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.322327807859507e-06,
+      "loss": 1.0849,
+      "step": 14897
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.321051554517407e-06,
+      "loss": 1.0584,
+      "step": 14898
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.319775370475084e-06,
+      "loss": 1.0028,
+      "step": 14899
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.318499255756581e-06,
+      "loss": 0.9235,
+      "step": 14900
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.317223210385933e-06,
+      "loss": 0.9919,
+      "step": 14901
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.31594723438719e-06,
+      "loss": 0.9462,
+      "step": 14902
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.314671327784367e-06,
+      "loss": 1.1554,
+      "step": 14903
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.313395490601512e-06,
+      "loss": 1.0419,
+      "step": 14904
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.312119722862657e-06,
+      "loss": 1.0738,
+      "step": 14905
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.310844024591827e-06,
+      "loss": 1.0542,
+      "step": 14906
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.309568395813055e-06,
+      "loss": 1.0348,
+      "step": 14907
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.30829283655037e-06,
+      "loss": 1.0612,
+      "step": 14908
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.307017346827803e-06,
+      "loss": 1.0372,
+      "step": 14909
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.305741926669375e-06,
+      "loss": 0.9994,
+      "step": 14910
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.304466576099112e-06,
+      "loss": 1.0,
+      "step": 14911
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.303191295141042e-06,
+      "loss": 1.0099,
+      "step": 14912
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.301916083819182e-06,
+      "loss": 1.078,
+      "step": 14913
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.30064094215755e-06,
+      "loss": 1.0403,
+      "step": 14914
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.299365870180176e-06,
+      "loss": 0.9785,
+      "step": 14915
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.298090867911073e-06,
+      "loss": 1.003,
+      "step": 14916
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.296815935374256e-06,
+      "loss": 1.0296,
+      "step": 14917
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.29554107259374e-06,
+      "loss": 1.0456,
+      "step": 14918
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.294266279593544e-06,
+      "loss": 1.0909,
+      "step": 14919
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.292991556397683e-06,
+      "loss": 0.9864,
+      "step": 14920
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.291716903030161e-06,
+      "loss": 0.9929,
+      "step": 14921
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.29044231951499e-06,
+      "loss": 0.9973,
+      "step": 14922
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.289167805876185e-06,
+      "loss": 1.0374,
+      "step": 14923
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.287893362137749e-06,
+      "loss": 1.0149,
+      "step": 14924
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.286618988323688e-06,
+      "loss": 0.9567,
+      "step": 14925
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.285344684458008e-06,
+      "loss": 1.0024,
+      "step": 14926
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.284070450564718e-06,
+      "loss": 1.0035,
+      "step": 14927
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.282796286667813e-06,
+      "loss": 0.9764,
+      "step": 14928
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.281522192791298e-06,
+      "loss": 1.0939,
+      "step": 14929
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.28024816895917e-06,
+      "loss": 0.9671,
+      "step": 14930
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.278974215195434e-06,
+      "loss": 1.0393,
+      "step": 14931
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.277700331524083e-06,
+      "loss": 0.9956,
+      "step": 14932
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.276426517969111e-06,
+      "loss": 1.081,
+      "step": 14933
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.275152774554519e-06,
+      "loss": 0.9992,
+      "step": 14934
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.273879101304288e-06,
+      "loss": 1.0178,
+      "step": 14935
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.272605498242423e-06,
+      "loss": 1.0542,
+      "step": 14936
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.27133196539291e-06,
+      "loss": 1.0328,
+      "step": 14937
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.270058502779739e-06,
+      "loss": 0.9855,
+      "step": 14938
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.268785110426893e-06,
+      "loss": 0.9485,
+      "step": 14939
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.267511788358365e-06,
+      "loss": 0.9554,
+      "step": 14940
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.266238536598136e-06,
+      "loss": 1.0971,
+      "step": 14941
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.264965355170199e-06,
+      "loss": 1.0204,
+      "step": 14942
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.263692244098522e-06,
+      "loss": 0.9776,
+      "step": 14943
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.262419203407098e-06,
+      "loss": 1.0389,
+      "step": 14944
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.261146233119905e-06,
+      "loss": 0.9802,
+      "step": 14945
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.259873333260917e-06,
+      "loss": 0.9162,
+      "step": 14946
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.258600503854114e-06,
+      "loss": 0.9257,
+      "step": 14947
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.257327744923473e-06,
+      "loss": 1.0122,
+      "step": 14948
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.256055056492972e-06,
+      "loss": 1.0048,
+      "step": 14949
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.254782438586579e-06,
+      "loss": 0.9803,
+      "step": 14950
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.253509891228265e-06,
+      "loss": 1.1121,
+      "step": 14951
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.252237414442006e-06,
+      "loss": 0.9992,
+      "step": 14952
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.25096500825177e-06,
+      "loss": 1.0772,
+      "step": 14953
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.249692672681525e-06,
+      "loss": 1.0445,
+      "step": 14954
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.24842040775523e-06,
+      "loss": 1.0288,
+      "step": 14955
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.247148213496867e-06,
+      "loss": 1.1042,
+      "step": 14956
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.245876089930385e-06,
+      "loss": 0.973,
+      "step": 14957
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.244604037079753e-06,
+      "loss": 1.0594,
+      "step": 14958
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.243332054968934e-06,
+      "loss": 0.9521,
+      "step": 14959
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.242060143621886e-06,
+      "loss": 0.9586,
+      "step": 14960
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.240788303062566e-06,
+      "loss": 1.0019,
+      "step": 14961
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.23951653331493e-06,
+      "loss": 0.9874,
+      "step": 14962
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.238244834402943e-06,
+      "loss": 1.0324,
+      "step": 14963
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.236973206350554e-06,
+      "loss": 0.9663,
+      "step": 14964
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.235701649181715e-06,
+      "loss": 1.0278,
+      "step": 14965
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.234430162920378e-06,
+      "loss": 0.9826,
+      "step": 14966
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.233158747590501e-06,
+      "loss": 0.9786,
+      "step": 14967
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.231887403216024e-06,
+      "loss": 0.9437,
+      "step": 14968
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.230616129820899e-06,
+      "loss": 1.007,
+      "step": 14969
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.22934492742907e-06,
+      "loss": 1.0375,
+      "step": 14970
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.228073796064493e-06,
+      "loss": 1.0623,
+      "step": 14971
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.226802735751098e-06,
+      "loss": 1.0362,
+      "step": 14972
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.225531746512838e-06,
+      "loss": 0.9353,
+      "step": 14973
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.224260828373647e-06,
+      "loss": 0.9707,
+      "step": 14974
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.222989981357472e-06,
+      "loss": 1.0458,
+      "step": 14975
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.221719205488248e-06,
+      "loss": 1.0268,
+      "step": 14976
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.220448500789914e-06,
+      "loss": 1.0485,
+      "step": 14977
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.219177867286402e-06,
+      "loss": 1.0453,
+      "step": 14978
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.217907305001652e-06,
+      "loss": 0.9895,
+      "step": 14979
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.216636813959594e-06,
+      "loss": 1.0194,
+      "step": 14980
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.21536639418416e-06,
+      "loss": 1.0505,
+      "step": 14981
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.214096045699285e-06,
+      "loss": 1.0858,
+      "step": 14982
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.212825768528892e-06,
+      "loss": 1.0036,
+      "step": 14983
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.211555562696914e-06,
+      "loss": 1.0,
+      "step": 14984
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.210285428227273e-06,
+      "loss": 1.0001,
+      "step": 14985
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.209015365143903e-06,
+      "loss": 0.9776,
+      "step": 14986
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.207745373470717e-06,
+      "loss": 1.087,
+      "step": 14987
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.206475453231643e-06,
+      "loss": 1.0187,
+      "step": 14988
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.205205604450607e-06,
+      "loss": 1.0052,
+      "step": 14989
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.203935827151519e-06,
+      "loss": 1.0948,
+      "step": 14990
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.202666121358302e-06,
+      "loss": 0.9997,
+      "step": 14991
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.201396487094877e-06,
+      "loss": 1.0495,
+      "step": 14992
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.200126924385159e-06,
+      "loss": 1.0039,
+      "step": 14993
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.198857433253056e-06,
+      "loss": 0.9646,
+      "step": 14994
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.197588013722486e-06,
+      "loss": 0.9958,
+      "step": 14995
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.196318665817362e-06,
+      "loss": 1.0202,
+      "step": 14996
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.195049389561592e-06,
+      "loss": 1.0916,
+      "step": 14997
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.193780184979087e-06,
+      "loss": 1.069,
+      "step": 14998
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.192511052093751e-06,
+      "loss": 0.9689,
+      "step": 14999
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.191241990929497e-06,
+      "loss": 1.04,
+      "step": 15000
+    },
+    {
+      "epoch": 0.64,
+      "eval_text_loss": 0.7747818231582642,
+      "eval_text_runtime": 1.4109,
+      "eval_text_samples_per_second": 52.449,
+      "eval_text_steps_per_second": 1.418,
+      "step": 15000
+    },
+    {
+      "epoch": 0.64,
+      "eval_svit_loss": 1.1021628379821777,
+      "eval_svit_runtime": 4.9611,
+      "eval_svit_samples_per_second": 137.267,
+      "eval_svit_steps_per_second": 2.217,
+      "step": 15000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.189973001510225e-06,
+      "loss": 1.0171,
+      "step": 15001
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.188704083859839e-06,
+      "loss": 0.9953,
+      "step": 15002
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.187435238002245e-06,
+      "loss": 0.9182,
+      "step": 15003
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.18616646396134e-06,
+      "loss": 1.0293,
+      "step": 15004
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.184897761761028e-06,
+      "loss": 1.0539,
+      "step": 15005
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.1836291314252e-06,
+      "loss": 1.0138,
+      "step": 15006
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.182360572977764e-06,
+      "loss": 1.0506,
+      "step": 15007
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.181092086442604e-06,
+      "loss": 1.0624,
+      "step": 15008
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.17982367184362e-06,
+      "loss": 0.98,
+      "step": 15009
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.178555329204705e-06,
+      "loss": 0.9824,
+      "step": 15010
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.177287058549755e-06,
+      "loss": 0.9847,
+      "step": 15011
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.17601885990265e-06,
+      "loss": 1.0881,
+      "step": 15012
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.174750733287285e-06,
+      "loss": 1.0309,
+      "step": 15013
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.173482678727544e-06,
+      "loss": 1.0183,
+      "step": 15014
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.172214696247323e-06,
+      "loss": 1.0076,
+      "step": 15015
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.170946785870495e-06,
+      "loss": 1.0217,
+      "step": 15016
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.16967894762095e-06,
+      "loss": 1.031,
+      "step": 15017
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.168411181522569e-06,
+      "loss": 1.027,
+      "step": 15018
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.167143487599227e-06,
+      "loss": 1.0176,
+      "step": 15019
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.165875865874811e-06,
+      "loss": 1.0932,
+      "step": 15020
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.164608316373197e-06,
+      "loss": 1.0402,
+      "step": 15021
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.163340839118261e-06,
+      "loss": 0.9885,
+      "step": 15022
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.162073434133876e-06,
+      "loss": 1.0285,
+      "step": 15023
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.16080610144392e-06,
+      "loss": 1.0683,
+      "step": 15024
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.159538841072264e-06,
+      "loss": 1.0121,
+      "step": 15025
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.158271653042778e-06,
+      "loss": 1.0489,
+      "step": 15026
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.157004537379333e-06,
+      "loss": 1.0177,
+      "step": 15027
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.155737494105794e-06,
+      "loss": 1.0575,
+      "step": 15028
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.154470523246037e-06,
+      "loss": 0.9623,
+      "step": 15029
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.153203624823918e-06,
+      "loss": 1.0828,
+      "step": 15030
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.151936798863302e-06,
+      "loss": 1.0302,
+      "step": 15031
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.15067004538806e-06,
+      "loss": 1.0092,
+      "step": 15032
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.149403364422051e-06,
+      "loss": 0.9531,
+      "step": 15033
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.148136755989129e-06,
+      "loss": 1.0627,
+      "step": 15034
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.146870220113158e-06,
+      "loss": 1.0547,
+      "step": 15035
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.145603756817994e-06,
+      "loss": 0.9899,
+      "step": 15036
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.144337366127497e-06,
+      "loss": 0.9874,
+      "step": 15037
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.143071048065517e-06,
+      "loss": 0.9487,
+      "step": 15038
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.141804802655905e-06,
+      "loss": 1.0102,
+      "step": 15039
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.140538629922523e-06,
+      "loss": 1.053,
+      "step": 15040
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.139272529889213e-06,
+      "loss": 1.0215,
+      "step": 15041
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.138006502579827e-06,
+      "loss": 1.0391,
+      "step": 15042
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.136740548018212e-06,
+      "loss": 0.9666,
+      "step": 15043
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.135474666228217e-06,
+      "loss": 0.9399,
+      "step": 15044
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.134208857233688e-06,
+      "loss": 1.0439,
+      "step": 15045
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.132943121058463e-06,
+      "loss": 1.0546,
+      "step": 15046
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.131677457726389e-06,
+      "loss": 1.0028,
+      "step": 15047
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.13041186726131e-06,
+      "loss": 0.9886,
+      "step": 15048
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.129146349687058e-06,
+      "loss": 0.9993,
+      "step": 15049
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.127880905027476e-06,
+      "loss": 1.0254,
+      "step": 15050
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.126615533306405e-06,
+      "loss": 0.9765,
+      "step": 15051
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.12535023454767e-06,
+      "loss": 0.9882,
+      "step": 15052
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.124085008775114e-06,
+      "loss": 1.0902,
+      "step": 15053
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.122819856012564e-06,
+      "loss": 0.9862,
+      "step": 15054
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.12155477628386e-06,
+      "loss": 0.9303,
+      "step": 15055
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.120289769612824e-06,
+      "loss": 1.0145,
+      "step": 15056
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.119024836023286e-06,
+      "loss": 1.1232,
+      "step": 15057
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.117759975539075e-06,
+      "loss": 1.0279,
+      "step": 15058
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.11649518818402e-06,
+      "loss": 0.9498,
+      "step": 15059
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.11523047398194e-06,
+      "loss": 1.0483,
+      "step": 15060
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.11396583295666e-06,
+      "loss": 1.0208,
+      "step": 15061
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.112701265132006e-06,
+      "loss": 1.0013,
+      "step": 15062
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.111436770531791e-06,
+      "loss": 1.0251,
+      "step": 15063
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.11017234917984e-06,
+      "loss": 1.0592,
+      "step": 15064
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.108908001099967e-06,
+      "loss": 0.9969,
+      "step": 15065
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.107643726315993e-06,
+      "loss": 0.9966,
+      "step": 15066
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.106379524851726e-06,
+      "loss": 0.9658,
+      "step": 15067
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.105115396730985e-06,
+      "loss": 0.9972,
+      "step": 15068
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.103851341977582e-06,
+      "loss": 0.9895,
+      "step": 15069
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.102587360615328e-06,
+      "loss": 1.0178,
+      "step": 15070
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.101323452668027e-06,
+      "loss": 0.9312,
+      "step": 15071
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.100059618159493e-06,
+      "loss": 1.0616,
+      "step": 15072
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.098795857113533e-06,
+      "loss": 0.9959,
+      "step": 15073
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.097532169553947e-06,
+      "loss": 0.9459,
+      "step": 15074
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.09626855550454e-06,
+      "loss": 0.9689,
+      "step": 15075
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.095005014989119e-06,
+      "loss": 1.0109,
+      "step": 15076
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.093741548031485e-06,
+      "loss": 1.031,
+      "step": 15077
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.092478154655432e-06,
+      "loss": 1.0082,
+      "step": 15078
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.091214834884758e-06,
+      "loss": 1.0451,
+      "step": 15079
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.08995158874327e-06,
+      "loss": 1.0164,
+      "step": 15080
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.088688416254755e-06,
+      "loss": 1.0256,
+      "step": 15081
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.087425317443007e-06,
+      "loss": 1.0141,
+      "step": 15082
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.086162292331822e-06,
+      "loss": 0.9948,
+      "step": 15083
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.084899340944993e-06,
+      "loss": 1.0468,
+      "step": 15084
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.083636463306306e-06,
+      "loss": 0.9832,
+      "step": 15085
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.08237365943955e-06,
+      "loss": 0.9786,
+      "step": 15086
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.081110929368512e-06,
+      "loss": 1.053,
+      "step": 15087
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.079848273116983e-06,
+      "loss": 1.0597,
+      "step": 15088
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.078585690708741e-06,
+      "loss": 1.0562,
+      "step": 15089
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.077323182167572e-06,
+      "loss": 1.0864,
+      "step": 15090
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.07606074751726e-06,
+      "loss": 1.0308,
+      "step": 15091
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.07479838678158e-06,
+      "loss": 0.9872,
+      "step": 15092
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.073536099984315e-06,
+      "loss": 0.9691,
+      "step": 15093
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.072273887149236e-06,
+      "loss": 1.0436,
+      "step": 15094
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.071011748300134e-06,
+      "loss": 1.0272,
+      "step": 15095
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.069749683460766e-06,
+      "loss": 1.0543,
+      "step": 15096
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.068487692654915e-06,
+      "loss": 1.057,
+      "step": 15097
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.067225775906349e-06,
+      "loss": 0.9921,
+      "step": 15098
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.065963933238847e-06,
+      "loss": 0.9346,
+      "step": 15099
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.064702164676166e-06,
+      "loss": 1.0022,
+      "step": 15100
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.063440470242081e-06,
+      "loss": 0.9909,
+      "step": 15101
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.06217884996036e-06,
+      "loss": 1.0664,
+      "step": 15102
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.06091730385476e-06,
+      "loss": 1.0343,
+      "step": 15103
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.059655831949051e-06,
+      "loss": 1.0971,
+      "step": 15104
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.058394434266995e-06,
+      "loss": 0.9539,
+      "step": 15105
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.057133110832352e-06,
+      "loss": 1.0389,
+      "step": 15106
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.055871861668876e-06,
+      "loss": 0.9901,
+      "step": 15107
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.054610686800333e-06,
+      "loss": 1.0184,
+      "step": 15108
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.053349586250477e-06,
+      "loss": 0.997,
+      "step": 15109
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.052088560043061e-06,
+      "loss": 1.0157,
+      "step": 15110
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.05082760820184e-06,
+      "loss": 1.0104,
+      "step": 15111
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.049566730750564e-06,
+      "loss": 1.0522,
+      "step": 15112
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.048305927712992e-06,
+      "loss": 0.9759,
+      "step": 15113
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.047045199112864e-06,
+      "loss": 0.9711,
+      "step": 15114
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.045784544973932e-06,
+      "loss": 1.0549,
+      "step": 15115
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.044523965319942e-06,
+      "loss": 1.0394,
+      "step": 15116
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.043263460174644e-06,
+      "loss": 1.0629,
+      "step": 15117
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.042003029561775e-06,
+      "loss": 1.0255,
+      "step": 15118
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.04074267350508e-06,
+      "loss": 0.9784,
+      "step": 15119
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.039482392028302e-06,
+      "loss": 0.9985,
+      "step": 15120
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.03822218515518e-06,
+      "loss": 0.9967,
+      "step": 15121
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.036962052909453e-06,
+      "loss": 1.0771,
+      "step": 15122
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.035701995314853e-06,
+      "loss": 1.027,
+      "step": 15123
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.034442012395123e-06,
+      "loss": 1.0375,
+      "step": 15124
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.03318210417399e-06,
+      "loss": 0.999,
+      "step": 15125
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.031922270675193e-06,
+      "loss": 0.975,
+      "step": 15126
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.030662511922456e-06,
+      "loss": 1.0594,
+      "step": 15127
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.029402827939519e-06,
+      "loss": 1.0044,
+      "step": 15128
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.028143218750102e-06,
+      "loss": 0.9955,
+      "step": 15129
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.026883684377935e-06,
+      "loss": 1.026,
+      "step": 15130
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.02562422484674e-06,
+      "loss": 1.1097,
+      "step": 15131
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.02436484018025e-06,
+      "loss": 1.0958,
+      "step": 15132
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.02310553040218e-06,
+      "loss": 0.9855,
+      "step": 15133
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.02184629553625e-06,
+      "loss": 0.9633,
+      "step": 15134
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.020587135606192e-06,
+      "loss": 1.0356,
+      "step": 15135
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.01932805063571e-06,
+      "loss": 1.0229,
+      "step": 15136
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.01806904064853e-06,
+      "loss": 0.9952,
+      "step": 15137
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.016810105668365e-06,
+      "loss": 1.0089,
+      "step": 15138
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.015551245718933e-06,
+      "loss": 0.951,
+      "step": 15139
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.014292460823938e-06,
+      "loss": 1.0096,
+      "step": 15140
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.013033751007102e-06,
+      "loss": 1.0691,
+      "step": 15141
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.011775116292125e-06,
+      "loss": 1.0287,
+      "step": 15142
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.010516556702731e-06,
+      "loss": 0.9937,
+      "step": 15143
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.009258072262607e-06,
+      "loss": 1.0127,
+      "step": 15144
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.007999662995474e-06,
+      "loss": 1.0122,
+      "step": 15145
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.006741328925033e-06,
+      "loss": 1.0552,
+      "step": 15146
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.005483070074982e-06,
+      "loss": 1.0609,
+      "step": 15147
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.004224886469027e-06,
+      "loss": 0.9937,
+      "step": 15148
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.002966778130869e-06,
+      "loss": 1.0423,
+      "step": 15149
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.00170874508421e-06,
+      "loss": 0.9964,
+      "step": 15150
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.000450787352734e-06,
+      "loss": 1.0732,
+      "step": 15151
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.999192904960151e-06,
+      "loss": 0.9508,
+      "step": 15152
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.99793509793015e-06,
+      "loss": 1.0193,
+      "step": 15153
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.996677366286424e-06,
+      "loss": 1.0142,
+      "step": 15154
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.995419710052663e-06,
+      "loss": 1.0512,
+      "step": 15155
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.99416212925256e-06,
+      "loss": 0.9227,
+      "step": 15156
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.992904623909807e-06,
+      "loss": 0.9986,
+      "step": 15157
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.991647194048087e-06,
+      "loss": 1.0303,
+      "step": 15158
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.990389839691081e-06,
+      "loss": 1.0984,
+      "step": 15159
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.989132560862484e-06,
+      "loss": 0.9835,
+      "step": 15160
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.987875357585975e-06,
+      "loss": 1.0756,
+      "step": 15161
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.986618229885233e-06,
+      "loss": 1.0407,
+      "step": 15162
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.98536117778394e-06,
+      "loss": 1.0068,
+      "step": 15163
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.984104201305779e-06,
+      "loss": 0.9837,
+      "step": 15164
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.98284730047442e-06,
+      "loss": 1.006,
+      "step": 15165
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.981590475313544e-06,
+      "loss": 1.0345,
+      "step": 15166
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.980333725846823e-06,
+      "loss": 1.0492,
+      "step": 15167
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.979077052097935e-06,
+      "loss": 1.0812,
+      "step": 15168
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.977820454090548e-06,
+      "loss": 1.0863,
+      "step": 15169
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.976563931848331e-06,
+      "loss": 1.0087,
+      "step": 15170
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.975307485394955e-06,
+      "loss": 1.0422,
+      "step": 15171
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.974051114754089e-06,
+      "loss": 0.9729,
+      "step": 15172
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.972794819949395e-06,
+      "loss": 1.0459,
+      "step": 15173
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.971538601004542e-06,
+      "loss": 1.0082,
+      "step": 15174
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.970282457943192e-06,
+      "loss": 1.0012,
+      "step": 15175
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.969026390789002e-06,
+      "loss": 1.0266,
+      "step": 15176
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.967770399565638e-06,
+      "loss": 0.991,
+      "step": 15177
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.966514484296755e-06,
+      "loss": 1.0129,
+      "step": 15178
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.96525864500602e-06,
+      "loss": 1.0398,
+      "step": 15179
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.964002881717074e-06,
+      "loss": 1.0329,
+      "step": 15180
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.962747194453582e-06,
+      "loss": 1.0555,
+      "step": 15181
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.961491583239192e-06,
+      "loss": 1.0532,
+      "step": 15182
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.960236048097565e-06,
+      "loss": 1.0397,
+      "step": 15183
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.958980589052337e-06,
+      "loss": 0.9697,
+      "step": 15184
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.957725206127169e-06,
+      "loss": 0.9935,
+      "step": 15185
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.956469899345703e-06,
+      "loss": 1.0415,
+      "step": 15186
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.955214668731587e-06,
+      "loss": 0.9747,
+      "step": 15187
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.95395951430846e-06,
+      "loss": 1.0475,
+      "step": 15188
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.952704436099972e-06,
+      "loss": 1.055,
+      "step": 15189
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.951449434129766e-06,
+      "loss": 1.0441,
+      "step": 15190
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.950194508421473e-06,
+      "loss": 1.0392,
+      "step": 15191
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.94893965899874e-06,
+      "loss": 0.9706,
+      "step": 15192
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.947684885885201e-06,
+      "loss": 1.0,
+      "step": 15193
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.946430189104498e-06,
+      "loss": 1.0404,
+      "step": 15194
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.945175568680253e-06,
+      "loss": 1.0486,
+      "step": 15195
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.94392102463611e-06,
+      "loss": 1.0328,
+      "step": 15196
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.942666556995698e-06,
+      "loss": 1.0518,
+      "step": 15197
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.941412165782645e-06,
+      "loss": 1.0269,
+      "step": 15198
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.94015785102058e-06,
+      "loss": 0.9961,
+      "step": 15199
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.938903612733132e-06,
+      "loss": 1.0863,
+      "step": 15200
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.93764945094393e-06,
+      "loss": 1.0435,
+      "step": 15201
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.93639536567659e-06,
+      "loss": 1.0688,
+      "step": 15202
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.935141356954741e-06,
+      "loss": 1.0295,
+      "step": 15203
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.933887424802003e-06,
+      "loss": 1.0253,
+      "step": 15204
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.932633569242e-06,
+      "loss": 0.9936,
+      "step": 15205
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.931379790298344e-06,
+      "loss": 0.9903,
+      "step": 15206
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.930126087994655e-06,
+      "loss": 0.9954,
+      "step": 15207
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.928872462354552e-06,
+      "loss": 1.043,
+      "step": 15208
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.927618913401644e-06,
+      "loss": 0.9732,
+      "step": 15209
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.926365441159547e-06,
+      "loss": 0.996,
+      "step": 15210
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.92511204565187e-06,
+      "loss": 1.0669,
+      "step": 15211
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.923858726902229e-06,
+      "loss": 1.0821,
+      "step": 15212
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.922605484934228e-06,
+      "loss": 1.0111,
+      "step": 15213
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.921352319771472e-06,
+      "loss": 0.9925,
+      "step": 15214
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.920099231437568e-06,
+      "loss": 1.0636,
+      "step": 15215
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.918846219956125e-06,
+      "loss": 1.003,
+      "step": 15216
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.917593285350742e-06,
+      "loss": 1.036,
+      "step": 15217
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.916340427645019e-06,
+      "loss": 1.0166,
+      "step": 15218
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.915087646862561e-06,
+      "loss": 0.9974,
+      "step": 15219
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.913834943026959e-06,
+      "loss": 1.0107,
+      "step": 15220
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.912582316161816e-06,
+      "loss": 0.9496,
+      "step": 15221
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.911329766290723e-06,
+      "loss": 1.116,
+      "step": 15222
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.910077293437281e-06,
+      "loss": 1.0877,
+      "step": 15223
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.908824897625075e-06,
+      "loss": 1.0277,
+      "step": 15224
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.9075725788777e-06,
+      "loss": 1.013,
+      "step": 15225
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.906320337218743e-06,
+      "loss": 1.0578,
+      "step": 15226
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.905068172671802e-06,
+      "loss": 1.0002,
+      "step": 15227
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.903816085260447e-06,
+      "loss": 0.9962,
+      "step": 15228
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.902564075008279e-06,
+      "loss": 1.0043,
+      "step": 15229
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.901312141938874e-06,
+      "loss": 0.9365,
+      "step": 15230
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.900060286075815e-06,
+      "loss": 0.9739,
+      "step": 15231
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.898808507442683e-06,
+      "loss": 0.9461,
+      "step": 15232
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.89755680606306e-06,
+      "loss": 1.0975,
+      "step": 15233
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.896305181960524e-06,
+      "loss": 0.9821,
+      "step": 15234
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.895053635158648e-06,
+      "loss": 0.9921,
+      "step": 15235
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.89380216568101e-06,
+      "loss": 0.946,
+      "step": 15236
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.892550773551183e-06,
+      "loss": 1.0258,
+      "step": 15237
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.891299458792744e-06,
+      "loss": 1.0845,
+      "step": 15238
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.890048221429253e-06,
+      "loss": 1.0001,
+      "step": 15239
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.888797061484289e-06,
+      "loss": 0.9509,
+      "step": 15240
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.887545978981417e-06,
+      "loss": 0.9924,
+      "step": 15241
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.886294973944202e-06,
+      "loss": 1.0251,
+      "step": 15242
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.885044046396208e-06,
+      "loss": 0.9815,
+      "step": 15243
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.883793196361004e-06,
+      "loss": 1.0012,
+      "step": 15244
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.88254242386215e-06,
+      "loss": 1.0073,
+      "step": 15245
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.881291728923202e-06,
+      "loss": 1.0579,
+      "step": 15246
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.880041111567722e-06,
+      "loss": 1.015,
+      "step": 15247
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.878790571819272e-06,
+      "loss": 1.0048,
+      "step": 15248
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.8775401097014e-06,
+      "loss": 1.0305,
+      "step": 15249
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.876289725237667e-06,
+      "loss": 1.0633,
+      "step": 15250
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.875039418451623e-06,
+      "loss": 1.0445,
+      "step": 15251
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.873789189366825e-06,
+      "loss": 1.0025,
+      "step": 15252
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.872539038006817e-06,
+      "loss": 1.0667,
+      "step": 15253
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.871288964395152e-06,
+      "loss": 1.0297,
+      "step": 15254
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.870038968555371e-06,
+      "loss": 1.0851,
+      "step": 15255
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.868789050511033e-06,
+      "loss": 0.9769,
+      "step": 15256
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.867539210285669e-06,
+      "loss": 1.0066,
+      "step": 15257
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.86628944790283e-06,
+      "loss": 0.984,
+      "step": 15258
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.865039763386055e-06,
+      "loss": 1.0828,
+      "step": 15259
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.86379015675888e-06,
+      "loss": 1.0426,
+      "step": 15260
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.86254062804485e-06,
+      "loss": 1.0129,
+      "step": 15261
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.861291177267502e-06,
+      "loss": 0.9251,
+      "step": 15262
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.86004180445037e-06,
+      "loss": 0.9724,
+      "step": 15263
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.858792509616985e-06,
+      "loss": 1.0525,
+      "step": 15264
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.857543292790884e-06,
+      "loss": 1.0363,
+      "step": 15265
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.856294153995596e-06,
+      "loss": 0.9491,
+      "step": 15266
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.855045093254656e-06,
+      "loss": 0.9795,
+      "step": 15267
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.853796110591583e-06,
+      "loss": 1.0083,
+      "step": 15268
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.852547206029912e-06,
+      "loss": 1.0677,
+      "step": 15269
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.851298379593167e-06,
+      "loss": 0.9632,
+      "step": 15270
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.850049631304869e-06,
+      "loss": 0.9968,
+      "step": 15271
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.848800961188538e-06,
+      "loss": 1.0374,
+      "step": 15272
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.847552369267704e-06,
+      "loss": 1.0284,
+      "step": 15273
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.846303855565882e-06,
+      "loss": 0.9778,
+      "step": 15274
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.845055420106587e-06,
+      "loss": 1.0514,
+      "step": 15275
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.843807062913338e-06,
+      "loss": 1.004,
+      "step": 15276
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.84255878400965e-06,
+      "loss": 1.0018,
+      "step": 15277
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.841310583419041e-06,
+      "loss": 1.0216,
+      "step": 15278
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.840062461165014e-06,
+      "loss": 0.9615,
+      "step": 15279
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.838814417271086e-06,
+      "loss": 1.0048,
+      "step": 15280
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.83756645176077e-06,
+      "loss": 1.0442,
+      "step": 15281
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.83631856465756e-06,
+      "loss": 0.9328,
+      "step": 15282
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.835070755984978e-06,
+      "loss": 1.0045,
+      "step": 15283
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.833823025766518e-06,
+      "loss": 0.9963,
+      "step": 15284
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.832575374025692e-06,
+      "loss": 0.9647,
+      "step": 15285
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.831327800785988e-06,
+      "loss": 0.986,
+      "step": 15286
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.830080306070922e-06,
+      "loss": 0.9925,
+      "step": 15287
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.828832889903983e-06,
+      "loss": 1.0671,
+      "step": 15288
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.827585552308677e-06,
+      "loss": 0.9921,
+      "step": 15289
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.826338293308486e-06,
+      "loss": 0.9826,
+      "step": 15290
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.825091112926916e-06,
+      "loss": 0.9987,
+      "step": 15291
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.823844011187462e-06,
+      "loss": 1.0555,
+      "step": 15292
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.822596988113607e-06,
+      "loss": 1.0723,
+      "step": 15293
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.82135004372884e-06,
+      "loss": 0.9989,
+      "step": 15294
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.820103178056656e-06,
+      "loss": 0.9981,
+      "step": 15295
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.818856391120543e-06,
+      "loss": 0.9699,
+      "step": 15296
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.817609682943982e-06,
+      "loss": 1.0675,
+      "step": 15297
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.816363053550454e-06,
+      "loss": 1.0367,
+      "step": 15298
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.815116502963447e-06,
+      "loss": 1.0208,
+      "step": 15299
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.813870031206448e-06,
+      "loss": 1.0376,
+      "step": 15300
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.81262363830292e-06,
+      "loss": 1.0716,
+      "step": 15301
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.811377324276354e-06,
+      "loss": 0.9964,
+      "step": 15302
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.810131089150228e-06,
+      "loss": 0.9766,
+      "step": 15303
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.808884932948008e-06,
+      "loss": 0.9215,
+      "step": 15304
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.807638855693168e-06,
+      "loss": 1.0512,
+      "step": 15305
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.806392857409189e-06,
+      "loss": 0.9929,
+      "step": 15306
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.80514693811954e-06,
+      "loss": 0.9385,
+      "step": 15307
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.803901097847683e-06,
+      "loss": 0.9766,
+      "step": 15308
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.802655336617086e-06,
+      "loss": 0.9795,
+      "step": 15309
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.801409654451224e-06,
+      "loss": 0.9901,
+      "step": 15310
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.80016405137356e-06,
+      "loss": 1.0772,
+      "step": 15311
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.798918527407548e-06,
+      "loss": 1.0491,
+      "step": 15312
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.797673082576654e-06,
+      "loss": 0.9375,
+      "step": 15313
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.796427716904347e-06,
+      "loss": 1.0042,
+      "step": 15314
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.795182430414075e-06,
+      "loss": 1.0034,
+      "step": 15315
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.7939372231293e-06,
+      "loss": 0.9779,
+      "step": 15316
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.79269209507347e-06,
+      "loss": 1.0795,
+      "step": 15317
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.791447046270056e-06,
+      "loss": 0.9218,
+      "step": 15318
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.790202076742497e-06,
+      "loss": 1.0296,
+      "step": 15319
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.788957186514246e-06,
+      "loss": 1.0299,
+      "step": 15320
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.787712375608754e-06,
+      "loss": 0.9803,
+      "step": 15321
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.786467644049476e-06,
+      "loss": 1.002,
+      "step": 15322
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.785222991859848e-06,
+      "loss": 0.9798,
+      "step": 15323
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.783978419063323e-06,
+      "loss": 0.9492,
+      "step": 15324
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.78273392568334e-06,
+      "loss": 0.9609,
+      "step": 15325
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.781489511743346e-06,
+      "loss": 1.1189,
+      "step": 15326
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.780245177266776e-06,
+      "loss": 1.0577,
+      "step": 15327
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.779000922277076e-06,
+      "loss": 0.9945,
+      "step": 15328
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.777756746797679e-06,
+      "loss": 0.9834,
+      "step": 15329
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.776512650852021e-06,
+      "loss": 1.0041,
+      "step": 15330
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.775268634463542e-06,
+      "loss": 1.0005,
+      "step": 15331
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.77402469765567e-06,
+      "loss": 1.0799,
+      "step": 15332
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.77278084045184e-06,
+      "loss": 0.9403,
+      "step": 15333
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.77153706287548e-06,
+      "loss": 0.9432,
+      "step": 15334
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.77029336495002e-06,
+      "loss": 0.9558,
+      "step": 15335
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.76904974669889e-06,
+      "loss": 1.0,
+      "step": 15336
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.76780620814551e-06,
+      "loss": 0.9517,
+      "step": 15337
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.76656274931331e-06,
+      "loss": 0.9885,
+      "step": 15338
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.765319370225708e-06,
+      "loss": 1.0061,
+      "step": 15339
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.764076070906128e-06,
+      "loss": 0.9772,
+      "step": 15340
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.762832851377991e-06,
+      "loss": 1.0581,
+      "step": 15341
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.761589711664714e-06,
+      "loss": 0.9939,
+      "step": 15342
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.760346651789712e-06,
+      "loss": 1.0041,
+      "step": 15343
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.759103671776404e-06,
+      "loss": 1.0302,
+      "step": 15344
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.757860771648202e-06,
+      "loss": 1.0432,
+      "step": 15345
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.756617951428517e-06,
+      "loss": 0.9694,
+      "step": 15346
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.755375211140762e-06,
+      "loss": 0.9994,
+      "step": 15347
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.754132550808344e-06,
+      "loss": 1.0776,
+      "step": 15348
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.752889970454673e-06,
+      "loss": 0.9762,
+      "step": 15349
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.751647470103154e-06,
+      "loss": 1.0449,
+      "step": 15350
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.750405049777194e-06,
+      "loss": 1.0274,
+      "step": 15351
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.749162709500192e-06,
+      "loss": 1.0528,
+      "step": 15352
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.747920449295554e-06,
+      "loss": 0.9322,
+      "step": 15353
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.746678269186682e-06,
+      "loss": 1.0194,
+      "step": 15354
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.745436169196963e-06,
+      "loss": 0.9643,
+      "step": 15355
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.744194149349806e-06,
+      "loss": 0.9751,
+      "step": 15356
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.742952209668604e-06,
+      "loss": 1.027,
+      "step": 15357
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.74171035017675e-06,
+      "loss": 0.9666,
+      "step": 15358
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.740468570897639e-06,
+      "loss": 1.1194,
+      "step": 15359
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.73922687185466e-06,
+      "loss": 1.0281,
+      "step": 15360
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.737985253071202e-06,
+      "loss": 0.9966,
+      "step": 15361
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.736743714570655e-06,
+      "loss": 0.9922,
+      "step": 15362
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.735502256376403e-06,
+      "loss": 1.013,
+      "step": 15363
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.734260878511834e-06,
+      "loss": 1.0241,
+      "step": 15364
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.733019581000336e-06,
+      "loss": 1.086,
+      "step": 15365
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.731778363865279e-06,
+      "loss": 0.9822,
+      "step": 15366
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.730537227130051e-06,
+      "loss": 0.9699,
+      "step": 15367
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.729296170818032e-06,
+      "loss": 1.0131,
+      "step": 15368
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.728055194952603e-06,
+      "loss": 0.9976,
+      "step": 15369
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.726814299557127e-06,
+      "loss": 1.0213,
+      "step": 15370
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.725573484654992e-06,
+      "loss": 1.0082,
+      "step": 15371
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.724332750269563e-06,
+      "loss": 1.0164,
+      "step": 15372
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.72309209642422e-06,
+      "loss": 1.0031,
+      "step": 15373
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.72185152314232e-06,
+      "loss": 1.0379,
+      "step": 15374
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.72061103044724e-06,
+      "loss": 1.0189,
+      "step": 15375
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.719370618362353e-06,
+      "loss": 1.0355,
+      "step": 15376
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.718130286911012e-06,
+      "loss": 1.0595,
+      "step": 15377
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.716890036116582e-06,
+      "loss": 1.0064,
+      "step": 15378
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.715649866002433e-06,
+      "loss": 0.9794,
+      "step": 15379
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.714409776591929e-06,
+      "loss": 1.0062,
+      "step": 15380
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.713169767908418e-06,
+      "loss": 0.9784,
+      "step": 15381
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.711929839975256e-06,
+      "loss": 0.9844,
+      "step": 15382
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.710689992815812e-06,
+      "loss": 1.0338,
+      "step": 15383
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.70945022645344e-06,
+      "loss": 0.9213,
+      "step": 15384
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.708210540911484e-06,
+      "loss": 1.1233,
+      "step": 15385
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.706970936213294e-06,
+      "loss": 0.9543,
+      "step": 15386
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.705731412382235e-06,
+      "loss": 1.0253,
+      "step": 15387
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.704491969441643e-06,
+      "loss": 1.022,
+      "step": 15388
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.703252607414865e-06,
+      "loss": 1.0029,
+      "step": 15389
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.702013326325256e-06,
+      "loss": 0.9769,
+      "step": 15390
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.700774126196155e-06,
+      "loss": 0.9484,
+      "step": 15391
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.699535007050904e-06,
+      "loss": 1.0081,
+      "step": 15392
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.698295968912842e-06,
+      "loss": 0.9849,
+      "step": 15393
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.697057011805312e-06,
+      "loss": 1.0315,
+      "step": 15394
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.695818135751656e-06,
+      "loss": 0.9917,
+      "step": 15395
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.694579340775202e-06,
+      "loss": 0.997,
+      "step": 15396
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.693340626899283e-06,
+      "loss": 1.0081,
+      "step": 15397
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.69210199414725e-06,
+      "loss": 1.0701,
+      "step": 15398
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.690863442542416e-06,
+      "loss": 1.0703,
+      "step": 15399
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.68962497210812e-06,
+      "loss": 1.0699,
+      "step": 15400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.688386582867683e-06,
+      "loss": 0.9649,
+      "step": 15401
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.687148274844449e-06,
+      "loss": 1.0233,
+      "step": 15402
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.68591004806173e-06,
+      "loss": 1.1172,
+      "step": 15403
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.684671902542853e-06,
+      "loss": 0.9768,
+      "step": 15404
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.683433838311137e-06,
+      "loss": 0.9913,
+      "step": 15405
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.682195855389917e-06,
+      "loss": 1.0419,
+      "step": 15406
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.6809579538025e-06,
+      "loss": 1.1188,
+      "step": 15407
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.679720133572207e-06,
+      "loss": 1.0303,
+      "step": 15408
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.678482394722353e-06,
+      "loss": 1.0033,
+      "step": 15409
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.677244737276258e-06,
+      "loss": 1.0697,
+      "step": 15410
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.676007161257233e-06,
+      "loss": 0.9644,
+      "step": 15411
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.674769666688589e-06,
+      "loss": 1.0141,
+      "step": 15412
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.673532253593636e-06,
+      "loss": 0.9661,
+      "step": 15413
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.672294921995686e-06,
+      "loss": 1.0299,
+      "step": 15414
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.671057671918046e-06,
+      "loss": 1.1177,
+      "step": 15415
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.669820503384019e-06,
+      "loss": 0.9846,
+      "step": 15416
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.668583416416913e-06,
+      "loss": 1.0282,
+      "step": 15417
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.667346411040027e-06,
+      "loss": 1.0305,
+      "step": 15418
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.666109487276666e-06,
+      "loss": 1.0117,
+      "step": 15419
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.664872645150125e-06,
+      "loss": 1.1097,
+      "step": 15420
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.663635884683707e-06,
+      "loss": 1.0799,
+      "step": 15421
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.662399205900708e-06,
+      "loss": 0.9776,
+      "step": 15422
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.66116260882442e-06,
+      "loss": 0.9567,
+      "step": 15423
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.659926093478138e-06,
+      "loss": 1.0504,
+      "step": 15424
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.658689659885158e-06,
+      "loss": 0.9623,
+      "step": 15425
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.657453308068763e-06,
+      "loss": 1.0069,
+      "step": 15426
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.656217038052248e-06,
+      "loss": 0.9624,
+      "step": 15427
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.654980849858899e-06,
+      "loss": 1.0097,
+      "step": 15428
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.653744743512e-06,
+      "loss": 1.0158,
+      "step": 15429
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.652508719034838e-06,
+      "loss": 1.0277,
+      "step": 15430
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.651272776450694e-06,
+      "loss": 0.9627,
+      "step": 15431
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.650036915782849e-06,
+      "loss": 1.1522,
+      "step": 15432
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.648801137054584e-06,
+      "loss": 0.9604,
+      "step": 15433
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.647565440289176e-06,
+      "loss": 0.8898,
+      "step": 15434
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.646329825509904e-06,
+      "loss": 1.0371,
+      "step": 15435
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.645094292740039e-06,
+      "loss": 0.991,
+      "step": 15436
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.643858842002859e-06,
+      "loss": 1.0536,
+      "step": 15437
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.642623473321638e-06,
+      "loss": 1.0014,
+      "step": 15438
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.641388186719631e-06,
+      "loss": 0.94,
+      "step": 15439
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.640152982220126e-06,
+      "loss": 1.0292,
+      "step": 15440
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.638917859846381e-06,
+      "loss": 1.0183,
+      "step": 15441
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.637682819621667e-06,
+      "loss": 1.0018,
+      "step": 15442
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.63644786156924e-06,
+      "loss": 1.05,
+      "step": 15443
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.635212985712367e-06,
+      "loss": 1.0638,
+      "step": 15444
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.63397819207431e-06,
+      "loss": 1.1306,
+      "step": 15445
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.632743480678331e-06,
+      "loss": 1.0151,
+      "step": 15446
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.631508851547682e-06,
+      "loss": 0.9833,
+      "step": 15447
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.630274304705625e-06,
+      "loss": 0.9905,
+      "step": 15448
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.629039840175417e-06,
+      "loss": 0.9562,
+      "step": 15449
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.627805457980298e-06,
+      "loss": 1.0048,
+      "step": 15450
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.626571158143533e-06,
+      "loss": 1.0166,
+      "step": 15451
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.625336940688367e-06,
+      "loss": 0.9357,
+      "step": 15452
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.624102805638057e-06,
+      "loss": 1.0128,
+      "step": 15453
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.622868753015833e-06,
+      "loss": 1.0048,
+      "step": 15454
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.621634782844956e-06,
+      "loss": 0.9248,
+      "step": 15455
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.620400895148663e-06,
+      "loss": 0.9818,
+      "step": 15456
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.619167089950204e-06,
+      "loss": 0.9455,
+      "step": 15457
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.617933367272807e-06,
+      "loss": 0.942,
+      "step": 15458
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.616699727139723e-06,
+      "loss": 1.0701,
+      "step": 15459
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.615466169574191e-06,
+      "loss": 1.0932,
+      "step": 15460
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.614232694599436e-06,
+      "loss": 0.8714,
+      "step": 15461
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.6129993022386964e-06,
+      "loss": 1.0214,
+      "step": 15462
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.611765992515213e-06,
+      "loss": 1.0303,
+      "step": 15463
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.610532765452214e-06,
+      "loss": 0.9706,
+      "step": 15464
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.609299621072925e-06,
+      "loss": 1.007,
+      "step": 15465
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.608066559400573e-06,
+      "loss": 0.9756,
+      "step": 15466
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.606833580458396e-06,
+      "loss": 1.0363,
+      "step": 15467
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.605600684269614e-06,
+      "loss": 0.9958,
+      "step": 15468
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.604367870857447e-06,
+      "loss": 1.0108,
+      "step": 15469
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.603135140245118e-06,
+      "loss": 1.0328,
+      "step": 15470
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.601902492455855e-06,
+      "loss": 0.9957,
+      "step": 15471
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.60066992751287e-06,
+      "loss": 1.0044,
+      "step": 15472
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.599437445439382e-06,
+      "loss": 1.0581,
+      "step": 15473
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.598205046258604e-06,
+      "loss": 0.9743,
+      "step": 15474
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.596972729993762e-06,
+      "loss": 0.9111,
+      "step": 15475
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.595740496668057e-06,
+      "loss": 1.0597,
+      "step": 15476
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.594508346304701e-06,
+      "loss": 0.9796,
+      "step": 15477
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.593276278926913e-06,
+      "loss": 0.9413,
+      "step": 15478
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.592044294557898e-06,
+      "loss": 0.9572,
+      "step": 15479
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.590812393220856e-06,
+      "loss": 0.9878,
+      "step": 15480
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.589580574938995e-06,
+      "loss": 1.0447,
+      "step": 15481
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.5883488397355275e-06,
+      "loss": 0.9372,
+      "step": 15482
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.587117187633641e-06,
+      "loss": 1.1011,
+      "step": 15483
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.585885618656546e-06,
+      "loss": 1.0003,
+      "step": 15484
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.584654132827436e-06,
+      "loss": 1.037,
+      "step": 15485
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.583422730169517e-06,
+      "loss": 1.0159,
+      "step": 15486
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.582191410705973e-06,
+      "loss": 1.0406,
+      "step": 15487
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.580960174460006e-06,
+      "loss": 0.979,
+      "step": 15488
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.579729021454805e-06,
+      "loss": 1.0437,
+      "step": 15489
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.578497951713562e-06,
+      "loss": 1.0623,
+      "step": 15490
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.57726696525947e-06,
+      "loss": 1.0106,
+      "step": 15491
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.5760360621157095e-06,
+      "loss": 1.0039,
+      "step": 15492
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.574805242305473e-06,
+      "loss": 0.9637,
+      "step": 15493
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.573574505851944e-06,
+      "loss": 1.0365,
+      "step": 15494
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.572343852778304e-06,
+      "loss": 1.0848,
+      "step": 15495
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.571113283107735e-06,
+      "loss": 1.0026,
+      "step": 15496
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.569882796863419e-06,
+      "loss": 1.0885,
+      "step": 15497
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.5686523940685325e-06,
+      "loss": 0.9336,
+      "step": 15498
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.567422074746253e-06,
+      "loss": 1.002,
+      "step": 15499
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.5661918389197566e-06,
+      "loss": 1.0062,
+      "step": 15500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.5649616866122165e-06,
+      "loss": 1.0113,
+      "step": 15501
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.563731617846804e-06,
+      "loss": 1.0461,
+      "step": 15502
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.562501632646691e-06,
+      "loss": 0.9821,
+      "step": 15503
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.561271731035045e-06,
+      "loss": 1.0289,
+      "step": 15504
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.5600419130350345e-06,
+      "loss": 1.0065,
+      "step": 15505
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.558812178669827e-06,
+      "loss": 1.022,
+      "step": 15506
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.5575825279625844e-06,
+      "loss": 1.0978,
+      "step": 15507
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.556352960936469e-06,
+      "loss": 1.0476,
+      "step": 15508
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.555123477614646e-06,
+      "loss": 0.9621,
+      "step": 15509
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.55389407802027e-06,
+      "loss": 0.9832,
+      "step": 15510
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.5526647621765e-06,
+      "loss": 1.0853,
+      "step": 15511
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.551435530106496e-06,
+      "loss": 1.0234,
+      "step": 15512
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.550206381833409e-06,
+      "loss": 1.0438,
+      "step": 15513
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.548977317380395e-06,
+      "loss": 1.0063,
+      "step": 15514
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.547748336770605e-06,
+      "loss": 0.9877,
+      "step": 15515
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.546519440027185e-06,
+      "loss": 1.0527,
+      "step": 15516
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.545290627173289e-06,
+      "loss": 0.9492,
+      "step": 15517
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.544061898232063e-06,
+      "loss": 1.0269,
+      "step": 15518
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.542833253226651e-06,
+      "loss": 1.0222,
+      "step": 15519
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.541604692180198e-06,
+      "loss": 1.0115,
+      "step": 15520
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.540376215115844e-06,
+      "loss": 1.0093,
+      "step": 15521
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.539147822056736e-06,
+      "loss": 0.9963,
+      "step": 15522
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.537919513025999e-06,
+      "loss": 1.0506,
+      "step": 15523
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.536691288046785e-06,
+      "loss": 1.0675,
+      "step": 15524
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.535463147142223e-06,
+      "loss": 1.0374,
+      "step": 15525
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.534235090335453e-06,
+      "loss": 1.0697,
+      "step": 15526
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.533007117649598e-06,
+      "loss": 1.1016,
+      "step": 15527
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.531779229107796e-06,
+      "loss": 1.0543,
+      "step": 15528
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.530551424733175e-06,
+      "loss": 0.9675,
+      "step": 15529
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.5293237045488685e-06,
+      "loss": 0.9584,
+      "step": 15530
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.528096068577989e-06,
+      "loss": 1.0067,
+      "step": 15531
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.526868516843673e-06,
+      "loss": 0.9906,
+      "step": 15532
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.525641049369046e-06,
+      "loss": 1.0285,
+      "step": 15533
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.524413666177216e-06,
+      "loss": 1.0092,
+      "step": 15534
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.523186367291314e-06,
+      "loss": 1.0372,
+      "step": 15535
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.521959152734455e-06,
+      "loss": 1.0209,
+      "step": 15536
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.520732022529762e-06,
+      "loss": 1.0512,
+      "step": 15537
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.5195049767003345e-06,
+      "loss": 1.0366,
+      "step": 15538
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.518278015269302e-06,
+      "loss": 1.0292,
+      "step": 15539
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.517051138259771e-06,
+      "loss": 1.0303,
+      "step": 15540
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.515824345694855e-06,
+      "loss": 1.0111,
+      "step": 15541
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.5145976375976536e-06,
+      "loss": 1.0955,
+      "step": 15542
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.513371013991282e-06,
+      "loss": 1.0463,
+      "step": 15543
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.51214447489885e-06,
+      "loss": 1.0498,
+      "step": 15544
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.510918020343452e-06,
+      "loss": 0.9658,
+      "step": 15545
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.509691650348189e-06,
+      "loss": 1.0273,
+      "step": 15546
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.508465364936173e-06,
+      "loss": 1.0669,
+      "step": 15547
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.507239164130501e-06,
+      "loss": 1.077,
+      "step": 15548
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.506013047954263e-06,
+      "loss": 1.0265,
+      "step": 15549
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.504787016430558e-06,
+      "loss": 1.0416,
+      "step": 15550
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.503561069582483e-06,
+      "loss": 0.9565,
+      "step": 15551
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.502335207433136e-06,
+      "loss": 1.0329,
+      "step": 15552
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.5011094300055995e-06,
+      "loss": 1.017,
+      "step": 15553
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.499883737322961e-06,
+      "loss": 0.9659,
+      "step": 15554
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.498658129408323e-06,
+      "loss": 0.9703,
+      "step": 15555
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.497432606284758e-06,
+      "loss": 1.0342,
+      "step": 15556
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.4962071679753565e-06,
+      "loss": 0.9902,
+      "step": 15557
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.494981814503199e-06,
+      "loss": 1.0579,
+      "step": 15558
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.493756545891375e-06,
+      "loss": 1.0076,
+      "step": 15559
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.492531362162958e-06,
+      "loss": 1.0274,
+      "step": 15560
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.4913062633410274e-06,
+      "loss": 1.009,
+      "step": 15561
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.490081249448658e-06,
+      "loss": 1.0082,
+      "step": 15562
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.488856320508934e-06,
+      "loss": 1.0268,
+      "step": 15563
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.487631476544921e-06,
+      "loss": 0.9423,
+      "step": 15564
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.486406717579689e-06,
+      "loss": 0.9946,
+      "step": 15565
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.48518204363632e-06,
+      "loss": 1.0054,
+      "step": 15566
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.483957454737873e-06,
+      "loss": 1.0183,
+      "step": 15567
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.4827329509074195e-06,
+      "loss": 1.085,
+      "step": 15568
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.481508532168019e-06,
+      "loss": 1.0385,
+      "step": 15569
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.480284198542749e-06,
+      "loss": 1.074,
+      "step": 15570
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.47905995005466e-06,
+      "loss": 1.1027,
+      "step": 15571
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.477835786726818e-06,
+      "loss": 1.0235,
+      "step": 15572
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.47661170858228e-06,
+      "loss": 1.0171,
+      "step": 15573
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.475387715644108e-06,
+      "loss": 1.0174,
+      "step": 15574
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.474163807935352e-06,
+      "loss": 0.9106,
+      "step": 15575
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.4729399854790706e-06,
+      "loss": 1.0404,
+      "step": 15576
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.471716248298317e-06,
+      "loss": 1.0333,
+      "step": 15577
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.470492596416141e-06,
+      "loss": 0.9773,
+      "step": 15578
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.469269029855594e-06,
+      "loss": 0.9899,
+      "step": 15579
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.468045548639722e-06,
+      "loss": 1.0503,
+      "step": 15580
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.466822152791574e-06,
+      "loss": 0.9821,
+      "step": 15581
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.465598842334193e-06,
+      "loss": 0.9697,
+      "step": 15582
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.4643756172906196e-06,
+      "loss": 1.0129,
+      "step": 15583
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.463152477683902e-06,
+      "loss": 1.0634,
+      "step": 15584
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.461929423537077e-06,
+      "loss": 1.0036,
+      "step": 15585
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.460706454873182e-06,
+      "loss": 0.9663,
+      "step": 15586
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.459483571715256e-06,
+      "loss": 1.0415,
+      "step": 15587
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.458260774086333e-06,
+      "loss": 0.9722,
+      "step": 15588
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.4570380620094446e-06,
+      "loss": 1.0512,
+      "step": 15589
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.455815435507628e-06,
+      "loss": 0.9438,
+      "step": 15590
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.4545928946039095e-06,
+      "loss": 0.917,
+      "step": 15591
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.453370439321321e-06,
+      "loss": 0.979,
+      "step": 15592
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.4521480696828856e-06,
+      "loss": 0.9975,
+      "step": 15593
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.450925785711631e-06,
+      "loss": 1.0014,
+      "step": 15594
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.449703587430584e-06,
+      "loss": 0.9707,
+      "step": 15595
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.448481474862762e-06,
+      "loss": 0.9764,
+      "step": 15596
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.447259448031188e-06,
+      "loss": 1.0184,
+      "step": 15597
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.446037506958881e-06,
+      "loss": 0.9942,
+      "step": 15598
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.44481565166886e-06,
+      "loss": 0.9949,
+      "step": 15599
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.443593882184139e-06,
+      "loss": 1.0278,
+      "step": 15600
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.4423721985277314e-06,
+      "loss": 1.0111,
+      "step": 15601
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.441150600722653e-06,
+      "loss": 0.9236,
+      "step": 15602
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.439929088791913e-06,
+      "loss": 1.0743,
+      "step": 15603
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.43870766275852e-06,
+      "loss": 0.9709,
+      "step": 15604
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.4374863226454845e-06,
+      "loss": 0.9945,
+      "step": 15605
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.436265068475815e-06,
+      "loss": 0.9688,
+      "step": 15606
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.435043900272504e-06,
+      "loss": 1.0479,
+      "step": 15607
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.433822818058564e-06,
+      "loss": 0.9926,
+      "step": 15608
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.432601821856997e-06,
+      "loss": 1.0888,
+      "step": 15609
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.431380911690804e-06,
+      "loss": 0.9474,
+      "step": 15610
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.430160087582972e-06,
+      "loss": 1.0591,
+      "step": 15611
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.428939349556507e-06,
+      "loss": 1.0112,
+      "step": 15612
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.427718697634405e-06,
+      "loss": 1.0571,
+      "step": 15613
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.426498131839659e-06,
+      "loss": 1.006,
+      "step": 15614
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.425277652195251e-06,
+      "loss": 0.9426,
+      "step": 15615
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.4240572587241805e-06,
+      "loss": 1.0923,
+      "step": 15616
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.422836951449438e-06,
+      "loss": 0.9738,
+      "step": 15617
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.4216167303940005e-06,
+      "loss": 0.9951,
+      "step": 15618
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.420396595580854e-06,
+      "loss": 0.9661,
+      "step": 15619
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.419176547032991e-06,
+      "loss": 1.0261,
+      "step": 15620
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.417956584773392e-06,
+      "loss": 0.9393,
+      "step": 15621
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.416736708825026e-06,
+      "loss": 0.9859,
+      "step": 15622
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.415516919210882e-06,
+      "loss": 0.9839,
+      "step": 15623
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.414297215953936e-06,
+      "loss": 1.0752,
+      "step": 15624
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.413077599077165e-06,
+      "loss": 1.0804,
+      "step": 15625
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.4118580686035305e-06,
+      "loss": 0.9119,
+      "step": 15626
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.410638624556017e-06,
+      "loss": 0.9923,
+      "step": 15627
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.409419266957598e-06,
+      "loss": 1.0036,
+      "step": 15628
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.408199995831229e-06,
+      "loss": 1.0769,
+      "step": 15629
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.4069808111998815e-06,
+      "loss": 0.9591,
+      "step": 15630
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.405761713086528e-06,
+      "loss": 0.9934,
+      "step": 15631
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.40454270151413e-06,
+      "loss": 1.01,
+      "step": 15632
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.403323776505645e-06,
+      "loss": 1.0022,
+      "step": 15633
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.402104938084035e-06,
+      "loss": 0.9827,
+      "step": 15634
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.400886186272261e-06,
+      "loss": 1.0108,
+      "step": 15635
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.3996675210932845e-06,
+      "loss": 1.0033,
+      "step": 15636
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.398448942570054e-06,
+      "loss": 1.0011,
+      "step": 15637
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.397230450725523e-06,
+      "loss": 0.9914,
+      "step": 15638
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.396012045582654e-06,
+      "loss": 1.0193,
+      "step": 15639
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.3947937271643885e-06,
+      "loss": 1.0653,
+      "step": 15640
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.393575495493679e-06,
+      "loss": 0.9772,
+      "step": 15641
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.392357350593469e-06,
+      "loss": 0.9823,
+      "step": 15642
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.391139292486715e-06,
+      "loss": 1.0526,
+      "step": 15643
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.389921321196355e-06,
+      "loss": 0.9802,
+      "step": 15644
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.388703436745327e-06,
+      "loss": 1.0487,
+      "step": 15645
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.387485639156575e-06,
+      "loss": 0.9741,
+      "step": 15646
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.386267928453046e-06,
+      "loss": 0.9383,
+      "step": 15647
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.385050304657669e-06,
+      "loss": 1.0172,
+      "step": 15648
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.383832767793384e-06,
+      "loss": 0.9497,
+      "step": 15649
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.382615317883124e-06,
+      "loss": 0.9641,
+      "step": 15650
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.381397954949822e-06,
+      "loss": 0.9892,
+      "step": 15651
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.380180679016411e-06,
+      "loss": 0.9426,
+      "step": 15652
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.378963490105815e-06,
+      "loss": 0.9407,
+      "step": 15653
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.377746388240975e-06,
+      "loss": 1.0363,
+      "step": 15654
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.376529373444804e-06,
+      "loss": 0.9615,
+      "step": 15655
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.375312445740233e-06,
+      "loss": 0.9231,
+      "step": 15656
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.374095605150183e-06,
+      "loss": 0.9831,
+      "step": 15657
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.372878851697573e-06,
+      "loss": 1.0598,
+      "step": 15658
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.371662185405331e-06,
+      "loss": 0.9995,
+      "step": 15659
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.370445606296368e-06,
+      "loss": 0.998,
+      "step": 15660
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.369229114393602e-06,
+      "loss": 1.0391,
+      "step": 15661
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.368012709719947e-06,
+      "loss": 0.994,
+      "step": 15662
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.36679639229832e-06,
+      "loss": 1.0279,
+      "step": 15663
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.36558016215163e-06,
+      "loss": 1.0262,
+      "step": 15664
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.3643640193027865e-06,
+      "loss": 1.0297,
+      "step": 15665
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.3631479637747e-06,
+      "loss": 0.9785,
+      "step": 15666
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.361931995590274e-06,
+      "loss": 0.9551,
+      "step": 15667
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.360716114772415e-06,
+      "loss": 1.0409,
+      "step": 15668
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.359500321344027e-06,
+      "loss": 1.1163,
+      "step": 15669
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.358284615328011e-06,
+      "loss": 1.0371,
+      "step": 15670
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.357068996747268e-06,
+      "loss": 1.0107,
+      "step": 15671
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.3558534656246955e-06,
+      "loss": 1.0199,
+      "step": 15672
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.35463802198319e-06,
+      "loss": 1.0119,
+      "step": 15673
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.353422665845647e-06,
+      "loss": 1.0124,
+      "step": 15674
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.35220739723496e-06,
+      "loss": 1.1157,
+      "step": 15675
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.3509922161740205e-06,
+      "loss": 1.0608,
+      "step": 15676
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.34977712268572e-06,
+      "loss": 1.0627,
+      "step": 15677
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.348562116792945e-06,
+      "loss": 0.9891,
+      "step": 15678
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.3473471985185886e-06,
+      "loss": 0.9329,
+      "step": 15679
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.346132367885523e-06,
+      "loss": 0.9739,
+      "step": 15680
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.344917624916644e-06,
+      "loss": 0.9757,
+      "step": 15681
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.343702969634828e-06,
+      "loss": 1.0273,
+      "step": 15682
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.342488402062957e-06,
+      "loss": 1.0673,
+      "step": 15683
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.341273922223908e-06,
+      "loss": 1.0441,
+      "step": 15684
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.34005953014056e-06,
+      "loss": 0.9632,
+      "step": 15685
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.3388452258357886e-06,
+      "loss": 1.0225,
+      "step": 15686
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.337631009332464e-06,
+      "loss": 1.0683,
+      "step": 15687
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.3364168806534605e-06,
+      "loss": 0.9858,
+      "step": 15688
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.33520283982165e-06,
+      "loss": 1.0382,
+      "step": 15689
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.333988886859903e-06,
+      "loss": 0.9642,
+      "step": 15690
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.332775021791075e-06,
+      "loss": 1.0418,
+      "step": 15691
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.331561244638044e-06,
+      "loss": 1.131,
+      "step": 15692
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.3303475554236695e-06,
+      "loss": 0.9948,
+      "step": 15693
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.329133954170815e-06,
+      "loss": 1.0127,
+      "step": 15694
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.3279204409023325e-06,
+      "loss": 1.0701,
+      "step": 15695
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.326707015641093e-06,
+      "loss": 1.0399,
+      "step": 15696
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.325493678409947e-06,
+      "loss": 1.0678,
+      "step": 15697
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.3242804292317545e-06,
+      "loss": 1.054,
+      "step": 15698
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.323067268129358e-06,
+      "loss": 0.9747,
+      "step": 15699
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.321854195125621e-06,
+      "loss": 1.0128,
+      "step": 15700
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.3206412102433975e-06,
+      "loss": 1.0606,
+      "step": 15701
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.319428313505523e-06,
+      "loss": 1.055,
+      "step": 15702
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.318215504934846e-06,
+      "loss": 1.0115,
+      "step": 15703
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.317002784554222e-06,
+      "loss": 0.9583,
+      "step": 15704
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.315790152386494e-06,
+      "loss": 1.0558,
+      "step": 15705
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.3145776084544945e-06,
+      "loss": 1.0475,
+      "step": 15706
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.313365152781067e-06,
+      "loss": 1.0899,
+      "step": 15707
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.312152785389055e-06,
+      "loss": 1.0035,
+      "step": 15708
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.310940506301299e-06,
+      "loss": 1.052,
+      "step": 15709
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.3097283155406205e-06,
+      "loss": 1.0116,
+      "step": 15710
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.308516213129864e-06,
+      "loss": 1.0299,
+      "step": 15711
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.307304199091865e-06,
+      "loss": 1.0894,
+      "step": 15712
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.306092273449446e-06,
+      "loss": 0.9854,
+      "step": 15713
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.304880436225433e-06,
+      "loss": 0.9961,
+      "step": 15714
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.30366868744266e-06,
+      "loss": 1.0101,
+      "step": 15715
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.302457027123957e-06,
+      "loss": 1.0754,
+      "step": 15716
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.301245455292139e-06,
+      "loss": 1.1604,
+      "step": 15717
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.300033971970027e-06,
+      "loss": 1.0667,
+      "step": 15718
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.298822577180448e-06,
+      "loss": 1.0199,
+      "step": 15719
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.297611270946224e-06,
+      "loss": 1.0571,
+      "step": 15720
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.29640005329016e-06,
+      "loss": 1.0571,
+      "step": 15721
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.295188924235077e-06,
+      "loss": 0.9795,
+      "step": 15722
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.2939778838037965e-06,
+      "loss": 1.0423,
+      "step": 15723
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.2927669320191215e-06,
+      "loss": 0.9989,
+      "step": 15724
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.291556068903863e-06,
+      "loss": 1.0579,
+      "step": 15725
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.29034529448083e-06,
+      "loss": 0.9975,
+      "step": 15726
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.289134608772839e-06,
+      "loss": 1.0219,
+      "step": 15727
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.287924011802685e-06,
+      "loss": 1.0194,
+      "step": 15728
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.286713503593174e-06,
+      "loss": 1.0199,
+      "step": 15729
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.285503084167106e-06,
+      "loss": 1.0842,
+      "step": 15730
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.28429275354729e-06,
+      "loss": 1.0394,
+      "step": 15731
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.283082511756519e-06,
+      "loss": 1.0291,
+      "step": 15732
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.2818723588175905e-06,
+      "loss": 0.9956,
+      "step": 15733
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.280662294753299e-06,
+      "loss": 1.0598,
+      "step": 15734
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.279452319586438e-06,
+      "loss": 0.9799,
+      "step": 15735
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.278242433339802e-06,
+      "loss": 1.0874,
+      "step": 15736
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.27703263603618e-06,
+      "loss": 0.9758,
+      "step": 15737
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.275822927698362e-06,
+      "loss": 1.052,
+      "step": 15738
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.2746133083491345e-06,
+      "loss": 1.016,
+      "step": 15739
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.273403778011282e-06,
+      "loss": 1.0413,
+      "step": 15740
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.27219433670759e-06,
+      "loss": 1.0471,
+      "step": 15741
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.270984984460838e-06,
+      "loss": 1.0437,
+      "step": 15742
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.2697757212938075e-06,
+      "loss": 1.0394,
+      "step": 15743
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.26856654722928e-06,
+      "loss": 0.9909,
+      "step": 15744
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.2673574622900285e-06,
+      "loss": 1.0447,
+      "step": 15745
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.26614846649883e-06,
+      "loss": 1.0035,
+      "step": 15746
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.2649395598784584e-06,
+      "loss": 1.0144,
+      "step": 15747
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.263730742451685e-06,
+      "loss": 1.0288,
+      "step": 15748
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.262522014241281e-06,
+      "loss": 1.0027,
+      "step": 15749
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.261313375270014e-06,
+      "loss": 0.9727,
+      "step": 15750
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.260104825560653e-06,
+      "loss": 0.9615,
+      "step": 15751
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.258896365135959e-06,
+      "loss": 0.9594,
+      "step": 15752
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.257687994018701e-06,
+      "loss": 1.0227,
+      "step": 15753
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.256479712231636e-06,
+      "loss": 1.063,
+      "step": 15754
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.255271519797527e-06,
+      "loss": 1.0035,
+      "step": 15755
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.254063416739132e-06,
+      "loss": 1.0685,
+      "step": 15756
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.252855403079208e-06,
+      "loss": 0.9905,
+      "step": 15757
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.251647478840511e-06,
+      "loss": 1.0093,
+      "step": 15758
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.2504396440457935e-06,
+      "loss": 0.9774,
+      "step": 15759
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.2492318987178065e-06,
+      "loss": 1.0493,
+      "step": 15760
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.248024242879299e-06,
+      "loss": 1.0166,
+      "step": 15761
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.246816676553023e-06,
+      "loss": 1.0048,
+      "step": 15762
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.245609199761728e-06,
+      "loss": 1.0196,
+      "step": 15763
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.244401812528146e-06,
+      "loss": 0.9555,
+      "step": 15764
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.243194514875032e-06,
+      "loss": 1.079,
+      "step": 15765
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.241987306825128e-06,
+      "loss": 1.022,
+      "step": 15766
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.240780188401172e-06,
+      "loss": 1.0193,
+      "step": 15767
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.239573159625892e-06,
+      "loss": 0.9554,
+      "step": 15768
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.238366220522038e-06,
+      "loss": 0.9904,
+      "step": 15769
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.237159371112341e-06,
+      "loss": 1.0164,
+      "step": 15770
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.235952611419535e-06,
+      "loss": 0.9779,
+      "step": 15771
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.2347459414663484e-06,
+      "loss": 1.0132,
+      "step": 15772
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.233539361275515e-06,
+      "loss": 0.9785,
+      "step": 15773
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.232332870869763e-06,
+      "loss": 0.9655,
+      "step": 15774
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.231126470271808e-06,
+      "loss": 0.993,
+      "step": 15775
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.2299201595043895e-06,
+      "loss": 1.0542,
+      "step": 15776
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.228713938590224e-06,
+      "loss": 0.9981,
+      "step": 15777
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.2275078075520375e-06,
+      "loss": 1.0655,
+      "step": 15778
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.226301766412536e-06,
+      "loss": 1.021,
+      "step": 15779
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.225095815194454e-06,
+      "loss": 1.0202,
+      "step": 15780
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.223889953920499e-06,
+      "loss": 0.9585,
+      "step": 15781
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.222684182613392e-06,
+      "loss": 1.0365,
+      "step": 15782
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.221478501295833e-06,
+      "loss": 1.0269,
+      "step": 15783
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.2202729099905475e-06,
+      "loss": 1.0191,
+      "step": 15784
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.219067408720243e-06,
+      "loss": 1.0762,
+      "step": 15785
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.217861997507618e-06,
+      "loss": 1.0382,
+      "step": 15786
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.216656676375383e-06,
+      "loss": 1.0354,
+      "step": 15787
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.215451445346246e-06,
+      "loss": 1.0368,
+      "step": 15788
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.2142463044429116e-06,
+      "loss": 1.0166,
+      "step": 15789
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.213041253688074e-06,
+      "loss": 0.8918,
+      "step": 15790
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.211836293104428e-06,
+      "loss": 0.9784,
+      "step": 15791
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.210631422714686e-06,
+      "loss": 1.0688,
+      "step": 15792
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.20942664254154e-06,
+      "loss": 1.0263,
+      "step": 15793
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.208221952607675e-06,
+      "loss": 1.0352,
+      "step": 15794
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.2070173529357885e-06,
+      "loss": 1.0174,
+      "step": 15795
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.2058128435485764e-06,
+      "loss": 0.9644,
+      "step": 15796
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.204608424468722e-06,
+      "loss": 1.0125,
+      "step": 15797
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.2034040957189095e-06,
+      "loss": 0.9906,
+      "step": 15798
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.202199857321835e-06,
+      "loss": 0.9621,
+      "step": 15799
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.20099570930018e-06,
+      "loss": 0.9932,
+      "step": 15800
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.19979165167662e-06,
+      "loss": 1.0783,
+      "step": 15801
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.198587684473835e-06,
+      "loss": 1.0964,
+      "step": 15802
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.197383807714513e-06,
+      "loss": 1.0353,
+      "step": 15803
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.196180021421332e-06,
+      "loss": 1.0208,
+      "step": 15804
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.194976325616959e-06,
+      "loss": 0.983,
+      "step": 15805
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.193772720324064e-06,
+      "loss": 0.9754,
+      "step": 15806
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.1925692055653364e-06,
+      "loss": 1.0639,
+      "step": 15807
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.191365781363431e-06,
+      "loss": 0.9524,
+      "step": 15808
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.1901624477410234e-06,
+      "loss": 0.9992,
+      "step": 15809
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.188959204720777e-06,
+      "loss": 0.9547,
+      "step": 15810
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.1877560523253626e-06,
+      "loss": 0.9977,
+      "step": 15811
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.186552990577441e-06,
+      "loss": 0.9863,
+      "step": 15812
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.1853500194996714e-06,
+      "loss": 1.1009,
+      "step": 15813
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.184147139114712e-06,
+      "loss": 1.007,
+      "step": 15814
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.182944349445234e-06,
+      "loss": 1.0656,
+      "step": 15815
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.181741650513883e-06,
+      "loss": 1.0193,
+      "step": 15816
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.180539042343314e-06,
+      "loss": 1.0531,
+      "step": 15817
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.179336524956185e-06,
+      "loss": 1.0084,
+      "step": 15818
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.178134098375145e-06,
+      "loss": 1.0151,
+      "step": 15819
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.176931762622845e-06,
+      "loss": 0.9402,
+      "step": 15820
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.175729517721932e-06,
+      "loss": 1.02,
+      "step": 15821
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.174527363695054e-06,
+      "loss": 1.0615,
+      "step": 15822
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.173325300564855e-06,
+      "loss": 1.0271,
+      "step": 15823
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.172123328353979e-06,
+      "loss": 1.0128,
+      "step": 15824
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.170921447085064e-06,
+      "loss": 0.9747,
+      "step": 15825
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.169719656780754e-06,
+      "loss": 0.9675,
+      "step": 15826
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.168517957463686e-06,
+      "loss": 0.9702,
+      "step": 15827
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.167316349156495e-06,
+      "loss": 0.9808,
+      "step": 15828
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.166114831881814e-06,
+      "loss": 0.9624,
+      "step": 15829
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.16491340566228e-06,
+      "loss": 1.0181,
+      "step": 15830
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.163712070520519e-06,
+      "loss": 0.9858,
+      "step": 15831
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.162510826479166e-06,
+      "loss": 1.0099,
+      "step": 15832
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.161309673560844e-06,
+      "loss": 1.0063,
+      "step": 15833
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.1601086117881805e-06,
+      "loss": 0.9717,
+      "step": 15834
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.1589076411838e-06,
+      "loss": 1.0095,
+      "step": 15835
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.157706761770325e-06,
+      "loss": 1.0003,
+      "step": 15836
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.156505973570376e-06,
+      "loss": 0.978,
+      "step": 15837
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.155305276606574e-06,
+      "loss": 1.1202,
+      "step": 15838
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.154104670901534e-06,
+      "loss": 1.0063,
+      "step": 15839
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.15290415647787e-06,
+      "loss": 1.0798,
+      "step": 15840
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.1517037333582e-06,
+      "loss": 0.975,
+      "step": 15841
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.150503401565134e-06,
+      "loss": 0.9467,
+      "step": 15842
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.149303161121282e-06,
+      "loss": 0.9739,
+      "step": 15843
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.1481030120492555e-06,
+      "loss": 1.015,
+      "step": 15844
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.146902954371657e-06,
+      "loss": 1.0198,
+      "step": 15845
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.145702988111095e-06,
+      "loss": 0.9669,
+      "step": 15846
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.144503113290176e-06,
+      "loss": 1.063,
+      "step": 15847
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.143303329931491e-06,
+      "loss": 0.9926,
+      "step": 15848
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.142103638057651e-06,
+      "loss": 1.044,
+      "step": 15849
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.140904037691252e-06,
+      "loss": 1.0661,
+      "step": 15850
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.139704528854892e-06,
+      "loss": 1.0672,
+      "step": 15851
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.138505111571158e-06,
+      "loss": 1.022,
+      "step": 15852
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.137305785862649e-06,
+      "loss": 1.0623,
+      "step": 15853
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.1361065517519585e-06,
+      "loss": 0.9786,
+      "step": 15854
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.1349074092616774e-06,
+      "loss": 1.0079,
+      "step": 15855
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.133708358414381e-06,
+      "loss": 0.9876,
+      "step": 15856
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.1325093992326704e-06,
+      "loss": 0.9239,
+      "step": 15857
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.1313105317391286e-06,
+      "loss": 0.9671,
+      "step": 15858
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.130111755956327e-06,
+      "loss": 0.9694,
+      "step": 15859
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.128913071906857e-06,
+      "loss": 0.9988,
+      "step": 15860
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.127714479613297e-06,
+      "loss": 0.9849,
+      "step": 15861
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.126515979098228e-06,
+      "loss": 1.0219,
+      "step": 15862
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.12531757038421e-06,
+      "loss": 0.9709,
+      "step": 15863
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.124119253493836e-06,
+      "loss": 1.0237,
+      "step": 15864
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.122921028449668e-06,
+      "loss": 1.0501,
+      "step": 15865
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.121722895274284e-06,
+      "loss": 1.0973,
+      "step": 15866
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.12052485399024e-06,
+      "loss": 1.095,
+      "step": 15867
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.1193269046201195e-06,
+      "loss": 1.0625,
+      "step": 15868
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.118129047186481e-06,
+      "loss": 1.0502,
+      "step": 15869
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.116931281711885e-06,
+      "loss": 1.0301,
+      "step": 15870
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.115733608218893e-06,
+      "loss": 1.0046,
+      "step": 15871
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.114536026730072e-06,
+      "loss": 1.023,
+      "step": 15872
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.113338537267981e-06,
+      "loss": 1.0481,
+      "step": 15873
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.1121411398551716e-06,
+      "loss": 1.0504,
+      "step": 15874
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.110943834514198e-06,
+      "loss": 1.0667,
+      "step": 15875
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.109746621267618e-06,
+      "loss": 1.0034,
+      "step": 15876
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.108549500137985e-06,
+      "loss": 1.0487,
+      "step": 15877
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.1073524711478445e-06,
+      "loss": 1.022,
+      "step": 15878
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.10615553431974e-06,
+      "loss": 0.9998,
+      "step": 15879
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.104958689676234e-06,
+      "loss": 1.0314,
+      "step": 15880
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.103761937239855e-06,
+      "loss": 0.983,
+      "step": 15881
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.102565277033155e-06,
+      "loss": 0.9738,
+      "step": 15882
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.101368709078667e-06,
+      "loss": 1.0272,
+      "step": 15883
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.100172233398946e-06,
+      "loss": 1.0005,
+      "step": 15884
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.098975850016515e-06,
+      "loss": 0.9511,
+      "step": 15885
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.0977795589539105e-06,
+      "loss": 0.9644,
+      "step": 15886
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.096583360233676e-06,
+      "loss": 1.0082,
+      "step": 15887
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.095387253878345e-06,
+      "loss": 1.0029,
+      "step": 15888
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.094191239910438e-06,
+      "loss": 1.0,
+      "step": 15889
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.092995318352484e-06,
+      "loss": 1.0872,
+      "step": 15890
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.0917994892270274e-06,
+      "loss": 1.0503,
+      "step": 15891
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.090603752556576e-06,
+      "loss": 1.0614,
+      "step": 15892
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.089408108363658e-06,
+      "loss": 1.032,
+      "step": 15893
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.0882125566707954e-06,
+      "loss": 1.0621,
+      "step": 15894
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.0870170975005174e-06,
+      "loss": 0.988,
+      "step": 15895
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.085821730875331e-06,
+      "loss": 1.0209,
+      "step": 15896
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.08462645681776e-06,
+      "loss": 1.0257,
+      "step": 15897
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.0834312753503115e-06,
+      "loss": 1.1091,
+      "step": 15898
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.082236186495515e-06,
+      "loss": 0.9557,
+      "step": 15899
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.081041190275868e-06,
+      "loss": 1.0256,
+      "step": 15900
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.0798462867138835e-06,
+      "loss": 1.0492,
+      "step": 15901
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.078651475832071e-06,
+      "loss": 1.0171,
+      "step": 15902
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.077456757652937e-06,
+      "loss": 0.9766,
+      "step": 15903
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.0762621321989864e-06,
+      "loss": 1.0174,
+      "step": 15904
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.075067599492719e-06,
+      "loss": 0.9664,
+      "step": 15905
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.073873159556643e-06,
+      "loss": 1.0344,
+      "step": 15906
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.072678812413251e-06,
+      "loss": 0.9954,
+      "step": 15907
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.071484558085045e-06,
+      "loss": 0.9948,
+      "step": 15908
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.070290396594519e-06,
+      "loss": 1.0295,
+      "step": 15909
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.069096327964166e-06,
+      "loss": 1.0397,
+      "step": 15910
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.067902352216482e-06,
+      "loss": 1.0075,
+      "step": 15911
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.066708469373958e-06,
+      "loss": 1.0702,
+      "step": 15912
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.065514679459078e-06,
+      "loss": 0.9207,
+      "step": 15913
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.064320982494331e-06,
+      "loss": 1.0344,
+      "step": 15914
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.063127378502206e-06,
+      "loss": 0.9417,
+      "step": 15915
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.061933867505184e-06,
+      "loss": 1.0216,
+      "step": 15916
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.060740449525746e-06,
+      "loss": 0.9953,
+      "step": 15917
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.059547124586374e-06,
+      "loss": 1.0238,
+      "step": 15918
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.058353892709546e-06,
+      "loss": 1.0234,
+      "step": 15919
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.057160753917741e-06,
+      "loss": 1.0177,
+      "step": 15920
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.055967708233429e-06,
+      "loss": 0.97,
+      "step": 15921
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.0547747556790864e-06,
+      "loss": 0.915,
+      "step": 15922
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.053581896277184e-06,
+      "loss": 1.0655,
+      "step": 15923
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.052389130050193e-06,
+      "loss": 1.0395,
+      "step": 15924
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.051196457020579e-06,
+      "loss": 1.017,
+      "step": 15925
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.05000387721081e-06,
+      "loss": 1.0587,
+      "step": 15926
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.048811390643348e-06,
+      "loss": 0.9165,
+      "step": 15927
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.047618997340659e-06,
+      "loss": 0.9436,
+      "step": 15928
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.046426697325202e-06,
+      "loss": 1.0661,
+      "step": 15929
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.045234490619435e-06,
+      "loss": 1.0165,
+      "step": 15930
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.044042377245824e-06,
+      "loss": 0.9891,
+      "step": 15931
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.042850357226808e-06,
+      "loss": 1.0208,
+      "step": 15932
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.0416584305848526e-06,
+      "loss": 0.9928,
+      "step": 15933
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.04046659734241e-06,
+      "loss": 1.0277,
+      "step": 15934
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.03927485752193e-06,
+      "loss": 1.0257,
+      "step": 15935
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.038083211145854e-06,
+      "loss": 1.0217,
+      "step": 15936
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.036891658236638e-06,
+      "loss": 1.0523,
+      "step": 15937
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.035700198816721e-06,
+      "loss": 1.0149,
+      "step": 15938
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.034508832908556e-06,
+      "loss": 1.0221,
+      "step": 15939
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.033317560534569e-06,
+      "loss": 0.9909,
+      "step": 15940
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.03212638171721e-06,
+      "loss": 1.0529,
+      "step": 15941
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.030935296478923e-06,
+      "loss": 1.0205,
+      "step": 15942
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.02974430484213e-06,
+      "loss": 1.0581,
+      "step": 15943
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.028553406829268e-06,
+      "loss": 0.9468,
+      "step": 15944
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.027362602462779e-06,
+      "loss": 1.0785,
+      "step": 15945
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.026171891765092e-06,
+      "loss": 0.9659,
+      "step": 15946
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.024981274758624e-06,
+      "loss": 0.9801,
+      "step": 15947
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.023790751465818e-06,
+      "loss": 0.9916,
+      "step": 15948
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.022600321909093e-06,
+      "loss": 0.995,
+      "step": 15949
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.021409986110875e-06,
+      "loss": 1.0783,
+      "step": 15950
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.020219744093578e-06,
+      "loss": 1.0397,
+      "step": 15951
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.019029595879633e-06,
+      "loss": 1.0309,
+      "step": 15952
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.017839541491459e-06,
+      "loss": 0.9792,
+      "step": 15953
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.0166495809514625e-06,
+      "loss": 1.0409,
+      "step": 15954
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.01545971428206e-06,
+      "loss": 0.9813,
+      "step": 15955
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.0142699415056745e-06,
+      "loss": 0.972,
+      "step": 15956
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.013080262644716e-06,
+      "loss": 1.0276,
+      "step": 15957
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.011890677721587e-06,
+      "loss": 0.9775,
+      "step": 15958
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.0107011867586945e-06,
+      "loss": 1.0036,
+      "step": 15959
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.0095117897784545e-06,
+      "loss": 0.9951,
+      "step": 15960
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.008322486803269e-06,
+      "loss": 0.9303,
+      "step": 15961
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.007133277855534e-06,
+      "loss": 1.0181,
+      "step": 15962
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.005944162957652e-06,
+      "loss": 0.9204,
+      "step": 15963
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.004755142132032e-06,
+      "loss": 0.995,
+      "step": 15964
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.003566215401058e-06,
+      "loss": 1.0168,
+      "step": 15965
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.0023773827871346e-06,
+      "loss": 0.9516,
+      "step": 15966
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.001188644312647e-06,
+      "loss": 1.0365,
+      "step": 15967
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.0000000000000024e-06,
+      "loss": 0.9572,
+      "step": 15968
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.998811449871577e-06,
+      "loss": 1.0294,
+      "step": 15969
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.997622993949764e-06,
+      "loss": 1.0175,
+      "step": 15970
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.9964346322569465e-06,
+      "loss": 1.0105,
+      "step": 15971
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.995246364815521e-06,
+      "loss": 1.0427,
+      "step": 15972
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.994058191647858e-06,
+      "loss": 0.98,
+      "step": 15973
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.992870112776342e-06,
+      "loss": 1.0883,
+      "step": 15974
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.991682128223362e-06,
+      "loss": 1.0019,
+      "step": 15975
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.990494238011285e-06,
+      "loss": 0.9646,
+      "step": 15976
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.9893064421624885e-06,
+      "loss": 1.0341,
+      "step": 15977
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.988118740699346e-06,
+      "loss": 0.9903,
+      "step": 15978
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.986931133644241e-06,
+      "loss": 0.9793,
+      "step": 15979
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.985743621019531e-06,
+      "loss": 0.9505,
+      "step": 15980
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.984556202847591e-06,
+      "loss": 1.0192,
+      "step": 15981
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.983368879150787e-06,
+      "loss": 1.1073,
+      "step": 15982
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.9821816499514845e-06,
+      "loss": 1.0021,
+      "step": 15983
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.9809945152720495e-06,
+      "loss": 1.0912,
+      "step": 15984
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.979807475134841e-06,
+      "loss": 1.051,
+      "step": 15985
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.978620529562218e-06,
+      "loss": 1.0209,
+      "step": 15986
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.977433678576541e-06,
+      "loss": 1.0585,
+      "step": 15987
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.976246922200169e-06,
+      "loss": 1.0375,
+      "step": 15988
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.975060260455453e-06,
+      "loss": 1.0101,
+      "step": 15989
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.9738736933647455e-06,
+      "loss": 1.0123,
+      "step": 15990
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.972687220950399e-06,
+      "loss": 0.998,
+      "step": 15991
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.971500843234764e-06,
+      "loss": 0.9476,
+      "step": 15992
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.970314560240188e-06,
+      "loss": 0.9998,
+      "step": 15993
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.969128371989016e-06,
+      "loss": 1.0696,
+      "step": 15994
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.967942278503591e-06,
+      "loss": 0.9858,
+      "step": 15995
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.966756279806255e-06,
+      "loss": 1.0215,
+      "step": 15996
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.9655703759193506e-06,
+      "loss": 1.0999,
+      "step": 15997
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.964384566865216e-06,
+      "loss": 0.9975,
+      "step": 15998
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.963198852666188e-06,
+      "loss": 1.0588,
+      "step": 15999
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.962013233344603e-06,
+      "loss": 0.9436,
+      "step": 16000
+    },
+    {
+      "epoch": 0.68,
+      "eval_text_loss": 0.7689434885978699,
+      "eval_text_runtime": 1.4142,
+      "eval_text_samples_per_second": 52.325,
+      "eval_text_steps_per_second": 1.414,
+      "step": 16000
+    },
+    {
+      "epoch": 0.68,
+      "eval_svit_loss": 1.0955743789672852,
+      "eval_svit_runtime": 4.9564,
+      "eval_svit_samples_per_second": 137.399,
+      "eval_svit_steps_per_second": 2.219,
+      "step": 16000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.96082770892279e-06,
+      "loss": 0.9958,
+      "step": 16001
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.9596422794230845e-06,
+      "loss": 1.0261,
+      "step": 16002
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.958456944867815e-06,
+      "loss": 1.0876,
+      "step": 16003
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.9572717052793096e-06,
+      "loss": 1.0278,
+      "step": 16004
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.9560865606798944e-06,
+      "loss": 0.9869,
+      "step": 16005
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.954901511091894e-06,
+      "loss": 1.0173,
+      "step": 16006
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.953716556537629e-06,
+      "loss": 1.0592,
+      "step": 16007
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.952531697039424e-06,
+      "loss": 1.0499,
+      "step": 16008
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.9513469326195955e-06,
+      "loss": 0.9915,
+      "step": 16009
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.950162263300462e-06,
+      "loss": 1.0282,
+      "step": 16010
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.948977689104336e-06,
+      "loss": 1.0093,
+      "step": 16011
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.947793210053536e-06,
+      "loss": 1.0724,
+      "step": 16012
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.946608826170371e-06,
+      "loss": 1.0125,
+      "step": 16013
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.945424537477149e-06,
+      "loss": 1.0028,
+      "step": 16014
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.944240343996187e-06,
+      "loss": 1.005,
+      "step": 16015
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.943056245749775e-06,
+      "loss": 0.9626,
+      "step": 16016
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.941872242760231e-06,
+      "loss": 1.0592,
+      "step": 16017
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.940688335049857e-06,
+      "loss": 1.0192,
+      "step": 16018
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.939504522640953e-06,
+      "loss": 1.0052,
+      "step": 16019
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.938320805555811e-06,
+      "loss": 0.9832,
+      "step": 16020
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.937137183816737e-06,
+      "loss": 0.9826,
+      "step": 16021
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.935953657446024e-06,
+      "loss": 1.0894,
+      "step": 16022
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.934770226465971e-06,
+      "loss": 0.9465,
+      "step": 16023
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.933586890898857e-06,
+      "loss": 1.0418,
+      "step": 16024
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.9324036507669835e-06,
+      "loss": 0.9531,
+      "step": 16025
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.93122050609264e-06,
+      "loss": 1.0338,
+      "step": 16026
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.930037456898106e-06,
+      "loss": 0.9772,
+      "step": 16027
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.928854503205665e-06,
+      "loss": 1.0017,
+      "step": 16028
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.9276716450376075e-06,
+      "loss": 1.0033,
+      "step": 16029
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.926488882416218e-06,
+      "loss": 1.0111,
+      "step": 16030
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.925306215363766e-06,
+      "loss": 1.0369,
+      "step": 16031
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.924123643902527e-06,
+      "loss": 0.9724,
+      "step": 16032
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.922941168054789e-06,
+      "loss": 1.0047,
+      "step": 16033
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.921758787842822e-06,
+      "loss": 0.9778,
+      "step": 16034
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.920576503288892e-06,
+      "loss": 0.9602,
+      "step": 16035
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.919394314415276e-06,
+      "loss": 1.0415,
+      "step": 16036
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.918212221244244e-06,
+      "loss": 1.0018,
+      "step": 16037
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.917030223798057e-06,
+      "loss": 0.9709,
+      "step": 16038
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.915848322098978e-06,
+      "loss": 1.0292,
+      "step": 16039
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.91466651616928e-06,
+      "loss": 1.0288,
+      "step": 16040
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.913484806031222e-06,
+      "loss": 1.0113,
+      "step": 16041
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.912303191707059e-06,
+      "loss": 1.0751,
+      "step": 16042
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.911121673219045e-06,
+      "loss": 0.9812,
+      "step": 16043
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.909940250589448e-06,
+      "loss": 1.0368,
+      "step": 16044
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.90875892384052e-06,
+      "loss": 0.9929,
+      "step": 16045
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.907577692994505e-06,
+      "loss": 1.0836,
+      "step": 16046
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.9063965580736555e-06,
+      "loss": 0.9747,
+      "step": 16047
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.905215519100231e-06,
+      "loss": 1.0531,
+      "step": 16048
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.904034576096467e-06,
+      "loss": 0.9158,
+      "step": 16049
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.902853729084615e-06,
+      "loss": 1.033,
+      "step": 16050
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.90167297808691e-06,
+      "loss": 1.0307,
+      "step": 16051
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.900492323125608e-06,
+      "loss": 0.9438,
+      "step": 16052
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.899311764222936e-06,
+      "loss": 0.9952,
+      "step": 16053
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.898131301401137e-06,
+      "loss": 1.0128,
+      "step": 16054
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.896950934682445e-06,
+      "loss": 0.983,
+      "step": 16055
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.8957706640891015e-06,
+      "loss": 1.0195,
+      "step": 16056
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.8945904896433306e-06,
+      "loss": 0.9988,
+      "step": 16057
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.893410411367366e-06,
+      "loss": 1.001,
+      "step": 16058
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.892230429283438e-06,
+      "loss": 0.9942,
+      "step": 16059
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.891050543413771e-06,
+      "loss": 1.0002,
+      "step": 16060
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.8898707537805925e-06,
+      "loss": 1.0378,
+      "step": 16061
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.888691060406122e-06,
+      "loss": 1.1378,
+      "step": 16062
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.887511463312594e-06,
+      "loss": 0.9685,
+      "step": 16063
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.886331962522212e-06,
+      "loss": 1.0382,
+      "step": 16064
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.885152558057202e-06,
+      "loss": 1.0757,
+      "step": 16065
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.88397324993978e-06,
+      "loss": 0.9709,
+      "step": 16066
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.8827940381921615e-06,
+      "loss": 0.9831,
+      "step": 16067
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.881614922836555e-06,
+      "loss": 1.0307,
+      "step": 16068
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.880435903895174e-06,
+      "loss": 1.042,
+      "step": 16069
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.87925698139023e-06,
+      "loss": 1.007,
+      "step": 16070
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.878078155343924e-06,
+      "loss": 0.9634,
+      "step": 16071
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.876899425778466e-06,
+      "loss": 1.0012,
+      "step": 16072
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.875720792716059e-06,
+      "loss": 0.9433,
+      "step": 16073
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.874542256178903e-06,
+      "loss": 1.0055,
+      "step": 16074
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.873363816189199e-06,
+      "loss": 1.0399,
+      "step": 16075
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.8721854727691464e-06,
+      "loss": 1.0121,
+      "step": 16076
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.8710072259409395e-06,
+      "loss": 1.0399,
+      "step": 16077
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.869829075726774e-06,
+      "loss": 1.0269,
+      "step": 16078
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.868651022148841e-06,
+      "loss": 1.023,
+      "step": 16079
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.867473065229332e-06,
+      "loss": 1.1187,
+      "step": 16080
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.866295204990437e-06,
+      "loss": 1.0528,
+      "step": 16081
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.865117441454342e-06,
+      "loss": 1.0104,
+      "step": 16082
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.8639397746432336e-06,
+      "loss": 1.0484,
+      "step": 16083
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.862762204579295e-06,
+      "loss": 0.9566,
+      "step": 16084
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.861584731284706e-06,
+      "loss": 1.0464,
+      "step": 16085
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.860407354781648e-06,
+      "loss": 1.0505,
+      "step": 16086
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.859230075092299e-06,
+      "loss": 1.0227,
+      "step": 16087
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.85805289223884e-06,
+      "loss": 0.9642,
+      "step": 16088
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.856875806243432e-06,
+      "loss": 1.0437,
+      "step": 16089
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.85569881712826e-06,
+      "loss": 1.0116,
+      "step": 16090
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.85452192491549e-06,
+      "loss": 1.0557,
+      "step": 16091
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.853345129627294e-06,
+      "loss": 1.0114,
+      "step": 16092
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.852168431285835e-06,
+      "loss": 1.0122,
+      "step": 16093
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.850991829913281e-06,
+      "loss": 1.0202,
+      "step": 16094
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.8498153255317966e-06,
+      "loss": 1.0004,
+      "step": 16095
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.84863891816354e-06,
+      "loss": 1.0551,
+      "step": 16096
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.847462607830671e-06,
+      "loss": 0.9949,
+      "step": 16097
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.8462863945553525e-06,
+      "loss": 0.9984,
+      "step": 16098
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.845110278359738e-06,
+      "loss": 0.9687,
+      "step": 16099
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.843934259265974e-06,
+      "loss": 1.0523,
+      "step": 16100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.842758337296224e-06,
+      "loss": 0.9267,
+      "step": 16101
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.841582512472636e-06,
+      "loss": 0.9915,
+      "step": 16102
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.840406784817362e-06,
+      "loss": 1.049,
+      "step": 16103
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.839231154352534e-06,
+      "loss": 1.0727,
+      "step": 16104
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.8380556211003155e-06,
+      "loss": 1.101,
+      "step": 16105
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.836880185082839e-06,
+      "loss": 1.062,
+      "step": 16106
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.835704846322253e-06,
+      "loss": 1.0643,
+      "step": 16107
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.834529604840686e-06,
+      "loss": 1.0507,
+      "step": 16108
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.8333544606602885e-06,
+      "loss": 1.0003,
+      "step": 16109
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.832179413803192e-06,
+      "loss": 1.0648,
+      "step": 16110
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.831004464291529e-06,
+      "loss": 1.0142,
+      "step": 16111
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.8298296121474274e-06,
+      "loss": 1.0256,
+      "step": 16112
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.828654857393026e-06,
+      "loss": 1.0761,
+      "step": 16113
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.8274802000504565e-06,
+      "loss": 1.0336,
+      "step": 16114
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.826305640141834e-06,
+      "loss": 1.0901,
+      "step": 16115
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.825131177689286e-06,
+      "loss": 1.014,
+      "step": 16116
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.823956812714942e-06,
+      "loss": 0.9241,
+      "step": 16117
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.822782545240925e-06,
+      "loss": 1.0003,
+      "step": 16118
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.821608375289344e-06,
+      "loss": 1.001,
+      "step": 16119
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.82043430288232e-06,
+      "loss": 1.0448,
+      "step": 16120
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.819260328041979e-06,
+      "loss": 0.9524,
+      "step": 16121
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.818086450790423e-06,
+      "loss": 1.0083,
+      "step": 16122
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.8169126711497646e-06,
+      "loss": 0.9606,
+      "step": 16123
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.81573898914212e-06,
+      "loss": 0.9335,
+      "step": 16124
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.8145654047896e-06,
+      "loss": 0.977,
+      "step": 16125
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.813391918114304e-06,
+      "loss": 1.0589,
+      "step": 16126
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.812218529138334e-06,
+      "loss": 0.9989,
+      "step": 16127
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.811045237883804e-06,
+      "loss": 0.9886,
+      "step": 16128
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.809872044372813e-06,
+      "loss": 1.0167,
+      "step": 16129
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.808698948627452e-06,
+      "loss": 1.0099,
+      "step": 16130
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.807525950669819e-06,
+      "loss": 1.0128,
+      "step": 16131
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.806353050522022e-06,
+      "loss": 0.9412,
+      "step": 16132
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.805180248206143e-06,
+      "loss": 0.9293,
+      "step": 16133
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.804007543744276e-06,
+      "loss": 1.0171,
+      "step": 16134
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.80283493715851e-06,
+      "loss": 0.9949,
+      "step": 16135
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.801662428470943e-06,
+      "loss": 0.9665,
+      "step": 16136
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.800490017703649e-06,
+      "loss": 1.0646,
+      "step": 16137
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.7993177048787206e-06,
+      "loss": 0.9932,
+      "step": 16138
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.798145490018231e-06,
+      "loss": 1.0535,
+      "step": 16139
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.796973373144275e-06,
+      "loss": 1.0081,
+      "step": 16140
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.795801354278921e-06,
+      "loss": 0.9934,
+      "step": 16141
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.794629433444249e-06,
+      "loss": 0.9718,
+      "step": 16142
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.7934576106623345e-06,
+      "loss": 1.0068,
+      "step": 16143
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.792285885955251e-06,
+      "loss": 1.0849,
+      "step": 16144
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.79111425934507e-06,
+      "loss": 1.0522,
+      "step": 16145
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.78994273085386e-06,
+      "loss": 0.9784,
+      "step": 16146
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.788771300503692e-06,
+      "loss": 0.9734,
+      "step": 16147
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.78759996831663e-06,
+      "loss": 1.0645,
+      "step": 16148
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.786428734314737e-06,
+      "loss": 1.0656,
+      "step": 16149
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.785257598520077e-06,
+      "loss": 1.0349,
+      "step": 16150
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.7840865609547115e-06,
+      "loss": 1.0151,
+      "step": 16151
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.782915621640697e-06,
+      "loss": 1.0001,
+      "step": 16152
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.781744780600091e-06,
+      "loss": 0.9237,
+      "step": 16153
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.780574037854949e-06,
+      "loss": 1.0126,
+      "step": 16154
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.779403393427324e-06,
+      "loss": 1.0027,
+      "step": 16155
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.778232847339266e-06,
+      "loss": 0.9729,
+      "step": 16156
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.777062399612825e-06,
+      "loss": 0.9907,
+      "step": 16157
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.775892050270051e-06,
+      "loss": 0.9919,
+      "step": 16158
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.774721799332987e-06,
+      "loss": 1.0475,
+      "step": 16159
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.773551646823678e-06,
+      "loss": 1.0134,
+      "step": 16160
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.772381592764164e-06,
+      "loss": 0.9903,
+      "step": 16161
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.771211637176486e-06,
+      "loss": 1.0062,
+      "step": 16162
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.770041780082684e-06,
+      "loss": 1.0226,
+      "step": 16163
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.768872021504795e-06,
+      "loss": 1.0379,
+      "step": 16164
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.767702361464849e-06,
+      "loss": 1.006,
+      "step": 16165
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.766532799984883e-06,
+      "loss": 1.0265,
+      "step": 16166
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.765363337086926e-06,
+      "loss": 1.0429,
+      "step": 16167
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.764193972793006e-06,
+      "loss": 1.0525,
+      "step": 16168
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.763024707125153e-06,
+      "loss": 0.909,
+      "step": 16169
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.761855540105391e-06,
+      "loss": 0.9867,
+      "step": 16170
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.7606864717557415e-06,
+      "loss": 1.0306,
+      "step": 16171
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.759517502098234e-06,
+      "loss": 1.0515,
+      "step": 16172
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.758348631154873e-06,
+      "loss": 0.9727,
+      "step": 16173
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.757179858947688e-06,
+      "loss": 1.0667,
+      "step": 16174
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.756011185498695e-06,
+      "loss": 0.9701,
+      "step": 16175
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.754842610829907e-06,
+      "loss": 0.9526,
+      "step": 16176
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.753674134963328e-06,
+      "loss": 0.9885,
+      "step": 16177
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.752505757920977e-06,
+      "loss": 1.0093,
+      "step": 16178
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.751337479724864e-06,
+      "loss": 1.0083,
+      "step": 16179
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.750169300396989e-06,
+      "loss": 0.9948,
+      "step": 16180
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.749001219959361e-06,
+      "loss": 1.0582,
+      "step": 16181
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.747833238433983e-06,
+      "loss": 1.1009,
+      "step": 16182
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.7466653558428595e-06,
+      "loss": 0.9797,
+      "step": 16183
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.745497572207975e-06,
+      "loss": 1.033,
+      "step": 16184
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.744329887551343e-06,
+      "loss": 1.0037,
+      "step": 16185
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.743162301894953e-06,
+      "loss": 0.9258,
+      "step": 16186
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.741994815260802e-06,
+      "loss": 1.0082,
+      "step": 16187
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.740827427670871e-06,
+      "loss": 0.9523,
+      "step": 16188
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.739660139147162e-06,
+      "loss": 0.9888,
+      "step": 16189
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.738492949711658e-06,
+      "loss": 0.9451,
+      "step": 16190
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.737325859386349e-06,
+      "loss": 1.0502,
+      "step": 16191
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.736158868193209e-06,
+      "loss": 0.9905,
+      "step": 16192
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.7349919761542324e-06,
+      "loss": 1.0403,
+      "step": 16193
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.733825183291396e-06,
+      "loss": 1.0411,
+      "step": 16194
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.732658489626674e-06,
+      "loss": 1.0623,
+      "step": 16195
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.731491895182044e-06,
+      "loss": 0.9906,
+      "step": 16196
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.730325399979487e-06,
+      "loss": 0.9764,
+      "step": 16197
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.729159004040977e-06,
+      "loss": 0.9649,
+      "step": 16198
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.727992707388475e-06,
+      "loss": 0.9803,
+      "step": 16199
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.726826510043953e-06,
+      "loss": 1.0119,
+      "step": 16200
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.725660412029385e-06,
+      "loss": 1.0213,
+      "step": 16201
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.724494413366739e-06,
+      "loss": 0.946,
+      "step": 16202
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.723328514077969e-06,
+      "loss": 0.9925,
+      "step": 16203
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.722162714185034e-06,
+      "loss": 0.9294,
+      "step": 16204
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.72099701370991e-06,
+      "loss": 0.9654,
+      "step": 16205
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.719831412674543e-06,
+      "loss": 1.0188,
+      "step": 16206
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.718665911100891e-06,
+      "loss": 1.0453,
+      "step": 16207
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.717500509010907e-06,
+      "loss": 1.0404,
+      "step": 16208
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.716335206426552e-06,
+      "loss": 1.0486,
+      "step": 16209
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.715170003369767e-06,
+      "loss": 1.0143,
+      "step": 16210
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.714004899862502e-06,
+      "loss": 1.0107,
+      "step": 16211
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.712839895926709e-06,
+      "loss": 0.9811,
+      "step": 16212
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.711674991584335e-06,
+      "loss": 1.0409,
+      "step": 16213
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.710510186857313e-06,
+      "loss": 0.9922,
+      "step": 16214
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.709345481767587e-06,
+      "loss": 1.0107,
+      "step": 16215
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.708180876337107e-06,
+      "loss": 0.9689,
+      "step": 16216
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.707016370587798e-06,
+      "loss": 1.0238,
+      "step": 16217
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.7058519645416e-06,
+      "loss": 1.0491,
+      "step": 16218
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.704687658220443e-06,
+      "loss": 1.0121,
+      "step": 16219
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.703523451646271e-06,
+      "loss": 1.0337,
+      "step": 16220
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.702359344841001e-06,
+      "loss": 1.0056,
+      "step": 16221
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.701195337826566e-06,
+      "loss": 1.0321,
+      "step": 16222
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.700031430624888e-06,
+      "loss": 0.9687,
+      "step": 16223
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.698867623257904e-06,
+      "loss": 0.9927,
+      "step": 16224
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.697703915747523e-06,
+      "loss": 1.035,
+      "step": 16225
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.696540308115672e-06,
+      "loss": 1.0223,
+      "step": 16226
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.695376800384266e-06,
+      "loss": 0.9925,
+      "step": 16227
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.694213392575224e-06,
+      "loss": 1.0573,
+      "step": 16228
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.6930500847104626e-06,
+      "loss": 0.9883,
+      "step": 16229
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.691886876811891e-06,
+      "loss": 1.0071,
+      "step": 16230
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.6907237689014235e-06,
+      "loss": 0.8947,
+      "step": 16231
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.689560761000968e-06,
+      "loss": 1.0271,
+      "step": 16232
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.688397853132432e-06,
+      "loss": 0.9641,
+      "step": 16233
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.687235045317723e-06,
+      "loss": 0.9994,
+      "step": 16234
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.6860723375787405e-06,
+      "loss": 1.0032,
+      "step": 16235
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.68490972993739e-06,
+      "loss": 0.9496,
+      "step": 16236
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.68374722241557e-06,
+      "loss": 0.985,
+      "step": 16237
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.682584815035177e-06,
+      "loss": 1.0454,
+      "step": 16238
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.681422507818109e-06,
+      "loss": 0.9743,
+      "step": 16239
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.6802603007862615e-06,
+      "loss": 1.0178,
+      "step": 16240
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.679098193961524e-06,
+      "loss": 1.0594,
+      "step": 16241
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.677936187365786e-06,
+      "loss": 1.0036,
+      "step": 16242
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.6767742810209416e-06,
+      "loss": 1.0151,
+      "step": 16243
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.675612474948872e-06,
+      "loss": 0.9836,
+      "step": 16244
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.674450769171463e-06,
+      "loss": 1.0603,
+      "step": 16245
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.673289163710599e-06,
+      "loss": 1.0404,
+      "step": 16246
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.672127658588161e-06,
+      "loss": 1.0554,
+      "step": 16247
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.670966253826026e-06,
+      "loss": 1.1251,
+      "step": 16248
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.669804949446074e-06,
+      "loss": 0.9995,
+      "step": 16249
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.668643745470178e-06,
+      "loss": 0.9827,
+      "step": 16250
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.667482641920212e-06,
+      "loss": 0.9687,
+      "step": 16251
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.666321638818049e-06,
+      "loss": 0.9871,
+      "step": 16252
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.665160736185556e-06,
+      "loss": 1.0754,
+      "step": 16253
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.663999934044605e-06,
+      "loss": 1.0389,
+      "step": 16254
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.662839232417055e-06,
+      "loss": 1.0067,
+      "step": 16255
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.6616786313247814e-06,
+      "loss": 1.0689,
+      "step": 16256
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.660518130789628e-06,
+      "loss": 0.9914,
+      "step": 16257
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.659357730833473e-06,
+      "loss": 1.0681,
+      "step": 16258
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.658197431478165e-06,
+      "loss": 0.9692,
+      "step": 16259
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.657037232745569e-06,
+      "loss": 0.9948,
+      "step": 16260
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.655877134657522e-06,
+      "loss": 0.9956,
+      "step": 16261
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.654717137235894e-06,
+      "loss": 1.0493,
+      "step": 16262
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.653557240502531e-06,
+      "loss": 1.0149,
+      "step": 16263
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.652397444479283e-06,
+      "loss": 0.9522,
+      "step": 16264
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.651237749187987e-06,
+      "loss": 1.0122,
+      "step": 16265
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.650078154650499e-06,
+      "loss": 0.9975,
+      "step": 16266
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.648918660888661e-06,
+      "loss": 0.9384,
+      "step": 16267
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.647759267924306e-06,
+      "loss": 0.9075,
+      "step": 16268
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.646599975779281e-06,
+      "loss": 1.0435,
+      "step": 16269
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.645440784475422e-06,
+      "loss": 1.0081,
+      "step": 16270
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.644281694034568e-06,
+      "loss": 0.9848,
+      "step": 16271
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.643122704478542e-06,
+      "loss": 1.0467,
+      "step": 16272
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.641963815829184e-06,
+      "loss": 1.028,
+      "step": 16273
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.640805028108322e-06,
+      "loss": 1.0278,
+      "step": 16274
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.6396463413377885e-06,
+      "loss": 0.9951,
+      "step": 16275
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.638487755539398e-06,
+      "loss": 1.0197,
+      "step": 16276
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.637329270734984e-06,
+      "loss": 0.9744,
+      "step": 16277
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.63617088694637e-06,
+      "loss": 1.0532,
+      "step": 16278
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.63501260419537e-06,
+      "loss": 0.9565,
+      "step": 16279
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.6338544225038e-06,
+      "loss": 1.0464,
+      "step": 16280
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.632696341893486e-06,
+      "loss": 1.0763,
+      "step": 16281
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.631538362386241e-06,
+      "loss": 0.9013,
+      "step": 16282
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.630380484003871e-06,
+      "loss": 0.9751,
+      "step": 16283
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.629222706768185e-06,
+      "loss": 0.992,
+      "step": 16284
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.628065030701005e-06,
+      "loss": 1.0617,
+      "step": 16285
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.626907455824131e-06,
+      "loss": 1.0731,
+      "step": 16286
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.6257499821593635e-06,
+      "loss": 1.0383,
+      "step": 16287
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.6245926097285075e-06,
+      "loss": 1.0,
+      "step": 16288
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.623435338553373e-06,
+      "loss": 0.979,
+      "step": 16289
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.622278168655748e-06,
+      "loss": 0.987,
+      "step": 16290
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.621121100057435e-06,
+      "loss": 0.9993,
+      "step": 16291
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.619964132780225e-06,
+      "loss": 1.017,
+      "step": 16292
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.618807266845923e-06,
+      "loss": 1.0307,
+      "step": 16293
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.61765050227631e-06,
+      "loss": 1.0399,
+      "step": 16294
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.616493839093178e-06,
+      "loss": 0.9247,
+      "step": 16295
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.615337277318313e-06,
+      "loss": 1.0533,
+      "step": 16296
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.614180816973512e-06,
+      "loss": 1.0455,
+      "step": 16297
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.613024458080547e-06,
+      "loss": 1.0286,
+      "step": 16298
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.6118682006612e-06,
+      "loss": 0.9642,
+      "step": 16299
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.6107120447372636e-06,
+      "loss": 1.0038,
+      "step": 16300
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.609555990330504e-06,
+      "loss": 0.9704,
+      "step": 16301
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.608400037462701e-06,
+      "loss": 1.0587,
+      "step": 16302
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.607244186155628e-06,
+      "loss": 0.9751,
+      "step": 16303
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.6060884364310645e-06,
+      "loss": 1.0107,
+      "step": 16304
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.604932788310771e-06,
+      "loss": 0.9551,
+      "step": 16305
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.603777241816524e-06,
+      "loss": 1.0164,
+      "step": 16306
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.602621796970082e-06,
+      "loss": 0.9883,
+      "step": 16307
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.601466453793224e-06,
+      "loss": 1.0482,
+      "step": 16308
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.6003112123077005e-06,
+      "loss": 0.9566,
+      "step": 16309
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.5991560725352776e-06,
+      "loss": 0.9736,
+      "step": 16310
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.598001034497712e-06,
+      "loss": 1.0039,
+      "step": 16311
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.596846098216764e-06,
+      "loss": 0.9711,
+      "step": 16312
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.595691263714186e-06,
+      "loss": 0.9968,
+      "step": 16313
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.594536531011732e-06,
+      "loss": 1.0304,
+      "step": 16314
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.593381900131156e-06,
+      "loss": 1.0296,
+      "step": 16315
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.592227371094206e-06,
+      "loss": 1.039,
+      "step": 16316
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.591072943922629e-06,
+      "loss": 0.9628,
+      "step": 16317
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.589918618638172e-06,
+      "loss": 1.0057,
+      "step": 16318
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.5887643952625786e-06,
+      "loss": 1.0903,
+      "step": 16319
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.58761027381759e-06,
+      "loss": 1.0403,
+      "step": 16320
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.586456254324946e-06,
+      "loss": 1.0382,
+      "step": 16321
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.585302336806387e-06,
+      "loss": 1.0542,
+      "step": 16322
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.584148521283645e-06,
+      "loss": 1.0339,
+      "step": 16323
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.582994807778459e-06,
+      "loss": 1.0496,
+      "step": 16324
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.581841196312558e-06,
+      "loss": 0.9419,
+      "step": 16325
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.580687686907673e-06,
+      "loss": 1.0415,
+      "step": 16326
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.579534279585533e-06,
+      "loss": 1.0294,
+      "step": 16327
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.578380974367866e-06,
+      "loss": 1.0688,
+      "step": 16328
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.577227771276393e-06,
+      "loss": 0.936,
+      "step": 16329
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.57607467033284e-06,
+      "loss": 0.9155,
+      "step": 16330
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.5749216715589264e-06,
+      "loss": 1.0065,
+      "step": 16331
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.5737687749763716e-06,
+      "loss": 1.034,
+      "step": 16332
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.572615980606892e-06,
+      "loss": 1.0553,
+      "step": 16333
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.571463288472202e-06,
+      "loss": 1.0301,
+      "step": 16334
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.570310698594016e-06,
+      "loss": 0.9968,
+      "step": 16335
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.569158210994045e-06,
+      "loss": 1.0436,
+      "step": 16336
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.568005825693998e-06,
+      "loss": 0.9809,
+      "step": 16337
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.566853542715581e-06,
+      "loss": 1.0659,
+      "step": 16338
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.565701362080501e-06,
+      "loss": 1.0558,
+      "step": 16339
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.5645492838104646e-06,
+      "loss": 1.0356,
+      "step": 16340
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.563397307927163e-06,
+      "loss": 0.991,
+      "step": 16341
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.562245434452305e-06,
+      "loss": 0.9834,
+      "step": 16342
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.561093663407588e-06,
+      "loss": 1.0776,
+      "step": 16343
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.559941994814708e-06,
+      "loss": 0.977,
+      "step": 16344
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.558790428695348e-06,
+      "loss": 1.0256,
+      "step": 16345
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.557638965071212e-06,
+      "loss": 0.9703,
+      "step": 16346
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.556487603963987e-06,
+      "loss": 0.9972,
+      "step": 16347
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.555336345395365e-06,
+      "loss": 0.9686,
+      "step": 16348
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.554185189387017e-06,
+      "loss": 0.9719,
+      "step": 16349
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.5530341359606425e-06,
+      "loss": 1.0699,
+      "step": 16350
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.5518831851379234e-06,
+      "loss": 1.0513,
+      "step": 16351
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.55073233694053e-06,
+      "loss": 0.9465,
+      "step": 16352
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.549581591390142e-06,
+      "loss": 0.9355,
+      "step": 16353
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.548430948508443e-06,
+      "loss": 1.028,
+      "step": 16354
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.547280408317109e-06,
+      "loss": 0.913,
+      "step": 16355
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.546129970837799e-06,
+      "loss": 0.9892,
+      "step": 16356
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.544979636092197e-06,
+      "loss": 1.0148,
+      "step": 16357
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.543829404101966e-06,
+      "loss": 0.9715,
+      "step": 16358
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.542679274888778e-06,
+      "loss": 1.0919,
+      "step": 16359
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.541529248474285e-06,
+      "loss": 0.9926,
+      "step": 16360
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.540379324880165e-06,
+      "loss": 1.0356,
+      "step": 16361
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.539229504128073e-06,
+      "loss": 0.9829,
+      "step": 16362
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.538079786239665e-06,
+      "loss": 1.0215,
+      "step": 16363
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.536930171236596e-06,
+      "loss": 1.048,
+      "step": 16364
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.535780659140529e-06,
+      "loss": 0.9665,
+      "step": 16365
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.5346312499731174e-06,
+      "loss": 0.9741,
+      "step": 16366
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.5334819437560045e-06,
+      "loss": 1.0517,
+      "step": 16367
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.532332740510842e-06,
+      "loss": 1.0057,
+      "step": 16368
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.53118364025928e-06,
+      "loss": 1.0787,
+      "step": 16369
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.5300346430229696e-06,
+      "loss": 0.9469,
+      "step": 16370
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.528885748823543e-06,
+      "loss": 0.9831,
+      "step": 16371
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.527736957682642e-06,
+      "loss": 0.9401,
+      "step": 16372
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.526588269621917e-06,
+      "loss": 1.0021,
+      "step": 16373
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.525439684662998e-06,
+      "loss": 1.0895,
+      "step": 16374
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.5242912028275194e-06,
+      "loss": 0.9415,
+      "step": 16375
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.523142824137115e-06,
+      "loss": 0.9543,
+      "step": 16376
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.521994548613428e-06,
+      "loss": 0.9572,
+      "step": 16377
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.520846376278074e-06,
+      "loss": 0.9771,
+      "step": 16378
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.519698307152687e-06,
+      "loss": 1.0527,
+      "step": 16379
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.518550341258889e-06,
+      "loss": 0.9626,
+      "step": 16380
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.517402478618313e-06,
+      "loss": 1.0155,
+      "step": 16381
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.516254719252574e-06,
+      "loss": 1.0494,
+      "step": 16382
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.515107063183293e-06,
+      "loss": 1.0892,
+      "step": 16383
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.513959510432088e-06,
+      "loss": 0.9212,
+      "step": 16384
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.512812061020578e-06,
+      "loss": 1.031,
+      "step": 16385
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.511664714970374e-06,
+      "loss": 1.0395,
+      "step": 16386
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.5105174723030875e-06,
+      "loss": 1.0612,
+      "step": 16387
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.509370333040338e-06,
+      "loss": 1.077,
+      "step": 16388
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.508223297203725e-06,
+      "loss": 1.0808,
+      "step": 16389
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.507076364814854e-06,
+      "loss": 0.98,
+      "step": 16390
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.505929535895332e-06,
+      "loss": 1.0219,
+      "step": 16391
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.50478281046677e-06,
+      "loss": 0.9724,
+      "step": 16392
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.503636188550757e-06,
+      "loss": 0.9786,
+      "step": 16393
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.502489670168896e-06,
+      "loss": 0.9601,
+      "step": 16394
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.501343255342784e-06,
+      "loss": 1.0099,
+      "step": 16395
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.5001969440940155e-06,
+      "loss": 0.991,
+      "step": 16396
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.499050736444183e-06,
+      "loss": 1.0294,
+      "step": 16397
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.497904632414878e-06,
+      "loss": 0.9938,
+      "step": 16398
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.496758632027691e-06,
+      "loss": 0.9601,
+      "step": 16399
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.495612735304206e-06,
+      "loss": 0.9993,
+      "step": 16400
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.494466942266011e-06,
+      "loss": 1.0681,
+      "step": 16401
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.493321252934685e-06,
+      "loss": 1.0781,
+      "step": 16402
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.492175667331814e-06,
+      "loss": 0.9444,
+      "step": 16403
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.491030185478977e-06,
+      "loss": 1.0952,
+      "step": 16404
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.489884807397746e-06,
+      "loss": 0.9989,
+      "step": 16405
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.4887395331097005e-06,
+      "loss": 0.9698,
+      "step": 16406
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.487594362636413e-06,
+      "loss": 0.9972,
+      "step": 16407
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.486449295999455e-06,
+      "loss": 1.0151,
+      "step": 16408
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.485304333220398e-06,
+      "loss": 1.0604,
+      "step": 16409
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.484159474320804e-06,
+      "loss": 0.9795,
+      "step": 16410
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.4830147193222445e-06,
+      "loss": 1.0252,
+      "step": 16411
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.481870068246277e-06,
+      "loss": 0.9809,
+      "step": 16412
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.48072552111447e-06,
+      "loss": 1.0011,
+      "step": 16413
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.4795810779483775e-06,
+      "loss": 1.0197,
+      "step": 16414
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.478436738769562e-06,
+      "loss": 1.0618,
+      "step": 16415
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.477292503599574e-06,
+      "loss": 1.0172,
+      "step": 16416
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.476148372459972e-06,
+      "loss": 0.9705,
+      "step": 16417
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.475004345372305e-06,
+      "loss": 1.0009,
+      "step": 16418
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.473860422358122e-06,
+      "loss": 1.0266,
+      "step": 16419
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.472716603438974e-06,
+      "loss": 1.0169,
+      "step": 16420
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.471572888636406e-06,
+      "loss": 0.9599,
+      "step": 16421
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.47042927797196e-06,
+      "loss": 0.9722,
+      "step": 16422
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.469285771467181e-06,
+      "loss": 0.9648,
+      "step": 16423
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.468142369143609e-06,
+      "loss": 0.9431,
+      "step": 16424
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.466999071022776e-06,
+      "loss": 0.9988,
+      "step": 16425
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.465855877126224e-06,
+      "loss": 0.9989,
+      "step": 16426
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.464712787475488e-06,
+      "loss": 1.0148,
+      "step": 16427
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.463569802092102e-06,
+      "loss": 0.9987,
+      "step": 16428
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.4624269209975824e-06,
+      "loss": 1.0414,
+      "step": 16429
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.461284144213473e-06,
+      "loss": 0.982,
+      "step": 16430
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.460141471761293e-06,
+      "loss": 0.9663,
+      "step": 16431
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.458998903662573e-06,
+      "loss": 0.9801,
+      "step": 16432
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.457856439938824e-06,
+      "loss": 1.0601,
+      "step": 16433
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.456714080611575e-06,
+      "loss": 1.0224,
+      "step": 16434
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.455571825702347e-06,
+      "loss": 1.0227,
+      "step": 16435
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.454429675232649e-06,
+      "loss": 0.9312,
+      "step": 16436
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.453287629223994e-06,
+      "loss": 1.0872,
+      "step": 16437
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.452145687697903e-06,
+      "loss": 0.9621,
+      "step": 16438
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.451003850675885e-06,
+      "loss": 0.9892,
+      "step": 16439
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.449862118179445e-06,
+      "loss": 1.0482,
+      "step": 16440
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.448720490230086e-06,
+      "loss": 0.9612,
+      "step": 16441
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.447578966849321e-06,
+      "loss": 1.0092,
+      "step": 16442
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.446437548058651e-06,
+      "loss": 1.0217,
+      "step": 16443
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.445296233879571e-06,
+      "loss": 1.0435,
+      "step": 16444
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.4441550243335865e-06,
+      "loss": 1.0038,
+      "step": 16445
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.443013919442194e-06,
+      "loss": 0.9843,
+      "step": 16446
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.4418729192268835e-06,
+      "loss": 1.0043,
+      "step": 16447
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.440732023709145e-06,
+      "loss": 0.9849,
+      "step": 16448
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.439591232910478e-06,
+      "loss": 1.0331,
+      "step": 16449
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.438450546852373e-06,
+      "loss": 1.0541,
+      "step": 16450
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.437309965556307e-06,
+      "loss": 1.0778,
+      "step": 16451
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.436169489043767e-06,
+      "loss": 1.0203,
+      "step": 16452
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.435029117336243e-06,
+      "loss": 0.9661,
+      "step": 16453
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.4338888504552145e-06,
+      "loss": 1.0597,
+      "step": 16454
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.4327486884221534e-06,
+      "loss": 1.0897,
+      "step": 16455
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.431608631258539e-06,
+      "loss": 1.0432,
+      "step": 16456
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.4304686789858555e-06,
+      "loss": 0.9836,
+      "step": 16457
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.429328831625565e-06,
+      "loss": 0.9187,
+      "step": 16458
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.428189089199143e-06,
+      "loss": 1.0484,
+      "step": 16459
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.427049451728055e-06,
+      "loss": 0.961,
+      "step": 16460
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.425909919233777e-06,
+      "loss": 1.0938,
+      "step": 16461
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.424770491737767e-06,
+      "loss": 0.9919,
+      "step": 16462
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.42363116926149e-06,
+      "loss": 1.0195,
+      "step": 16463
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.422491951826402e-06,
+      "loss": 1.0211,
+      "step": 16464
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.4213528394539755e-06,
+      "loss": 1.0308,
+      "step": 16465
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.420213832165656e-06,
+      "loss": 1.0074,
+      "step": 16466
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.419074929982902e-06,
+      "loss": 0.9548,
+      "step": 16467
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.417936132927169e-06,
+      "loss": 1.022,
+      "step": 16468
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.416797441019906e-06,
+      "loss": 0.9701,
+      "step": 16469
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.415658854282562e-06,
+      "loss": 1.0158,
+      "step": 16470
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.4145203727365866e-06,
+      "loss": 1.0697,
+      "step": 16471
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.413381996403423e-06,
+      "loss": 1.0337,
+      "step": 16472
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.412243725304516e-06,
+      "loss": 0.9201,
+      "step": 16473
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.411105559461307e-06,
+      "loss": 0.9818,
+      "step": 16474
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.409967498895231e-06,
+      "loss": 1.0064,
+      "step": 16475
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.4088295436277375e-06,
+      "loss": 1.0087,
+      "step": 16476
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.407691693680251e-06,
+      "loss": 1.063,
+      "step": 16477
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.406553949074207e-06,
+      "loss": 0.9481,
+      "step": 16478
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.405416309831037e-06,
+      "loss": 1.0044,
+      "step": 16479
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.404278775972174e-06,
+      "loss": 1.0774,
+      "step": 16480
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.403141347519041e-06,
+      "loss": 0.9118,
+      "step": 16481
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.402004024493069e-06,
+      "loss": 1.0657,
+      "step": 16482
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.400866806915675e-06,
+      "loss": 0.99,
+      "step": 16483
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.399729694808286e-06,
+      "loss": 1.0124,
+      "step": 16484
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.398592688192318e-06,
+      "loss": 0.9483,
+      "step": 16485
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.397455787089191e-06,
+      "loss": 0.9127,
+      "step": 16486
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.39631899152032e-06,
+      "loss": 0.9967,
+      "step": 16487
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.395182301507119e-06,
+      "loss": 1.0038,
+      "step": 16488
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.3940457170709964e-06,
+      "loss": 1.0456,
+      "step": 16489
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.392909238233367e-06,
+      "loss": 0.9579,
+      "step": 16490
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.391772865015635e-06,
+      "loss": 0.9882,
+      "step": 16491
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.390636597439207e-06,
+      "loss": 1.0377,
+      "step": 16492
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.389500435525487e-06,
+      "loss": 1.0153,
+      "step": 16493
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.388364379295878e-06,
+      "loss": 1.0051,
+      "step": 16494
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.387228428771776e-06,
+      "loss": 1.0166,
+      "step": 16495
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.386092583974581e-06,
+      "loss": 1.0328,
+      "step": 16496
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.384956844925694e-06,
+      "loss": 1.0281,
+      "step": 16497
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.383821211646495e-06,
+      "loss": 1.0006,
+      "step": 16498
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.382685684158388e-06,
+      "loss": 1.0012,
+      "step": 16499
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.381550262482759e-06,
+      "loss": 1.0099,
+      "step": 16500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.380414946640995e-06,
+      "loss": 1.0365,
+      "step": 16501
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.379279736654484e-06,
+      "loss": 1.0167,
+      "step": 16502
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.378144632544606e-06,
+      "loss": 1.0057,
+      "step": 16503
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.377009634332747e-06,
+      "loss": 1.0175,
+      "step": 16504
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.375874742040284e-06,
+      "loss": 1.0482,
+      "step": 16505
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.374739955688594e-06,
+      "loss": 1.0702,
+      "step": 16506
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.373605275299057e-06,
+      "loss": 0.9673,
+      "step": 16507
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.372470700893047e-06,
+      "loss": 0.9394,
+      "step": 16508
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.371336232491924e-06,
+      "loss": 1.0633,
+      "step": 16509
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.370201870117073e-06,
+      "loss": 1.0305,
+      "step": 16510
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.369067613789855e-06,
+      "loss": 1.0079,
+      "step": 16511
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.367933463531639e-06,
+      "loss": 1.0363,
+      "step": 16512
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.36679941936378e-06,
+      "loss": 1.0353,
+      "step": 16513
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.365665481307648e-06,
+      "loss": 0.9582,
+      "step": 16514
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.364531649384602e-06,
+      "loss": 1.0374,
+      "step": 16515
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.363397923616002e-06,
+      "loss": 0.9374,
+      "step": 16516
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.3622643040231945e-06,
+      "loss": 1.0501,
+      "step": 16517
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.3611307906275405e-06,
+      "loss": 1.0081,
+      "step": 16518
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.359997383450396e-06,
+      "loss": 1.025,
+      "step": 16519
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.3588640825131005e-06,
+      "loss": 0.9981,
+      "step": 16520
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.357730887837005e-06,
+      "loss": 1.0099,
+      "step": 16521
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.3565977994434594e-06,
+      "loss": 1.0073,
+      "step": 16522
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.3554648173538075e-06,
+      "loss": 1.0233,
+      "step": 16523
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.3543319415893876e-06,
+      "loss": 0.9169,
+      "step": 16524
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.353199172171536e-06,
+      "loss": 0.997,
+      "step": 16525
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.352066509121599e-06,
+      "loss": 1.0021,
+      "step": 16526
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.350933952460911e-06,
+      "loss": 1.0226,
+      "step": 16527
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.3498015022108006e-06,
+      "loss": 1.0116,
+      "step": 16528
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.348669158392599e-06,
+      "loss": 1.0374,
+      "step": 16529
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.3475369210276465e-06,
+      "loss": 0.9657,
+      "step": 16530
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.34640479013726e-06,
+      "loss": 1.0184,
+      "step": 16531
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.345272765742765e-06,
+      "loss": 1.0036,
+      "step": 16532
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.344140847865493e-06,
+      "loss": 0.9097,
+      "step": 16533
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.343009036526766e-06,
+      "loss": 0.9829,
+      "step": 16534
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.341877331747896e-06,
+      "loss": 0.9828,
+      "step": 16535
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.3407457335502e-06,
+      "loss": 1.0538,
+      "step": 16536
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.339614241955003e-06,
+      "loss": 1.0443,
+      "step": 16537
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.338482856983617e-06,
+      "loss": 1.0119,
+      "step": 16538
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.337351578657347e-06,
+      "loss": 1.0289,
+      "step": 16539
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.336220406997503e-06,
+      "loss": 1.0009,
+      "step": 16540
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.335089342025405e-06,
+      "loss": 1.0538,
+      "step": 16541
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.333958383762345e-06,
+      "loss": 0.9135,
+      "step": 16542
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.332827532229629e-06,
+      "loss": 1.0629,
+      "step": 16543
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.33169678744856e-06,
+      "loss": 0.9372,
+      "step": 16544
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.330566149440448e-06,
+      "loss": 1.0212,
+      "step": 16545
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.329435618226575e-06,
+      "loss": 1.0574,
+      "step": 16546
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.328305193828242e-06,
+      "loss": 1.0048,
+      "step": 16547
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.327174876266743e-06,
+      "loss": 0.9453,
+      "step": 16548
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.3260446655633775e-06,
+      "loss": 0.9871,
+      "step": 16549
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.3249145617394225e-06,
+      "loss": 1.0784,
+      "step": 16550
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.323784564816172e-06,
+      "loss": 1.083,
+      "step": 16551
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.322654674814912e-06,
+      "loss": 0.9713,
+      "step": 16552
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.3215248917569215e-06,
+      "loss": 1.0075,
+      "step": 16553
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.320395215663488e-06,
+      "loss": 0.9486,
+      "step": 16554
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.319265646555888e-06,
+      "loss": 0.9969,
+      "step": 16555
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.3181361844553985e-06,
+      "loss": 1.0684,
+      "step": 16556
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.317006829383295e-06,
+      "loss": 1.0519,
+      "step": 16557
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.315877581360854e-06,
+      "loss": 1.0216,
+      "step": 16558
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.314748440409343e-06,
+      "loss": 0.9396,
+      "step": 16559
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.313619406550035e-06,
+      "loss": 1.0317,
+      "step": 16560
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.312490479804195e-06,
+      "loss": 0.9494,
+      "step": 16561
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.311361660193087e-06,
+      "loss": 1.0928,
+      "step": 16562
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.310232947737979e-06,
+      "loss": 1.0606,
+      "step": 16563
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.309104342460128e-06,
+      "loss": 0.9439,
+      "step": 16564
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.307975844380797e-06,
+      "loss": 1.0248,
+      "step": 16565
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.306847453521241e-06,
+      "loss": 1.0142,
+      "step": 16566
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.305719169902715e-06,
+      "loss": 1.0116,
+      "step": 16567
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.304590993546474e-06,
+      "loss": 1.0205,
+      "step": 16568
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.303462924473769e-06,
+      "loss": 0.9909,
+      "step": 16569
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.302334962705847e-06,
+      "loss": 1.0299,
+      "step": 16570
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.301207108263959e-06,
+      "loss": 1.0126,
+      "step": 16571
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.3000793611693465e-06,
+      "loss": 1.0143,
+      "step": 16572
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.298951721443255e-06,
+      "loss": 1.039,
+      "step": 16573
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.297824189106926e-06,
+      "loss": 0.992,
+      "step": 16574
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.296696764181598e-06,
+      "loss": 0.9758,
+      "step": 16575
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.295569446688506e-06,
+      "loss": 1.0278,
+      "step": 16576
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.294442236648886e-06,
+      "loss": 1.0181,
+      "step": 16577
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.2933151340839744e-06,
+      "loss": 1.0229,
+      "step": 16578
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.292188139014999e-06,
+      "loss": 0.9417,
+      "step": 16579
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.291061251463191e-06,
+      "loss": 0.9554,
+      "step": 16580
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.289934471449779e-06,
+      "loss": 0.9852,
+      "step": 16581
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.288807798995978e-06,
+      "loss": 1.0581,
+      "step": 16582
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.287681234123021e-06,
+      "loss": 0.9983,
+      "step": 16583
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.286554776852125e-06,
+      "loss": 1.0543,
+      "step": 16584
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.285428427204515e-06,
+      "loss": 1.1338,
+      "step": 16585
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.284302185201395e-06,
+      "loss": 1.0125,
+      "step": 16586
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.283176050863992e-06,
+      "loss": 1.019,
+      "step": 16587
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.2820500242135145e-06,
+      "loss": 1.0347,
+      "step": 16588
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.2809241052711725e-06,
+      "loss": 0.9938,
+      "step": 16589
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.2797982940581754e-06,
+      "loss": 1.0284,
+      "step": 16590
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.27867259059573e-06,
+      "loss": 1.0321,
+      "step": 16591
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.277546994905045e-06,
+      "loss": 1.1185,
+      "step": 16592
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.276421507007312e-06,
+      "loss": 1.0426,
+      "step": 16593
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.275296126923743e-06,
+      "loss": 0.94,
+      "step": 16594
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.27417085467553e-06,
+      "loss": 1.0508,
+      "step": 16595
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.273045690283878e-06,
+      "loss": 1.0975,
+      "step": 16596
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.271920633769966e-06,
+      "loss": 1.0072,
+      "step": 16597
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.270795685155001e-06,
+      "loss": 1.0112,
+      "step": 16598
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.269670844460169e-06,
+      "loss": 0.9538,
+      "step": 16599
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.268546111706661e-06,
+      "loss": 1.0023,
+      "step": 16600
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.267421486915651e-06,
+      "loss": 1.0295,
+      "step": 16601
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.266296970108339e-06,
+      "loss": 1.0469,
+      "step": 16602
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.265172561305904e-06,
+      "loss": 1.0279,
+      "step": 16603
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.264048260529519e-06,
+      "loss": 1.041,
+      "step": 16604
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.262924067800364e-06,
+      "loss": 0.9877,
+      "step": 16605
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.261799983139621e-06,
+      "loss": 1.0447,
+      "step": 16606
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.260676006568465e-06,
+      "loss": 0.9798,
+      "step": 16607
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.25955213810806e-06,
+      "loss": 1.0016,
+      "step": 16608
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.258428377779577e-06,
+      "loss": 0.9459,
+      "step": 16609
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.257304725604192e-06,
+      "loss": 1.0985,
+      "step": 16610
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.25618118160307e-06,
+      "loss": 1.0087,
+      "step": 16611
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.255057745797368e-06,
+      "loss": 1.045,
+      "step": 16612
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.253934418208246e-06,
+      "loss": 0.9681,
+      "step": 16613
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.252811198856879e-06,
+      "loss": 1.0012,
+      "step": 16614
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.251688087764411e-06,
+      "loss": 0.9771,
+      "step": 16615
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.250565084952002e-06,
+      "loss": 0.9784,
+      "step": 16616
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.2494421904408025e-06,
+      "loss": 1.0415,
+      "step": 16617
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.248319404251976e-06,
+      "loss": 0.9688,
+      "step": 16618
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.247196726406659e-06,
+      "loss": 1.1394,
+      "step": 16619
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.246074156926003e-06,
+      "loss": 1.0751,
+      "step": 16620
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.244951695831157e-06,
+      "loss": 1.0097,
+      "step": 16621
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.243829343143266e-06,
+      "loss": 0.9652,
+      "step": 16622
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.242707098883466e-06,
+      "loss": 0.9916,
+      "step": 16623
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.241584963072894e-06,
+      "loss": 0.9659,
+      "step": 16624
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.240462935732701e-06,
+      "loss": 1.0158,
+      "step": 16625
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.239341016884008e-06,
+      "loss": 1.0005,
+      "step": 16626
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.238219206547957e-06,
+      "loss": 1.0133,
+      "step": 16627
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.237097504745671e-06,
+      "loss": 0.9989,
+      "step": 16628
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.235975911498293e-06,
+      "loss": 0.9649,
+      "step": 16629
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.234854426826937e-06,
+      "loss": 1.0102,
+      "step": 16630
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.233733050752735e-06,
+      "loss": 1.044,
+      "step": 16631
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.232611783296803e-06,
+      "loss": 1.0822,
+      "step": 16632
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.231490624480277e-06,
+      "loss": 1.0447,
+      "step": 16633
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.230369574324262e-06,
+      "loss": 0.9326,
+      "step": 16634
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.22924863284988e-06,
+      "loss": 1.1537,
+      "step": 16635
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.228127800078246e-06,
+      "loss": 1.0419,
+      "step": 16636
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.2270070760304725e-06,
+      "loss": 0.9607,
+      "step": 16637
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.22588646072767e-06,
+      "loss": 0.9754,
+      "step": 16638
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.224765954190951e-06,
+      "loss": 1.1136,
+      "step": 16639
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.223645556441416e-06,
+      "loss": 0.9167,
+      "step": 16640
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.222525267500174e-06,
+      "loss": 0.9611,
+      "step": 16641
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.221405087388328e-06,
+      "loss": 0.8871,
+      "step": 16642
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.220285016126976e-06,
+      "loss": 0.9951,
+      "step": 16643
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.219165053737219e-06,
+      "loss": 1.0748,
+      "step": 16644
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.2180452002401526e-06,
+      "loss": 0.9856,
+      "step": 16645
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.216925455656871e-06,
+      "loss": 1.0102,
+      "step": 16646
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.215805820008467e-06,
+      "loss": 0.9948,
+      "step": 16647
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.214686293316031e-06,
+      "loss": 1.0528,
+      "step": 16648
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.2135668756006526e-06,
+      "loss": 1.0085,
+      "step": 16649
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.2124475668834155e-06,
+      "loss": 0.954,
+      "step": 16650
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.211328367185406e-06,
+      "loss": 0.96,
+      "step": 16651
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.210209276527705e-06,
+      "loss": 0.904,
+      "step": 16652
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.209090294931394e-06,
+      "loss": 1.0218,
+      "step": 16653
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.2079714224175495e-06,
+      "loss": 0.9929,
+      "step": 16654
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.206852659007247e-06,
+      "loss": 1.0491,
+      "step": 16655
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.205734004721565e-06,
+      "loss": 1.0662,
+      "step": 16656
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.20461545958157e-06,
+      "loss": 0.9561,
+      "step": 16657
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.2034970236083356e-06,
+      "loss": 1.0321,
+      "step": 16658
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.202378696822926e-06,
+      "loss": 1.0298,
+      "step": 16659
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.20126047924641e-06,
+      "loss": 0.9705,
+      "step": 16660
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.200142370899851e-06,
+      "loss": 0.9678,
+      "step": 16661
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.199024371804309e-06,
+      "loss": 0.9951,
+      "step": 16662
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.197906481980846e-06,
+      "loss": 1.0045,
+      "step": 16663
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.196788701450517e-06,
+      "loss": 0.9602,
+      "step": 16664
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.195671030234383e-06,
+      "loss": 1.0193,
+      "step": 16665
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.194553468353485e-06,
+      "loss": 1.0327,
+      "step": 16666
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.193436015828885e-06,
+      "loss": 1.0032,
+      "step": 16667
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.192318672681631e-06,
+      "loss": 1.0436,
+      "step": 16668
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.191201438932772e-06,
+      "loss": 1.0284,
+      "step": 16669
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.190084314603342e-06,
+      "loss": 1.0028,
+      "step": 16670
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.1889672997143965e-06,
+      "loss": 1.0436,
+      "step": 16671
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.187850394286971e-06,
+      "loss": 0.9737,
+      "step": 16672
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.18673359834211e-06,
+      "loss": 1.0178,
+      "step": 16673
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.185616911900839e-06,
+      "loss": 0.9429,
+      "step": 16674
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.1845003349842036e-06,
+      "loss": 1.0018,
+      "step": 16675
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.183383867613235e-06,
+      "loss": 1.0249,
+      "step": 16676
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.1822675098089544e-06,
+      "loss": 1.0351,
+      "step": 16677
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.1811512615924025e-06,
+      "loss": 1.0068,
+      "step": 16678
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.180035122984601e-06,
+      "loss": 0.967,
+      "step": 16679
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.1789190940065775e-06,
+      "loss": 1.0758,
+      "step": 16680
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.177803174679346e-06,
+      "loss": 1.0463,
+      "step": 16681
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.176687365023933e-06,
+      "loss": 0.949,
+      "step": 16682
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.175571665061359e-06,
+      "loss": 0.9667,
+      "step": 16683
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.174456074812641e-06,
+      "loss": 1.0034,
+      "step": 16684
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.173340594298782e-06,
+      "loss": 0.968,
+      "step": 16685
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.172225223540804e-06,
+      "loss": 1.0235,
+      "step": 16686
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.171109962559721e-06,
+      "loss": 0.9619,
+      "step": 16687
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.16999481137653e-06,
+      "loss": 1.0136,
+      "step": 16688
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.168879770012237e-06,
+      "loss": 0.9802,
+      "step": 16689
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.167764838487857e-06,
+      "loss": 0.9761,
+      "step": 16690
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.166650016824386e-06,
+      "loss": 1.0021,
+      "step": 16691
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.165535305042822e-06,
+      "loss": 0.8864,
+      "step": 16692
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.16442070316416e-06,
+      "loss": 0.9142,
+      "step": 16693
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.163306211209404e-06,
+      "loss": 0.9739,
+      "step": 16694
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.162191829199545e-06,
+      "loss": 0.9951,
+      "step": 16695
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.161077557155569e-06,
+      "loss": 0.9607,
+      "step": 16696
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.1599633950984665e-06,
+      "loss": 0.9265,
+      "step": 16697
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.158849343049233e-06,
+      "loss": 1.0822,
+      "step": 16698
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.157735401028844e-06,
+      "loss": 1.0048,
+      "step": 16699
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.156621569058288e-06,
+      "loss": 1.0228,
+      "step": 16700
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.15550784715854e-06,
+      "loss": 1.0094,
+      "step": 16701
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.15439423535059e-06,
+      "loss": 1.0455,
+      "step": 16702
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.1532807336554065e-06,
+      "loss": 0.9972,
+      "step": 16703
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.152167342093964e-06,
+      "loss": 1.0148,
+      "step": 16704
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.151054060687236e-06,
+      "loss": 1.037,
+      "step": 16705
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.149940889456202e-06,
+      "loss": 1.0498,
+      "step": 16706
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.148827828421819e-06,
+      "loss": 1.0313,
+      "step": 16707
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.147714877605055e-06,
+      "loss": 0.8908,
+      "step": 16708
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.146602037026885e-06,
+      "loss": 0.985,
+      "step": 16709
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.145489306708261e-06,
+      "loss": 1.0832,
+      "step": 16710
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.144376686670143e-06,
+      "loss": 1.0249,
+      "step": 16711
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.143264176933493e-06,
+      "loss": 0.9929,
+      "step": 16712
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.14215177751927e-06,
+      "loss": 0.9814,
+      "step": 16713
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.1410394884484246e-06,
+      "loss": 0.9781,
+      "step": 16714
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.139927309741906e-06,
+      "loss": 1.0005,
+      "step": 16715
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.1388152414206655e-06,
+      "loss": 0.9951,
+      "step": 16716
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.13770328350566e-06,
+      "loss": 1.0256,
+      "step": 16717
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.136591436017823e-06,
+      "loss": 1.081,
+      "step": 16718
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.1354796989781035e-06,
+      "loss": 1.0043,
+      "step": 16719
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.1343680724074435e-06,
+      "loss": 1.0562,
+      "step": 16720
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.133256556326782e-06,
+      "loss": 1.1095,
+      "step": 16721
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.132145150757055e-06,
+      "loss": 1.0029,
+      "step": 16722
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.1310338557192e-06,
+      "loss": 0.9627,
+      "step": 16723
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.129922671234148e-06,
+      "loss": 0.9541,
+      "step": 16724
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.128811597322833e-06,
+      "loss": 0.9954,
+      "step": 16725
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.127700634006181e-06,
+      "loss": 1.0167,
+      "step": 16726
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.126589781305122e-06,
+      "loss": 1.0265,
+      "step": 16727
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.12547903924058e-06,
+      "loss": 1.016,
+      "step": 16728
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.124368407833477e-06,
+      "loss": 1.0697,
+      "step": 16729
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.123257887104733e-06,
+      "loss": 1.0617,
+      "step": 16730
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.12214747707527e-06,
+      "loss": 0.9455,
+      "step": 16731
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.121037177766002e-06,
+      "loss": 1.0065,
+      "step": 16732
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.119926989197844e-06,
+      "loss": 1.0119,
+      "step": 16733
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.1188169113917085e-06,
+      "loss": 0.9818,
+      "step": 16734
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.117706944368507e-06,
+      "loss": 1.0832,
+      "step": 16735
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.116597088149145e-06,
+      "loss": 0.9673,
+      "step": 16736
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.115487342754532e-06,
+      "loss": 0.9839,
+      "step": 16737
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.114377708205571e-06,
+      "loss": 0.9875,
+      "step": 16738
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.113268184523163e-06,
+      "loss": 0.9452,
+      "step": 16739
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.112158771728209e-06,
+      "loss": 1.025,
+      "step": 16740
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.111049469841605e-06,
+      "loss": 1.0006,
+      "step": 16741
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.109940278884249e-06,
+      "loss": 0.992,
+      "step": 16742
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.108831198877034e-06,
+      "loss": 0.9445,
+      "step": 16743
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.10772222984085e-06,
+      "loss": 1.0024,
+      "step": 16744
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.10661337179659e-06,
+      "loss": 1.0226,
+      "step": 16745
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.105504624765138e-06,
+      "loss": 0.949,
+      "step": 16746
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.10439598876738e-06,
+      "loss": 0.9375,
+      "step": 16747
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.103287463824201e-06,
+      "loss": 1.0616,
+      "step": 16748
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.1021790499564835e-06,
+      "loss": 1.0502,
+      "step": 16749
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.101070747185096e-06,
+      "loss": 1.0637,
+      "step": 16750
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.099962555530928e-06,
+      "loss": 0.9526,
+      "step": 16751
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.09885447501485e-06,
+      "loss": 1.0384,
+      "step": 16752
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.097746505657738e-06,
+      "loss": 1.0008,
+      "step": 16753
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.096638647480449e-06,
+      "loss": 0.9642,
+      "step": 16754
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.095530900503867e-06,
+      "loss": 0.9994,
+      "step": 16755
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.094423264748855e-06,
+      "loss": 0.901,
+      "step": 16756
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.093315740236276e-06,
+      "loss": 0.9407,
+      "step": 16757
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.0922083269869855e-06,
+      "loss": 1.0336,
+      "step": 16758
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.091101025021854e-06,
+      "loss": 0.9389,
+      "step": 16759
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.089993834361738e-06,
+      "loss": 0.9686,
+      "step": 16760
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.0888867550274885e-06,
+      "loss": 1.0134,
+      "step": 16761
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.087779787039957e-06,
+      "loss": 0.9292,
+      "step": 16762
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.086672930420004e-06,
+      "loss": 1.0552,
+      "step": 16763
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.085566185188479e-06,
+      "loss": 1.0456,
+      "step": 16764
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.0844595513662175e-06,
+      "loss": 0.9897,
+      "step": 16765
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.083353028974078e-06,
+      "loss": 0.9897,
+      "step": 16766
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.082246618032899e-06,
+      "loss": 0.9694,
+      "step": 16767
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.081140318563526e-06,
+      "loss": 0.9673,
+      "step": 16768
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.080034130586786e-06,
+      "loss": 0.9696,
+      "step": 16769
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.078928054123529e-06,
+      "loss": 1.0033,
+      "step": 16770
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.077822089194587e-06,
+      "loss": 1.0084,
+      "step": 16771
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.076716235820789e-06,
+      "loss": 0.9919,
+      "step": 16772
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.075610494022965e-06,
+      "loss": 0.9884,
+      "step": 16773
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.0745048638219485e-06,
+      "loss": 0.9703,
+      "step": 16774
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.073399345238568e-06,
+      "loss": 1.0613,
+      "step": 16775
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.072293938293641e-06,
+      "loss": 1.0048,
+      "step": 16776
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.07118864300799e-06,
+      "loss": 0.8234,
+      "step": 16777
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.070083459402442e-06,
+      "loss": 0.951,
+      "step": 16778
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.068978387497814e-06,
+      "loss": 1.0712,
+      "step": 16779
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.067873427314916e-06,
+      "loss": 0.9887,
+      "step": 16780
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.066768578874563e-06,
+      "loss": 1.0242,
+      "step": 16781
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.065663842197576e-06,
+      "loss": 0.9777,
+      "step": 16782
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.064559217304755e-06,
+      "loss": 0.9662,
+      "step": 16783
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.06345470421691e-06,
+      "loss": 1.0048,
+      "step": 16784
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.062350302954843e-06,
+      "loss": 1.0696,
+      "step": 16785
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.061246013539372e-06,
+      "loss": 0.9958,
+      "step": 16786
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.060141835991282e-06,
+      "loss": 0.9912,
+      "step": 16787
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.059037770331378e-06,
+      "loss": 1.007,
+      "step": 16788
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.057933816580455e-06,
+      "loss": 0.928,
+      "step": 16789
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.0568299747593175e-06,
+      "loss": 0.9947,
+      "step": 16790
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.055726244888748e-06,
+      "loss": 0.9311,
+      "step": 16791
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.054622626989539e-06,
+      "loss": 0.942,
+      "step": 16792
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.053519121082483e-06,
+      "loss": 0.9731,
+      "step": 16793
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.052415727188364e-06,
+      "loss": 1.035,
+      "step": 16794
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.051312445327965e-06,
+      "loss": 0.9646,
+      "step": 16795
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.050209275522067e-06,
+      "loss": 1.0792,
+      "step": 16796
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.049106217791464e-06,
+      "loss": 0.8829,
+      "step": 16797
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.048003272156916e-06,
+      "loss": 1.0221,
+      "step": 16798
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.046900438639208e-06,
+      "loss": 0.9749,
+      "step": 16799
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.045797717259108e-06,
+      "loss": 1.0195,
+      "step": 16800
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.0446951080374e-06,
+      "loss": 1.0565,
+      "step": 16801
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.043592610994844e-06,
+      "loss": 1.0395,
+      "step": 16802
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.042490226152208e-06,
+      "loss": 0.9543,
+      "step": 16803
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.041387953530258e-06,
+      "loss": 1.0351,
+      "step": 16804
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.040285793149759e-06,
+      "loss": 1.006,
+      "step": 16805
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.039183745031471e-06,
+      "loss": 1.0022,
+      "step": 16806
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.038081809196155e-06,
+      "loss": 1.0637,
+      "step": 16807
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.036979985664566e-06,
+      "loss": 0.9657,
+      "step": 16808
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.0358782744574595e-06,
+      "loss": 0.9785,
+      "step": 16809
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.034776675595589e-06,
+      "loss": 0.9207,
+      "step": 16810
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.0336751890997034e-06,
+      "loss": 1.0245,
+      "step": 16811
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.032573814990553e-06,
+      "loss": 1.0649,
+      "step": 16812
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.031472553288884e-06,
+      "loss": 1.0556,
+      "step": 16813
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.030371404015437e-06,
+      "loss": 1.0357,
+      "step": 16814
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.029270367190961e-06,
+      "loss": 1.0088,
+      "step": 16815
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.02816944283619e-06,
+      "loss": 1.0712,
+      "step": 16816
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.0270686309718655e-06,
+      "loss": 0.9718,
+      "step": 16817
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.025967931618722e-06,
+      "loss": 0.9622,
+      "step": 16818
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.024867344797492e-06,
+      "loss": 0.9184,
+      "step": 16819
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.0237668705289095e-06,
+      "loss": 1.0061,
+      "step": 16820
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.022666508833704e-06,
+      "loss": 1.0463,
+      "step": 16821
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.021566259732599e-06,
+      "loss": 0.9726,
+      "step": 16822
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.020466123246324e-06,
+      "loss": 1.0387,
+      "step": 16823
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.019366099395602e-06,
+      "loss": 1.0303,
+      "step": 16824
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.018266188201151e-06,
+      "loss": 1.0445,
+      "step": 16825
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.017166389683692e-06,
+      "loss": 0.9463,
+      "step": 16826
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.0160667038639405e-06,
+      "loss": 1.039,
+      "step": 16827
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.0149671307626126e-06,
+      "loss": 1.0422,
+      "step": 16828
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.013867670400418e-06,
+      "loss": 1.0257,
+      "step": 16829
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.012768322798071e-06,
+      "loss": 0.9826,
+      "step": 16830
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.011669087976278e-06,
+      "loss": 1.0518,
+      "step": 16831
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.0105699659557444e-06,
+      "loss": 0.9949,
+      "step": 16832
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.009470956757177e-06,
+      "loss": 1.0331,
+      "step": 16833
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.008372060401271e-06,
+      "loss": 1.0317,
+      "step": 16834
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.00727327690873e-06,
+      "loss": 1.0252,
+      "step": 16835
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.006174606300255e-06,
+      "loss": 0.977,
+      "step": 16836
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.005076048596542e-06,
+      "loss": 0.9824,
+      "step": 16837
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.003977603818274e-06,
+      "loss": 1.0139,
+      "step": 16838
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.002879271986153e-06,
+      "loss": 1.0216,
+      "step": 16839
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.001781053120864e-06,
+      "loss": 0.957,
+      "step": 16840
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.000682947243097e-06,
+      "loss": 1.003,
+      "step": 16841
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.999584954373528e-06,
+      "loss": 0.9548,
+      "step": 16842
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.998487074532848e-06,
+      "loss": 1.0072,
+      "step": 16843
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.99738930774174e-06,
+      "loss": 1.0162,
+      "step": 16844
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.996291654020875e-06,
+      "loss": 1.0717,
+      "step": 16845
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.9951941133909285e-06,
+      "loss": 1.0199,
+      "step": 16846
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.994096685872581e-06,
+      "loss": 1.049,
+      "step": 16847
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.992999371486508e-06,
+      "loss": 0.9441,
+      "step": 16848
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.991902170253368e-06,
+      "loss": 1.0167,
+      "step": 16849
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.9908050821938315e-06,
+      "loss": 1.0243,
+      "step": 16850
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.9897081073285695e-06,
+      "loss": 1.0156,
+      "step": 16851
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.988611245678248e-06,
+      "loss": 0.9917,
+      "step": 16852
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.9875144972635156e-06,
+      "loss": 1.0069,
+      "step": 16853
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.986417862105044e-06,
+      "loss": 0.9916,
+      "step": 16854
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.985321340223486e-06,
+      "loss": 1.0145,
+      "step": 16855
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.984224931639496e-06,
+      "loss": 0.96,
+      "step": 16856
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.9831286363737224e-06,
+      "loss": 0.9433,
+      "step": 16857
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.982032454446826e-06,
+      "loss": 1.0201,
+      "step": 16858
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.980936385879453e-06,
+      "loss": 1.0238,
+      "step": 16859
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.979840430692241e-06,
+      "loss": 0.969,
+      "step": 16860
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.978744588905841e-06,
+      "loss": 1.0178,
+      "step": 16861
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.9776488605408966e-06,
+      "loss": 1.058,
+      "step": 16862
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.976553245618048e-06,
+      "loss": 1.0755,
+      "step": 16863
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.975457744157929e-06,
+      "loss": 1.0339,
+      "step": 16864
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.974362356181172e-06,
+      "loss": 1.0518,
+      "step": 16865
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.973267081708426e-06,
+      "loss": 1.167,
+      "step": 16866
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.972171920760307e-06,
+      "loss": 0.9627,
+      "step": 16867
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.971076873357449e-06,
+      "loss": 1.0051,
+      "step": 16868
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.96998193952048e-06,
+      "loss": 1.0092,
+      "step": 16869
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.968887119270029e-06,
+      "loss": 0.9556,
+      "step": 16870
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.967792412626714e-06,
+      "loss": 1.0639,
+      "step": 16871
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.966697819611157e-06,
+      "loss": 0.9564,
+      "step": 16872
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.9656033402439725e-06,
+      "loss": 1.0129,
+      "step": 16873
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.964508974545787e-06,
+      "loss": 1.0621,
+      "step": 16874
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.963414722537208e-06,
+      "loss": 0.9879,
+      "step": 16875
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.962320584238848e-06,
+      "loss": 1.0653,
+      "step": 16876
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.961226559671319e-06,
+      "loss": 1.0129,
+      "step": 16877
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.960132648855225e-06,
+      "loss": 1.0094,
+      "step": 16878
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.959038851811178e-06,
+      "loss": 1.0204,
+      "step": 16879
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.9579451685597775e-06,
+      "loss": 0.9771,
+      "step": 16880
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.956851599121626e-06,
+      "loss": 1.0156,
+      "step": 16881
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.955758143517324e-06,
+      "loss": 1.0208,
+      "step": 16882
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.9546648017674665e-06,
+      "loss": 1.0407,
+      "step": 16883
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.953571573892646e-06,
+      "loss": 1.1221,
+      "step": 16884
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.952478459913467e-06,
+      "loss": 0.9857,
+      "step": 16885
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.9513854598505085e-06,
+      "loss": 1.0724,
+      "step": 16886
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.950292573724363e-06,
+      "loss": 1.0301,
+      "step": 16887
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.949199801555619e-06,
+      "loss": 0.9905,
+      "step": 16888
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.948107143364857e-06,
+      "loss": 1.0191,
+      "step": 16889
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.947014599172664e-06,
+      "loss": 1.003,
+      "step": 16890
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.945922168999616e-06,
+      "loss": 0.9652,
+      "step": 16891
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.9448298528662925e-06,
+      "loss": 1.0829,
+      "step": 16892
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.9437376507932704e-06,
+      "loss": 0.9452,
+      "step": 16893
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.942645562801121e-06,
+      "loss": 0.9518,
+      "step": 16894
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.941553588910418e-06,
+      "loss": 0.9618,
+      "step": 16895
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.940461729141728e-06,
+      "loss": 0.9664,
+      "step": 16896
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.9393699835156215e-06,
+      "loss": 1.0407,
+      "step": 16897
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.93827835205266e-06,
+      "loss": 1.0262,
+      "step": 16898
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.9371868347734105e-06,
+      "loss": 1.0108,
+      "step": 16899
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.936095431698429e-06,
+      "loss": 0.9982,
+      "step": 16900
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.9350041428482805e-06,
+      "loss": 1.0727,
+      "step": 16901
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.933912968243515e-06,
+      "loss": 1.0114,
+      "step": 16902
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.93282190790469e-06,
+      "loss": 1.0013,
+      "step": 16903
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.931730961852356e-06,
+      "loss": 1.0738,
+      "step": 16904
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.930640130107064e-06,
+      "loss": 0.9849,
+      "step": 16905
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.929549412689369e-06,
+      "loss": 0.9821,
+      "step": 16906
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.928458809619798e-06,
+      "loss": 1.0159,
+      "step": 16907
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.927368320918911e-06,
+      "loss": 0.9925,
+      "step": 16908
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.926277946607247e-06,
+      "loss": 0.9901,
+      "step": 16909
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.92518768670534e-06,
+      "loss": 1.0382,
+      "step": 16910
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.9240975412337295e-06,
+      "loss": 1.0037,
+      "step": 16911
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.923007510212952e-06,
+      "loss": 1.0069,
+      "step": 16912
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.921917593663537e-06,
+      "loss": 1.0446,
+      "step": 16913
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.920827791606018e-06,
+      "loss": 1.0338,
+      "step": 16914
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.919738104060922e-06,
+      "loss": 0.9813,
+      "step": 16915
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.918648531048776e-06,
+      "loss": 1.0249,
+      "step": 16916
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.917559072590107e-06,
+      "loss": 0.9557,
+      "step": 16917
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.916469728705428e-06,
+      "loss": 1.0267,
+      "step": 16918
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.915380499415267e-06,
+      "loss": 0.939,
+      "step": 16919
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.9142913847401394e-06,
+      "loss": 0.9602,
+      "step": 16920
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.913202384700562e-06,
+      "loss": 0.9672,
+      "step": 16921
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.9121134993170415e-06,
+      "loss": 1.0182,
+      "step": 16922
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.911024728610098e-06,
+      "loss": 1.0353,
+      "step": 16923
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.909936072600235e-06,
+      "loss": 1.0399,
+      "step": 16924
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.908847531307966e-06,
+      "loss": 1.0313,
+      "step": 16925
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.907759104753782e-06,
+      "loss": 0.9642,
+      "step": 16926
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.9066707929582e-06,
+      "loss": 1.039,
+      "step": 16927
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.905582595941716e-06,
+      "loss": 0.9819,
+      "step": 16928
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.904494513724823e-06,
+      "loss": 1.033,
+      "step": 16929
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.903406546328019e-06,
+      "loss": 1.0675,
+      "step": 16930
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.902318693771802e-06,
+      "loss": 1.034,
+      "step": 16931
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.9012309560766645e-06,
+      "loss": 1.0647,
+      "step": 16932
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.900143333263091e-06,
+      "loss": 1.0051,
+      "step": 16933
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.899055825351566e-06,
+      "loss": 1.0723,
+      "step": 16934
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.897968432362583e-06,
+      "loss": 1.0622,
+      "step": 16935
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.8968811543166254e-06,
+      "loss": 1.0498,
+      "step": 16936
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.895793991234167e-06,
+      "loss": 0.9802,
+      "step": 16937
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.894706943135685e-06,
+      "loss": 1.0119,
+      "step": 16938
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.893620010041669e-06,
+      "loss": 1.0052,
+      "step": 16939
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.8925331919725824e-06,
+      "loss": 1.0082,
+      "step": 16940
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.891446488948895e-06,
+      "loss": 0.9255,
+      "step": 16941
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.8903599009910855e-06,
+      "loss": 1.0738,
+      "step": 16942
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.8892734281196225e-06,
+      "loss": 1.0403,
+      "step": 16943
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.888187070354963e-06,
+      "loss": 1.0166,
+      "step": 16944
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.887100827717574e-06,
+      "loss": 0.9994,
+      "step": 16945
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.886014700227918e-06,
+      "loss": 1.0201,
+      "step": 16946
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.884928687906461e-06,
+      "loss": 0.9625,
+      "step": 16947
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.883842790773647e-06,
+      "loss": 1.029,
+      "step": 16948
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.8827570088499354e-06,
+      "loss": 0.9952,
+      "step": 16949
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.881671342155787e-06,
+      "loss": 1.0319,
+      "step": 16950
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.880585790711641e-06,
+      "loss": 0.9094,
+      "step": 16951
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.879500354537953e-06,
+      "loss": 1.0366,
+      "step": 16952
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.878415033655162e-06,
+      "loss": 1.0725,
+      "step": 16953
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.877329828083724e-06,
+      "loss": 1.0256,
+      "step": 16954
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.87624473784407e-06,
+      "loss": 1.0108,
+      "step": 16955
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.875159762956644e-06,
+      "loss": 0.9914,
+      "step": 16956
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.87407490344188e-06,
+      "loss": 0.9725,
+      "step": 16957
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.8729901593202216e-06,
+      "loss": 1.008,
+      "step": 16958
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.871905530612095e-06,
+      "loss": 1.0055,
+      "step": 16959
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.8708210173379315e-06,
+      "loss": 1.0608,
+      "step": 16960
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.86973661951816e-06,
+      "loss": 1.0439,
+      "step": 16961
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.86865233717321e-06,
+      "loss": 0.9945,
+      "step": 16962
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.867568170323505e-06,
+      "loss": 1.0499,
+      "step": 16963
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.866484118989466e-06,
+      "loss": 0.9771,
+      "step": 16964
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.865400183191512e-06,
+      "loss": 1.0022,
+      "step": 16965
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.864316362950065e-06,
+      "loss": 1.04,
+      "step": 16966
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.863232658285537e-06,
+      "loss": 1.0277,
+      "step": 16967
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.862149069218343e-06,
+      "loss": 0.97,
+      "step": 16968
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.861065595768895e-06,
+      "loss": 0.9863,
+      "step": 16969
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.8599822379576e-06,
+      "loss": 1.0218,
+      "step": 16970
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.858898995804869e-06,
+      "loss": 1.0457,
+      "step": 16971
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.8578158693311026e-06,
+      "loss": 0.9884,
+      "step": 16972
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.856732858556705e-06,
+      "loss": 1.1059,
+      "step": 16973
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.8556499635020775e-06,
+      "loss": 1.051,
+      "step": 16974
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.854567184187619e-06,
+      "loss": 1.0318,
+      "step": 16975
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.853484520633724e-06,
+      "loss": 1.0259,
+      "step": 16976
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.852401972860787e-06,
+      "loss": 1.0077,
+      "step": 16977
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.8513195408892e-06,
+      "loss": 0.9985,
+      "step": 16978
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.850237224739352e-06,
+      "loss": 1.1001,
+      "step": 16979
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.849155024431632e-06,
+      "loss": 1.063,
+      "step": 16980
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.848072939986423e-06,
+      "loss": 0.9836,
+      "step": 16981
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.846990971424111e-06,
+      "loss": 0.9327,
+      "step": 16982
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.845909118765074e-06,
+      "loss": 0.9647,
+      "step": 16983
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.844827382029692e-06,
+      "loss": 1.0863,
+      "step": 16984
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.84374576123834e-06,
+      "loss": 0.9966,
+      "step": 16985
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.842664256411394e-06,
+      "loss": 0.9937,
+      "step": 16986
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.841582867569226e-06,
+      "loss": 1.0026,
+      "step": 16987
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.840501594732206e-06,
+      "loss": 1.086,
+      "step": 16988
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.839420437920701e-06,
+      "loss": 0.9036,
+      "step": 16989
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.83833939715508e-06,
+      "loss": 1.0608,
+      "step": 16990
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.837258472455699e-06,
+      "loss": 1.0541,
+      "step": 16991
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.836177663842925e-06,
+      "loss": 1.0555,
+      "step": 16992
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.835096971337116e-06,
+      "loss": 1.0854,
+      "step": 16993
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.834016394958632e-06,
+      "loss": 0.9833,
+      "step": 16994
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.832935934727817e-06,
+      "loss": 0.9544,
+      "step": 16995
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.831855590665035e-06,
+      "loss": 0.9915,
+      "step": 16996
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.83077536279063e-06,
+      "loss": 0.9962,
+      "step": 16997
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.829695251124953e-06,
+      "loss": 0.9815,
+      "step": 16998
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.8286152556883485e-06,
+      "loss": 0.958,
+      "step": 16999
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.8275353765011616e-06,
+      "loss": 0.9661,
+      "step": 17000
+    },
+    {
+      "epoch": 0.72,
+      "eval_text_loss": 0.7659300565719604,
+      "eval_text_runtime": 1.4094,
+      "eval_text_samples_per_second": 52.504,
+      "eval_text_steps_per_second": 1.419,
+      "step": 17000
+    },
+    {
+      "epoch": 0.72,
+      "eval_svit_loss": 1.0897903442382812,
+      "eval_svit_runtime": 5.0051,
+      "eval_svit_samples_per_second": 136.062,
+      "eval_svit_steps_per_second": 2.198,
+      "step": 17000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.826455613583736e-06,
+      "loss": 0.9588,
+      "step": 17001
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.825375966956401e-06,
+      "loss": 1.0484,
+      "step": 17002
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.824296436639503e-06,
+      "loss": 1.0536,
+      "step": 17003
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.823217022653376e-06,
+      "loss": 1.1072,
+      "step": 17004
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.822137725018354e-06,
+      "loss": 0.9145,
+      "step": 17005
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.821058543754758e-06,
+      "loss": 1.0432,
+      "step": 17006
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.819979478882926e-06,
+      "loss": 1.0082,
+      "step": 17007
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.818900530423181e-06,
+      "loss": 1.0035,
+      "step": 17008
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.817821698395851e-06,
+      "loss": 1.0144,
+      "step": 17009
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.816742982821249e-06,
+      "loss": 0.9872,
+      "step": 17010
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.815664383719702e-06,
+      "loss": 1.0717,
+      "step": 17011
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.814585901111528e-06,
+      "loss": 1.0076,
+      "step": 17012
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.813507535017037e-06,
+      "loss": 1.0437,
+      "step": 17013
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.8124292854565405e-06,
+      "loss": 1.0079,
+      "step": 17014
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.811351152450355e-06,
+      "loss": 0.9697,
+      "step": 17015
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.810273136018792e-06,
+      "loss": 0.9503,
+      "step": 17016
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.809195236182151e-06,
+      "loss": 0.9865,
+      "step": 17017
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.808117452960734e-06,
+      "loss": 1.0788,
+      "step": 17018
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.807039786374851e-06,
+      "loss": 1.0087,
+      "step": 17019
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.805962236444802e-06,
+      "loss": 1.0037,
+      "step": 17020
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.8048848031908765e-06,
+      "loss": 0.9612,
+      "step": 17021
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.803807486633373e-06,
+      "loss": 0.9684,
+      "step": 17022
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.802730286792592e-06,
+      "loss": 1.1353,
+      "step": 17023
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.801653203688814e-06,
+      "loss": 0.9799,
+      "step": 17024
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.800576237342334e-06,
+      "loss": 1.0479,
+      "step": 17025
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.799499387773435e-06,
+      "loss": 1.0286,
+      "step": 17026
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.798422655002408e-06,
+      "loss": 1.016,
+      "step": 17027
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.797346039049529e-06,
+      "loss": 1.0668,
+      "step": 17028
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.796269539935076e-06,
+      "loss": 0.954,
+      "step": 17029
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.795193157679333e-06,
+      "loss": 1.0083,
+      "step": 17030
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.794116892302579e-06,
+      "loss": 1.0866,
+      "step": 17031
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.793040743825077e-06,
+      "loss": 1.0253,
+      "step": 17032
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.791964712267101e-06,
+      "loss": 1.025,
+      "step": 17033
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.790888797648929e-06,
+      "loss": 0.9935,
+      "step": 17034
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.789812999990816e-06,
+      "loss": 1.021,
+      "step": 17035
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.7887373193130326e-06,
+      "loss": 1.0481,
+      "step": 17036
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.787661755635836e-06,
+      "loss": 1.0163,
+      "step": 17037
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.786586308979498e-06,
+      "loss": 1.0395,
+      "step": 17038
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.785510979364267e-06,
+      "loss": 0.9632,
+      "step": 17039
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.7844357668104e-06,
+      "loss": 1.0387,
+      "step": 17040
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.783360671338149e-06,
+      "loss": 1.0171,
+      "step": 17041
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.782285692967776e-06,
+      "loss": 0.9977,
+      "step": 17042
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.781210831719516e-06,
+      "loss": 1.0075,
+      "step": 17043
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.780136087613624e-06,
+      "loss": 1.105,
+      "step": 17044
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.779061460670343e-06,
+      "loss": 1.021,
+      "step": 17045
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.7779869509099165e-06,
+      "loss": 1.045,
+      "step": 17046
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.776912558352583e-06,
+      "loss": 1.0164,
+      "step": 17047
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.775838283018582e-06,
+      "loss": 1.0071,
+      "step": 17048
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.77476412492815e-06,
+      "loss": 1.0007,
+      "step": 17049
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.773690084101519e-06,
+      "loss": 0.9782,
+      "step": 17050
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.772616160558922e-06,
+      "loss": 1.0479,
+      "step": 17051
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.771542354320587e-06,
+      "loss": 1.0098,
+      "step": 17052
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.770468665406743e-06,
+      "loss": 0.9853,
+      "step": 17053
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.769395093837614e-06,
+      "loss": 0.9447,
+      "step": 17054
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.768321639633422e-06,
+      "loss": 0.9595,
+      "step": 17055
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.767248302814388e-06,
+      "loss": 0.9078,
+      "step": 17056
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.76617508340073e-06,
+      "loss": 1.063,
+      "step": 17057
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.7651019814126655e-06,
+      "loss": 1.0192,
+      "step": 17058
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.764028996870407e-06,
+      "loss": 0.9999,
+      "step": 17059
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.762956129794166e-06,
+      "loss": 0.9753,
+      "step": 17060
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.761883380204152e-06,
+      "loss": 1.0593,
+      "step": 17061
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.760810748120573e-06,
+      "loss": 1.0178,
+      "step": 17062
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.759738233563634e-06,
+      "loss": 1.0122,
+      "step": 17063
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.758665836553537e-06,
+      "loss": 0.9891,
+      "step": 17064
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.757593557110483e-06,
+      "loss": 0.9859,
+      "step": 17065
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.756521395254669e-06,
+      "loss": 0.9974,
+      "step": 17066
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.755449351006294e-06,
+      "loss": 0.9922,
+      "step": 17067
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.7543774243855505e-06,
+      "loss": 0.9819,
+      "step": 17068
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.753305615412629e-06,
+      "loss": 1.0406,
+      "step": 17069
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.752233924107721e-06,
+      "loss": 1.0803,
+      "step": 17070
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.7511623504910116e-06,
+      "loss": 1.0592,
+      "step": 17071
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.750090894582689e-06,
+      "loss": 1.0316,
+      "step": 17072
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.749019556402933e-06,
+      "loss": 1.0026,
+      "step": 17073
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.747948335971928e-06,
+      "loss": 0.9366,
+      "step": 17074
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.7468772333098445e-06,
+      "loss": 0.9629,
+      "step": 17075
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.745806248436866e-06,
+      "loss": 1.0671,
+      "step": 17076
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.744735381373166e-06,
+      "loss": 1.0604,
+      "step": 17077
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.743664632138918e-06,
+      "loss": 1.0272,
+      "step": 17078
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.7425940007542804e-06,
+      "loss": 1.0108,
+      "step": 17079
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.741523487239432e-06,
+      "loss": 1.0429,
+      "step": 17080
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.740453091614535e-06,
+      "loss": 0.9577,
+      "step": 17081
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.739382813899755e-06,
+      "loss": 0.944,
+      "step": 17082
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.7383126541152415e-06,
+      "loss": 0.9795,
+      "step": 17083
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.737242612281166e-06,
+      "loss": 1.0852,
+      "step": 17084
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.736172688417681e-06,
+      "loss": 0.9753,
+      "step": 17085
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.7351028825449315e-06,
+      "loss": 0.9783,
+      "step": 17086
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.7340331946830825e-06,
+      "loss": 1.0657,
+      "step": 17087
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.7329636248522746e-06,
+      "loss": 1.0148,
+      "step": 17088
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.731894173072664e-06,
+      "loss": 1.0266,
+      "step": 17089
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.730824839364382e-06,
+      "loss": 1.017,
+      "step": 17090
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.729755623747584e-06,
+      "loss": 1.0472,
+      "step": 17091
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.728686526242405e-06,
+      "loss": 0.991,
+      "step": 17092
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.727617546868988e-06,
+      "loss": 1.0347,
+      "step": 17093
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.72654868564746e-06,
+      "loss": 1.0085,
+      "step": 17094
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.725479942597963e-06,
+      "loss": 0.9976,
+      "step": 17095
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.7244113177406304e-06,
+      "loss": 1.0104,
+      "step": 17096
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.723342811095584e-06,
+      "loss": 1.037,
+      "step": 17097
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.7222744226829525e-06,
+      "loss": 0.969,
+      "step": 17098
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.721206152522865e-06,
+      "loss": 1.0724,
+      "step": 17099
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.720138000635447e-06,
+      "loss": 0.9513,
+      "step": 17100
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.719069967040812e-06,
+      "loss": 1.0439,
+      "step": 17101
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.718002051759076e-06,
+      "loss": 1.0555,
+      "step": 17102
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.716934254810366e-06,
+      "loss": 1.0063,
+      "step": 17103
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.715866576214793e-06,
+      "loss": 0.9615,
+      "step": 17104
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.714799015992462e-06,
+      "loss": 1.0234,
+      "step": 17105
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.713731574163483e-06,
+      "loss": 0.9532,
+      "step": 17106
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.7126642507479725e-06,
+      "loss": 0.9746,
+      "step": 17107
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.711597045766028e-06,
+      "loss": 0.9567,
+      "step": 17108
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.710529959237753e-06,
+      "loss": 0.9358,
+      "step": 17109
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.709462991183246e-06,
+      "loss": 1.0512,
+      "step": 17110
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.708396141622615e-06,
+      "loss": 0.9773,
+      "step": 17111
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.707329410575946e-06,
+      "loss": 0.9682,
+      "step": 17112
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.706262798063336e-06,
+      "loss": 1.0608,
+      "step": 17113
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.705196304104873e-06,
+      "loss": 1.0432,
+      "step": 17114
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.704129928720658e-06,
+      "loss": 0.995,
+      "step": 17115
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.703063671930766e-06,
+      "loss": 0.9976,
+      "step": 17116
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.701997533755283e-06,
+      "loss": 0.9564,
+      "step": 17117
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.700931514214303e-06,
+      "loss": 0.98,
+      "step": 17118
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.699865613327894e-06,
+      "loss": 0.9491,
+      "step": 17119
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.698799831116139e-06,
+      "loss": 1.0456,
+      "step": 17120
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.697734167599109e-06,
+      "loss": 1.003,
+      "step": 17121
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.69666862279689e-06,
+      "loss": 0.9826,
+      "step": 17122
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.695603196729543e-06,
+      "loss": 0.9507,
+      "step": 17123
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.694537889417139e-06,
+      "loss": 0.9275,
+      "step": 17124
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.693472700879743e-06,
+      "loss": 0.9999,
+      "step": 17125
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.6924076311374296e-06,
+      "loss": 1.0712,
+      "step": 17126
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.691342680210253e-06,
+      "loss": 1.0656,
+      "step": 17127
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.690277848118273e-06,
+      "loss": 1.0211,
+      "step": 17128
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.6892131348815505e-06,
+      "loss": 1.0119,
+      "step": 17129
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.688148540520141e-06,
+      "loss": 1.0143,
+      "step": 17130
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.6870840650540986e-06,
+      "loss": 0.9647,
+      "step": 17131
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.6860197085034715e-06,
+      "loss": 1.0068,
+      "step": 17132
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.684955470888313e-06,
+      "loss": 1.0109,
+      "step": 17133
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.683891352228668e-06,
+      "loss": 0.8513,
+      "step": 17134
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.6828273525445805e-06,
+      "loss": 0.9669,
+      "step": 17135
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.681763471856094e-06,
+      "loss": 1.0312,
+      "step": 17136
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.6806997101832485e-06,
+      "loss": 1.0098,
+      "step": 17137
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.679636067546082e-06,
+      "loss": 0.9333,
+      "step": 17138
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.678572543964629e-06,
+      "loss": 1.0346,
+      "step": 17139
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.677509139458925e-06,
+      "loss": 0.9861,
+      "step": 17140
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.676445854048999e-06,
+      "loss": 1.0486,
+      "step": 17141
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.675382687754882e-06,
+      "loss": 1.0453,
+      "step": 17142
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.6743196405965985e-06,
+      "loss": 1.0299,
+      "step": 17143
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.673256712594175e-06,
+      "loss": 1.0108,
+      "step": 17144
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.672193903767632e-06,
+      "loss": 0.9864,
+      "step": 17145
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.671131214136991e-06,
+      "loss": 0.9784,
+      "step": 17146
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.670068643722269e-06,
+      "loss": 0.9619,
+      "step": 17147
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.669006192543482e-06,
+      "loss": 0.9891,
+      "step": 17148
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.667943860620641e-06,
+      "loss": 1.0018,
+      "step": 17149
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.666881647973762e-06,
+      "loss": 0.9641,
+      "step": 17150
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.665819554622847e-06,
+      "loss": 1.0637,
+      "step": 17151
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.664757580587908e-06,
+      "loss": 1.0159,
+      "step": 17152
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.663695725888946e-06,
+      "loss": 0.985,
+      "step": 17153
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.662633990545964e-06,
+      "loss": 0.9852,
+      "step": 17154
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.661572374578962e-06,
+      "loss": 0.9913,
+      "step": 17155
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.660510878007938e-06,
+      "loss": 0.9602,
+      "step": 17156
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.659449500852887e-06,
+      "loss": 1.0219,
+      "step": 17157
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.658388243133805e-06,
+      "loss": 1.0112,
+      "step": 17158
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.657327104870671e-06,
+      "loss": 1.0363,
+      "step": 17159
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.6562660860834865e-06,
+      "loss": 1.011,
+      "step": 17160
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.655205186792234e-06,
+      "loss": 1.1064,
+      "step": 17161
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.654144407016898e-06,
+      "loss": 0.9079,
+      "step": 17162
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.653083746777454e-06,
+      "loss": 0.9456,
+      "step": 17163
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.6520232060938885e-06,
+      "loss": 1.0016,
+      "step": 17164
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.650962784986177e-06,
+      "loss": 1.036,
+      "step": 17165
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.649902483474297e-06,
+      "loss": 1.0447,
+      "step": 17166
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.648842301578213e-06,
+      "loss": 1.0211,
+      "step": 17167
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.647782239317903e-06,
+      "loss": 1.1404,
+      "step": 17168
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.6467222967133365e-06,
+      "loss": 1.0364,
+      "step": 17169
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.6456624737844725e-06,
+      "loss": 0.9994,
+      "step": 17170
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.644602770551275e-06,
+      "loss": 0.9836,
+      "step": 17171
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.643543187033712e-06,
+      "loss": 1.0283,
+      "step": 17172
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.6424837232517425e-06,
+      "loss": 0.9657,
+      "step": 17173
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.6414243792253145e-06,
+      "loss": 1.0512,
+      "step": 17174
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.640365154974391e-06,
+      "loss": 1.0282,
+      "step": 17175
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.639306050518924e-06,
+      "loss": 1.0082,
+      "step": 17176
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.638247065878863e-06,
+      "loss": 1.0469,
+      "step": 17177
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.637188201074149e-06,
+      "loss": 0.9936,
+      "step": 17178
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.636129456124735e-06,
+      "loss": 1.0695,
+      "step": 17179
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.635070831050568e-06,
+      "loss": 1.0562,
+      "step": 17180
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.6340123258715794e-06,
+      "loss": 0.9661,
+      "step": 17181
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.632953940607709e-06,
+      "loss": 1.0222,
+      "step": 17182
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.631895675278902e-06,
+      "loss": 0.9552,
+      "step": 17183
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.630837529905089e-06,
+      "loss": 1.0198,
+      "step": 17184
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.629779504506199e-06,
+      "loss": 1.005,
+      "step": 17185
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.62872159910216e-06,
+      "loss": 0.9392,
+      "step": 17186
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.627663813712906e-06,
+      "loss": 0.9856,
+      "step": 17187
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.626606148358364e-06,
+      "loss": 0.9866,
+      "step": 17188
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.625548603058449e-06,
+      "loss": 1.0433,
+      "step": 17189
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.624491177833082e-06,
+      "loss": 0.956,
+      "step": 17190
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.623433872702192e-06,
+      "loss": 0.9079,
+      "step": 17191
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.6223766876856854e-06,
+      "loss": 0.993,
+      "step": 17192
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.6213196228034795e-06,
+      "loss": 0.9588,
+      "step": 17193
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.620262678075482e-06,
+      "loss": 0.9945,
+      "step": 17194
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.619205853521613e-06,
+      "loss": 0.9648,
+      "step": 17195
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.6181491491617705e-06,
+      "loss": 1.0314,
+      "step": 17196
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.617092565015861e-06,
+      "loss": 0.9684,
+      "step": 17197
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.616036101103784e-06,
+      "loss": 1.003,
+      "step": 17198
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.614979757445453e-06,
+      "loss": 0.947,
+      "step": 17199
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.6139235340607515e-06,
+      "loss": 1.0088,
+      "step": 17200
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.612867430969583e-06,
+      "loss": 0.9441,
+      "step": 17201
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.611811448191839e-06,
+      "loss": 0.9992,
+      "step": 17202
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.610755585747411e-06,
+      "loss": 1.0229,
+      "step": 17203
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.609699843656188e-06,
+      "loss": 1.029,
+      "step": 17204
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.6086442219380544e-06,
+      "loss": 0.9868,
+      "step": 17205
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.607588720612905e-06,
+      "loss": 0.9587,
+      "step": 17206
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.60653333970061e-06,
+      "loss": 1.0651,
+      "step": 17207
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.605478079221054e-06,
+      "loss": 1.0308,
+      "step": 17208
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.6044229391941115e-06,
+      "loss": 0.9597,
+      "step": 17209
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.60336791963967e-06,
+      "loss": 0.9619,
+      "step": 17210
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.60231302057759e-06,
+      "loss": 0.9603,
+      "step": 17211
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.601258242027746e-06,
+      "loss": 0.9046,
+      "step": 17212
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.600203584010007e-06,
+      "loss": 0.9739,
+      "step": 17213
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.599149046544241e-06,
+      "loss": 1.0659,
+      "step": 17214
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.59809462965031e-06,
+      "loss": 1.013,
+      "step": 17215
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5970403333480785e-06,
+      "loss": 0.9791,
+      "step": 17216
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5959861576574025e-06,
+      "loss": 0.9948,
+      "step": 17217
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.594932102598142e-06,
+      "loss": 0.9758,
+      "step": 17218
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5938781681901525e-06,
+      "loss": 0.9668,
+      "step": 17219
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.592824354453284e-06,
+      "loss": 0.9522,
+      "step": 17220
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.591770661407389e-06,
+      "loss": 0.9253,
+      "step": 17221
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5907170890723164e-06,
+      "loss": 1.0411,
+      "step": 17222
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5896636374679105e-06,
+      "loss": 1.0415,
+      "step": 17223
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.588610306614016e-06,
+      "loss": 1.0293,
+      "step": 17224
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.587557096530473e-06,
+      "loss": 1.0459,
+      "step": 17225
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.586504007237123e-06,
+      "loss": 0.9868,
+      "step": 17226
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5854510387538e-06,
+      "loss": 1.0714,
+      "step": 17227
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.584398191100341e-06,
+      "loss": 0.9506,
+      "step": 17228
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.583345464296578e-06,
+      "loss": 0.9974,
+      "step": 17229
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.58229285836234e-06,
+      "loss": 1.0537,
+      "step": 17230
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.581240373317456e-06,
+      "loss": 0.9904,
+      "step": 17231
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.580188009181751e-06,
+      "loss": 0.9841,
+      "step": 17232
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5791357659750464e-06,
+      "loss": 0.942,
+      "step": 17233
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.578083643717168e-06,
+      "loss": 0.9601,
+      "step": 17234
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.577031642427928e-06,
+      "loss": 1.035,
+      "step": 17235
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.575979762127147e-06,
+      "loss": 0.918,
+      "step": 17236
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.574928002834639e-06,
+      "loss": 1.0277,
+      "step": 17237
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5738763645702145e-06,
+      "loss": 0.9856,
+      "step": 17238
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.572824847353683e-06,
+      "loss": 0.981,
+      "step": 17239
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.571773451204853e-06,
+      "loss": 0.9987,
+      "step": 17240
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.570722176143528e-06,
+      "loss": 1.0481,
+      "step": 17241
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.569671022189516e-06,
+      "loss": 1.04,
+      "step": 17242
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.568619989362605e-06,
+      "loss": 1.0494,
+      "step": 17243
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5675690776826054e-06,
+      "loss": 0.9827,
+      "step": 17244
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.566518287169308e-06,
+      "loss": 0.9437,
+      "step": 17245
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.565467617842511e-06,
+      "loss": 1.103,
+      "step": 17246
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.564417069721994e-06,
+      "loss": 0.9766,
+      "step": 17247
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.563366642827558e-06,
+      "loss": 0.9554,
+      "step": 17248
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.562316337178986e-06,
+      "loss": 1.092,
+      "step": 17249
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.561266152796064e-06,
+      "loss": 0.9557,
+      "step": 17250
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.560216089698567e-06,
+      "loss": 0.9813,
+      "step": 17251
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.559166147906282e-06,
+      "loss": 1.0093,
+      "step": 17252
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.558116327438988e-06,
+      "loss": 0.9788,
+      "step": 17253
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.557066628316455e-06,
+      "loss": 0.9427,
+      "step": 17254
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5560170505584535e-06,
+      "loss": 0.9191,
+      "step": 17255
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.554967594184762e-06,
+      "loss": 1.0307,
+      "step": 17256
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.553918259215149e-06,
+      "loss": 0.94,
+      "step": 17257
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.552869045669375e-06,
+      "loss": 1.0345,
+      "step": 17258
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.551819953567202e-06,
+      "loss": 0.9828,
+      "step": 17259
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5507709829283975e-06,
+      "loss": 1.0017,
+      "step": 17260
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.549722133772724e-06,
+      "loss": 0.9271,
+      "step": 17261
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5486734061199265e-06,
+      "loss": 0.9738,
+      "step": 17262
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.547624799989771e-06,
+      "loss": 1.0253,
+      "step": 17263
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.546576315402007e-06,
+      "loss": 0.9398,
+      "step": 17264
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.545527952376382e-06,
+      "loss": 1.0367,
+      "step": 17265
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.54447971093264e-06,
+      "loss": 0.9893,
+      "step": 17266
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.543431591090537e-06,
+      "loss": 0.8952,
+      "step": 17267
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5423835928698125e-06,
+      "loss": 1.0036,
+      "step": 17268
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.541335716290204e-06,
+      "loss": 0.9986,
+      "step": 17269
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.540287961371448e-06,
+      "loss": 0.9888,
+      "step": 17270
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.539240328133288e-06,
+      "loss": 1.012,
+      "step": 17271
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.538192816595461e-06,
+      "loss": 1.0449,
+      "step": 17272
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.537145426777687e-06,
+      "loss": 1.0144,
+      "step": 17273
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.536098158699699e-06,
+      "loss": 0.966,
+      "step": 17274
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.535051012381233e-06,
+      "loss": 1.0179,
+      "step": 17275
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.534003987842005e-06,
+      "loss": 1.0074,
+      "step": 17276
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.532957085101741e-06,
+      "loss": 1.0362,
+      "step": 17277
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.531910304180156e-06,
+      "loss": 1.032,
+      "step": 17278
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.53086364509698e-06,
+      "loss": 0.9315,
+      "step": 17279
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.529817107871918e-06,
+      "loss": 1.0219,
+      "step": 17280
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.528770692524686e-06,
+      "loss": 0.9869,
+      "step": 17281
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.527724399074994e-06,
+      "loss": 0.966,
+      "step": 17282
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.526678227542559e-06,
+      "loss": 1.0345,
+      "step": 17283
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.525632177947078e-06,
+      "loss": 1.089,
+      "step": 17284
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.524586250308259e-06,
+      "loss": 1.0805,
+      "step": 17285
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.523540444645804e-06,
+      "loss": 1.02,
+      "step": 17286
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.522494760979413e-06,
+      "loss": 0.9845,
+      "step": 17287
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.521449199328781e-06,
+      "loss": 1.0165,
+      "step": 17288
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.520403759713607e-06,
+      "loss": 0.9372,
+      "step": 17289
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.51935844215358e-06,
+      "loss": 1.0395,
+      "step": 17290
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.518313246668393e-06,
+      "loss": 0.974,
+      "step": 17291
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.517268173277733e-06,
+      "loss": 1.0092,
+      "step": 17292
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.516223222001284e-06,
+      "loss": 0.9836,
+      "step": 17293
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.515178392858737e-06,
+      "loss": 0.9587,
+      "step": 17294
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.514133685869764e-06,
+      "loss": 0.9756,
+      "step": 17295
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.513089101054051e-06,
+      "loss": 0.9845,
+      "step": 17296
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.51204463843127e-06,
+      "loss": 0.999,
+      "step": 17297
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.511000298021097e-06,
+      "loss": 1.0016,
+      "step": 17298
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.509956079843205e-06,
+      "loss": 1.019,
+      "step": 17299
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.508911983917262e-06,
+      "loss": 1.0242,
+      "step": 17300
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5078680102629385e-06,
+      "loss": 1.029,
+      "step": 17301
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.506824158899896e-06,
+      "loss": 1.0578,
+      "step": 17302
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5057804298477994e-06,
+      "loss": 1.0068,
+      "step": 17303
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.504736823126309e-06,
+      "loss": 0.9755,
+      "step": 17304
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.503693338755081e-06,
+      "loss": 1.024,
+      "step": 17305
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.502649976753777e-06,
+      "loss": 0.9951,
+      "step": 17306
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.501606737142045e-06,
+      "loss": 0.9667,
+      "step": 17307
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.500563619939538e-06,
+      "loss": 1.0615,
+      "step": 17308
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.499520625165905e-06,
+      "loss": 0.9864,
+      "step": 17309
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.498477752840794e-06,
+      "loss": 1.0049,
+      "step": 17310
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.497435002983848e-06,
+      "loss": 1.0016,
+      "step": 17311
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.49639237561471e-06,
+      "loss": 1.0045,
+      "step": 17312
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.495349870753019e-06,
+      "loss": 1.0089,
+      "step": 17313
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.494307488418412e-06,
+      "loss": 1.0364,
+      "step": 17314
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.49326522863053e-06,
+      "loss": 0.9694,
+      "step": 17315
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.492223091408994e-06,
+      "loss": 1.0004,
+      "step": 17316
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.491181076773444e-06,
+      "loss": 0.9438,
+      "step": 17317
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.490139184743507e-06,
+      "loss": 1.0334,
+      "step": 17318
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.489097415338807e-06,
+      "loss": 0.964,
+      "step": 17319
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.4880557685789695e-06,
+      "loss": 0.9707,
+      "step": 17320
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.487014244483614e-06,
+      "loss": 1.0394,
+      "step": 17321
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.48597284307236e-06,
+      "loss": 0.9927,
+      "step": 17322
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.484931564364825e-06,
+      "loss": 0.9427,
+      "step": 17323
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.483890408380623e-06,
+      "loss": 1.0387,
+      "step": 17324
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.482849375139365e-06,
+      "loss": 0.9912,
+      "step": 17325
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.481808464660666e-06,
+      "loss": 0.9553,
+      "step": 17326
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.480767676964123e-06,
+      "loss": 1.0536,
+      "step": 17327
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.479727012069349e-06,
+      "loss": 1.005,
+      "step": 17328
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.478686469995948e-06,
+      "loss": 0.9333,
+      "step": 17329
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.477646050763519e-06,
+      "loss": 0.9661,
+      "step": 17330
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.4766057543916525e-06,
+      "loss": 1.0281,
+      "step": 17331
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.475565580899955e-06,
+      "loss": 1.054,
+      "step": 17332
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.474525530308016e-06,
+      "loss": 1.0056,
+      "step": 17333
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.47348560263543e-06,
+      "loss": 1.0355,
+      "step": 17334
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.472445797901775e-06,
+      "loss": 1.0445,
+      "step": 17335
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.47140611612665e-06,
+      "loss": 0.932,
+      "step": 17336
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.470366557329637e-06,
+      "loss": 0.975,
+      "step": 17337
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.469327121530314e-06,
+      "loss": 1.0012,
+      "step": 17338
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.4682878087482575e-06,
+      "loss": 1.0572,
+      "step": 17339
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.467248619003054e-06,
+      "loss": 1.0009,
+      "step": 17340
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.466209552314277e-06,
+      "loss": 0.9303,
+      "step": 17341
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.465170608701494e-06,
+      "loss": 1.0065,
+      "step": 17342
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.464131788184273e-06,
+      "loss": 0.9301,
+      "step": 17343
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.463093090782192e-06,
+      "loss": 1.0001,
+      "step": 17344
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.462054516514815e-06,
+      "loss": 0.9582,
+      "step": 17345
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.461016065401699e-06,
+      "loss": 0.9562,
+      "step": 17346
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.459977737462406e-06,
+      "loss": 1.0664,
+      "step": 17347
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.458939532716502e-06,
+      "loss": 0.9987,
+      "step": 17348
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.457901451183536e-06,
+      "loss": 0.9768,
+      "step": 17349
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.456863492883062e-06,
+      "loss": 0.9937,
+      "step": 17350
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.455825657834638e-06,
+      "loss": 1.0412,
+      "step": 17351
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.454787946057814e-06,
+      "loss": 0.9611,
+      "step": 17352
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.45375035757213e-06,
+      "loss": 1.0409,
+      "step": 17353
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.452712892397132e-06,
+      "loss": 1.0375,
+      "step": 17354
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.4516755505523666e-06,
+      "loss": 1.0063,
+      "step": 17355
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.450638332057378e-06,
+      "loss": 0.9932,
+      "step": 17356
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.449601236931693e-06,
+      "loss": 1.022,
+      "step": 17357
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.4485642651948516e-06,
+      "loss": 0.9987,
+      "step": 17358
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.447527416866393e-06,
+      "loss": 0.9682,
+      "step": 17359
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.446490691965842e-06,
+      "loss": 0.9813,
+      "step": 17360
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.445454090512728e-06,
+      "loss": 1.0117,
+      "step": 17361
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.444417612526573e-06,
+      "loss": 0.9602,
+      "step": 17362
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.443381258026914e-06,
+      "loss": 0.9881,
+      "step": 17363
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.4423450270332625e-06,
+      "loss": 0.9732,
+      "step": 17364
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.4413089195651384e-06,
+      "loss": 1.0628,
+      "step": 17365
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.440272935642058e-06,
+      "loss": 0.9851,
+      "step": 17366
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.439237075283543e-06,
+      "loss": 0.979,
+      "step": 17367
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.438201338509098e-06,
+      "loss": 1.0419,
+      "step": 17368
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.437165725338237e-06,
+      "loss": 0.9886,
+      "step": 17369
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.4361302357904656e-06,
+      "loss": 0.984,
+      "step": 17370
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.4350948698852915e-06,
+      "loss": 0.9965,
+      "step": 17371
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.434059627642216e-06,
+      "loss": 0.9664,
+      "step": 17372
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.433024509080739e-06,
+      "loss": 1.1019,
+      "step": 17373
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.431989514220361e-06,
+      "loss": 0.9507,
+      "step": 17374
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.430954643080577e-06,
+      "loss": 1.0242,
+      "step": 17375
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.429919895680881e-06,
+      "loss": 0.9638,
+      "step": 17376
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.428885272040764e-06,
+      "loss": 0.9583,
+      "step": 17377
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.427850772179715e-06,
+      "loss": 1.0344,
+      "step": 17378
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.426816396117222e-06,
+      "loss": 0.9891,
+      "step": 17379
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.4257821438727666e-06,
+      "loss": 0.9394,
+      "step": 17380
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.424748015465834e-06,
+      "loss": 1.0263,
+      "step": 17381
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.423714010915902e-06,
+      "loss": 1.0407,
+      "step": 17382
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.422680130242448e-06,
+      "loss": 0.9677,
+      "step": 17383
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.4216463734649485e-06,
+      "loss": 1.03,
+      "step": 17384
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.420612740602874e-06,
+      "loss": 1.0141,
+      "step": 17385
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.419579231675698e-06,
+      "loss": 0.9409,
+      "step": 17386
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.418545846702885e-06,
+      "loss": 1.085,
+      "step": 17387
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.417512585703903e-06,
+      "loss": 1.032,
+      "step": 17388
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.416479448698215e-06,
+      "loss": 1.0404,
+      "step": 17389
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.415446435705282e-06,
+      "loss": 1.0916,
+      "step": 17390
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.4144135467445614e-06,
+      "loss": 1.0722,
+      "step": 17391
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.413380781835512e-06,
+      "loss": 1.0084,
+      "step": 17392
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.412348140997584e-06,
+      "loss": 1.0136,
+      "step": 17393
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.411315624250234e-06,
+      "loss": 0.9584,
+      "step": 17394
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.410283231612909e-06,
+      "loss": 0.9827,
+      "step": 17395
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.409250963105055e-06,
+      "loss": 1.0271,
+      "step": 17396
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.408218818746119e-06,
+      "loss": 0.9959,
+      "step": 17397
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.40718679855554e-06,
+      "loss": 1.0168,
+      "step": 17398
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.406154902552764e-06,
+      "loss": 1.0746,
+      "step": 17399
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.405123130757219e-06,
+      "loss": 1.0202,
+      "step": 17400
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.404091483188349e-06,
+      "loss": 1.0398,
+      "step": 17401
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.403059959865583e-06,
+      "loss": 1.0483,
+      "step": 17402
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.402028560808357e-06,
+      "loss": 1.0647,
+      "step": 17403
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.400997286036087e-06,
+      "loss": 0.9631,
+      "step": 17404
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.399966135568212e-06,
+      "loss": 0.9991,
+      "step": 17405
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3989351094241495e-06,
+      "loss": 0.9994,
+      "step": 17406
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.397904207623322e-06,
+      "loss": 1.0331,
+      "step": 17407
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.396873430185147e-06,
+      "loss": 1.0009,
+      "step": 17408
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.395842777129044e-06,
+      "loss": 1.0442,
+      "step": 17409
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.394812248474429e-06,
+      "loss": 1.0308,
+      "step": 17410
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.393781844240704e-06,
+      "loss": 0.9916,
+      "step": 17411
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.392751564447287e-06,
+      "loss": 1.0012,
+      "step": 17412
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.391721409113584e-06,
+      "loss": 1.0304,
+      "step": 17413
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.390691378259002e-06,
+      "loss": 0.927,
+      "step": 17414
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.389661471902935e-06,
+      "loss": 1.0228,
+      "step": 17415
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.388631690064792e-06,
+      "loss": 0.9991,
+      "step": 17416
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3876020327639674e-06,
+      "loss": 0.9503,
+      "step": 17417
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3865725000198595e-06,
+      "loss": 1.0046,
+      "step": 17418
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3855430918518535e-06,
+      "loss": 1.0106,
+      "step": 17419
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.384513808279348e-06,
+      "loss": 0.9811,
+      "step": 17420
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3834846493217335e-06,
+      "loss": 0.9872,
+      "step": 17421
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.382455614998388e-06,
+      "loss": 1.0236,
+      "step": 17422
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.381426705328696e-06,
+      "loss": 0.9576,
+      "step": 17423
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.380397920332045e-06,
+      "loss": 0.9843,
+      "step": 17424
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.379369260027814e-06,
+      "loss": 1.0592,
+      "step": 17425
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.378340724435372e-06,
+      "loss": 1.0326,
+      "step": 17426
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.377312313574097e-06,
+      "loss": 1.0098,
+      "step": 17427
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3762840274633646e-06,
+      "loss": 0.9643,
+      "step": 17428
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.375255866122544e-06,
+      "loss": 1.0382,
+      "step": 17429
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3742278295709995e-06,
+      "loss": 0.9845,
+      "step": 17430
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.373199917828092e-06,
+      "loss": 1.0037,
+      "step": 17431
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.372172130913196e-06,
+      "loss": 1.0431,
+      "step": 17432
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.371144468845662e-06,
+      "loss": 0.9279,
+      "step": 17433
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.370116931644851e-06,
+      "loss": 1.0347,
+      "step": 17434
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.369089519330113e-06,
+      "loss": 0.9856,
+      "step": 17435
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.368062231920816e-06,
+      "loss": 0.9805,
+      "step": 17436
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.367035069436296e-06,
+      "loss": 0.9563,
+      "step": 17437
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.366008031895904e-06,
+      "loss": 1.0013,
+      "step": 17438
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.364981119318993e-06,
+      "loss": 1.0324,
+      "step": 17439
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3639543317249055e-06,
+      "loss": 1.0147,
+      "step": 17440
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3629276691329785e-06,
+      "loss": 0.9804,
+      "step": 17441
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.361901131562547e-06,
+      "loss": 1.0259,
+      "step": 17442
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.360874719032962e-06,
+      "loss": 0.9811,
+      "step": 17443
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.359848431563546e-06,
+      "loss": 0.9861,
+      "step": 17444
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.358822269173634e-06,
+      "loss": 1.0352,
+      "step": 17445
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.357796231882553e-06,
+      "loss": 0.9019,
+      "step": 17446
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3567703197096405e-06,
+      "loss": 0.9142,
+      "step": 17447
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3557445326742115e-06,
+      "loss": 0.9877,
+      "step": 17448
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.35471887079559e-06,
+      "loss": 0.9961,
+      "step": 17449
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3536933340930955e-06,
+      "loss": 1.0046,
+      "step": 17450
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3526679225860535e-06,
+      "loss": 0.9959,
+      "step": 17451
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.351642636293772e-06,
+      "loss": 1.0328,
+      "step": 17452
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.350617475235566e-06,
+      "loss": 1.0263,
+      "step": 17453
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3495924394307465e-06,
+      "loss": 0.9556,
+      "step": 17454
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.348567528898622e-06,
+      "loss": 0.9889,
+      "step": 17455
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3475427436584984e-06,
+      "loss": 1.0297,
+      "step": 17456
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.346518083729679e-06,
+      "loss": 0.963,
+      "step": 17457
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.345493549131467e-06,
+      "loss": 1.0095,
+      "step": 17458
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.344469139883161e-06,
+      "loss": 1.0276,
+      "step": 17459
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.343444856004054e-06,
+      "loss": 1.0407,
+      "step": 17460
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.342420697513446e-06,
+      "loss": 0.9684,
+      "step": 17461
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.341396664430624e-06,
+      "loss": 0.9554,
+      "step": 17462
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3403727567748806e-06,
+      "loss": 1.0005,
+      "step": 17463
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.339348974565502e-06,
+      "loss": 0.9314,
+      "step": 17464
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.338325317821772e-06,
+      "loss": 0.9704,
+      "step": 17465
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.337301786562974e-06,
+      "loss": 1.0027,
+      "step": 17466
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.33627838080839e-06,
+      "loss": 0.9549,
+      "step": 17467
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.335255100577293e-06,
+      "loss": 0.9418,
+      "step": 17468
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3342319458889635e-06,
+      "loss": 1.0864,
+      "step": 17469
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.333208916762672e-06,
+      "loss": 0.9067,
+      "step": 17470
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.332186013217689e-06,
+      "loss": 0.9777,
+      "step": 17471
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.331163235273282e-06,
+      "loss": 1.0325,
+      "step": 17472
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.330140582948719e-06,
+      "loss": 1.0041,
+      "step": 17473
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.329118056263264e-06,
+      "loss": 0.9651,
+      "step": 17474
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3280956552361755e-06,
+      "loss": 0.9739,
+      "step": 17475
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.327073379886714e-06,
+      "loss": 0.997,
+      "step": 17476
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.326051230234135e-06,
+      "loss": 0.98,
+      "step": 17477
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.325029206297694e-06,
+      "loss": 1.0051,
+      "step": 17478
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3240073080966405e-06,
+      "loss": 1.0455,
+      "step": 17479
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.322985535650226e-06,
+      "loss": 1.021,
+      "step": 17480
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.321963888977697e-06,
+      "loss": 1.075,
+      "step": 17481
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.320942368098296e-06,
+      "loss": 1.0207,
+      "step": 17482
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.319920973031272e-06,
+      "loss": 0.9512,
+      "step": 17483
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.318899703795854e-06,
+      "loss": 1.0735,
+      "step": 17484
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.317878560411286e-06,
+      "loss": 1.0012,
+      "step": 17485
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.316857542896804e-06,
+      "loss": 1.0433,
+      "step": 17486
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.315836651271642e-06,
+      "loss": 0.9444,
+      "step": 17487
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.314815885555021e-06,
+      "loss": 0.9975,
+      "step": 17488
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.313795245766179e-06,
+      "loss": 1.0031,
+      "step": 17489
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.312774731924338e-06,
+      "loss": 0.994,
+      "step": 17490
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3117543440487265e-06,
+      "loss": 1.026,
+      "step": 17491
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3107340821585514e-06,
+      "loss": 1.0265,
+      "step": 17492
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.309713946273044e-06,
+      "loss": 0.934,
+      "step": 17493
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.308693936411421e-06,
+      "loss": 0.999,
+      "step": 17494
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3076740525928844e-06,
+      "loss": 0.939,
+      "step": 17495
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.306654294836657e-06,
+      "loss": 0.9809,
+      "step": 17496
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.305634663161942e-06,
+      "loss": 0.9637,
+      "step": 17497
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3046151575879525e-06,
+      "loss": 0.9663,
+      "step": 17498
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.303595778133881e-06,
+      "loss": 0.9945,
+      "step": 17499
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.30257652481894e-06,
+      "loss": 0.964,
+      "step": 17500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.301557397662325e-06,
+      "loss": 0.9937,
+      "step": 17501
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.300538396683238e-06,
+      "loss": 0.9803,
+      "step": 17502
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.2995195219008625e-06,
+      "loss": 1.0359,
+      "step": 17503
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.2985007733344e-06,
+      "loss": 1.0363,
+      "step": 17504
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.297482151003042e-06,
+      "loss": 1.0145,
+      "step": 17505
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.296463654925969e-06,
+      "loss": 1.0696,
+      "step": 17506
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.295445285122368e-06,
+      "loss": 1.0228,
+      "step": 17507
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.2944270416114255e-06,
+      "loss": 1.0242,
+      "step": 17508
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.2934089244123235e-06,
+      "loss": 1.0192,
+      "step": 17509
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.2923909335442335e-06,
+      "loss": 1.0654,
+      "step": 17510
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.291373069026331e-06,
+      "loss": 0.9693,
+      "step": 17511
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.290355330877797e-06,
+      "loss": 1.0062,
+      "step": 17512
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.2893377191178e-06,
+      "loss": 1.0689,
+      "step": 17513
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.288320233765504e-06,
+      "loss": 0.9858,
+      "step": 17514
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.287302874840076e-06,
+      "loss": 1.0471,
+      "step": 17515
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.286285642360687e-06,
+      "loss": 1.0412,
+      "step": 17516
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.285268536346489e-06,
+      "loss": 0.9933,
+      "step": 17517
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.284251556816648e-06,
+      "loss": 0.9271,
+      "step": 17518
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.283234703790313e-06,
+      "loss": 0.9984,
+      "step": 17519
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.2822179772866515e-06,
+      "loss": 1.0073,
+      "step": 17520
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.281201377324803e-06,
+      "loss": 1.0569,
+      "step": 17521
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.2801849039239205e-06,
+      "loss": 1.0147,
+      "step": 17522
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.27916855710315e-06,
+      "loss": 1.0015,
+      "step": 17523
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.278152336881645e-06,
+      "loss": 0.9797,
+      "step": 17524
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.277136243278537e-06,
+      "loss": 1.0072,
+      "step": 17525
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.276120276312967e-06,
+      "loss": 0.9889,
+      "step": 17526
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.275104436004083e-06,
+      "loss": 0.9808,
+      "step": 17527
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.274088722371009e-06,
+      "loss": 1.0007,
+      "step": 17528
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.273073135432882e-06,
+      "loss": 1.0099,
+      "step": 17529
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.27205767520883e-06,
+      "loss": 1.0304,
+      "step": 17530
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.271042341717988e-06,
+      "loss": 1.0421,
+      "step": 17531
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.270027134979474e-06,
+      "loss": 1.0034,
+      "step": 17532
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.269012055012416e-06,
+      "loss": 1.0571,
+      "step": 17533
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.267997101835928e-06,
+      "loss": 1.0274,
+      "step": 17534
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.266982275469141e-06,
+      "loss": 1.0145,
+      "step": 17535
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.265967575931162e-06,
+      "loss": 0.9906,
+      "step": 17536
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.264953003241105e-06,
+      "loss": 0.9766,
+      "step": 17537
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.2639385574180825e-06,
+      "loss": 0.9628,
+      "step": 17538
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.262924238481204e-06,
+      "loss": 0.9506,
+      "step": 17539
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.261910046449578e-06,
+      "loss": 1.0153,
+      "step": 17540
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.260895981342305e-06,
+      "loss": 0.9676,
+      "step": 17541
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.259882043178488e-06,
+      "loss": 0.993,
+      "step": 17542
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.258868231977228e-06,
+      "loss": 0.9594,
+      "step": 17543
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.25785454775762e-06,
+      "loss": 1.0218,
+      "step": 17544
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.25684099053876e-06,
+      "loss": 0.9983,
+      "step": 17545
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.255827560339739e-06,
+      "loss": 1.13,
+      "step": 17546
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.254814257179648e-06,
+      "loss": 1.0123,
+      "step": 17547
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.253801081077574e-06,
+      "loss": 1.0252,
+      "step": 17548
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.252788032052601e-06,
+      "loss": 0.9662,
+      "step": 17549
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.251775110123813e-06,
+      "loss": 1.0152,
+      "step": 17550
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.2507623153102914e-06,
+      "loss": 1.0673,
+      "step": 17551
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.249749647631112e-06,
+      "loss": 0.9821,
+      "step": 17552
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.24873710710535e-06,
+      "loss": 0.9248,
+      "step": 17553
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.247724693752079e-06,
+      "loss": 0.9454,
+      "step": 17554
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.24671240759037e-06,
+      "loss": 1.0266,
+      "step": 17555
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.245700248639292e-06,
+      "loss": 1.0242,
+      "step": 17556
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.244688216917909e-06,
+      "loss": 1.1134,
+      "step": 17557
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.243676312445287e-06,
+      "loss": 1.0055,
+      "step": 17558
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.242664535240485e-06,
+      "loss": 0.9882,
+      "step": 17559
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.241652885322561e-06,
+      "loss": 1.0429,
+      "step": 17560
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.240641362710574e-06,
+      "loss": 1.0035,
+      "step": 17561
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.239629967423577e-06,
+      "loss": 1.0572,
+      "step": 17562
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.238618699480621e-06,
+      "loss": 1.0234,
+      "step": 17563
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.2376075589007545e-06,
+      "loss": 0.9993,
+      "step": 17564
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.236596545703025e-06,
+      "loss": 1.0373,
+      "step": 17565
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.235585659906476e-06,
+      "loss": 1.0252,
+      "step": 17566
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.234574901530154e-06,
+      "loss": 1.0969,
+      "step": 17567
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.233564270593088e-06,
+      "loss": 0.9413,
+      "step": 17568
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.232553767114324e-06,
+      "loss": 1.0135,
+      "step": 17569
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.231543391112896e-06,
+      "loss": 0.9782,
+      "step": 17570
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.230533142607836e-06,
+      "loss": 1.0705,
+      "step": 17571
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.229523021618167e-06,
+      "loss": 1.0137,
+      "step": 17572
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.228513028162926e-06,
+      "loss": 0.9387,
+      "step": 17573
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.227503162261133e-06,
+      "loss": 0.9441,
+      "step": 17574
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.226493423931815e-06,
+      "loss": 1.0262,
+      "step": 17575
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.225483813193984e-06,
+      "loss": 0.9262,
+      "step": 17576
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.224474330066665e-06,
+      "loss": 1.0258,
+      "step": 17577
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.2234649745688735e-06,
+      "loss": 1.0404,
+      "step": 17578
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.2224557467196195e-06,
+      "loss": 1.0275,
+      "step": 17579
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.2214466465379105e-06,
+      "loss": 0.9968,
+      "step": 17580
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.220437674042761e-06,
+      "loss": 1.0246,
+      "step": 17581
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.21942882925318e-06,
+      "loss": 1.1083,
+      "step": 17582
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.218420112188157e-06,
+      "loss": 0.975,
+      "step": 17583
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.2174115228667064e-06,
+      "loss": 0.9396,
+      "step": 17584
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.216403061307823e-06,
+      "loss": 0.9706,
+      "step": 17585
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.2153947275305045e-06,
+      "loss": 0.951,
+      "step": 17586
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.214386521553736e-06,
+      "loss": 1.0156,
+      "step": 17587
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.213378443396519e-06,
+      "loss": 0.9837,
+      "step": 17588
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.212370493077842e-06,
+      "loss": 1.0278,
+      "step": 17589
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.211362670616686e-06,
+      "loss": 0.9565,
+      "step": 17590
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.2103549760320325e-06,
+      "loss": 0.9468,
+      "step": 17591
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.209347409342873e-06,
+      "loss": 0.9898,
+      "step": 17592
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.208339970568186e-06,
+      "loss": 1.0216,
+      "step": 17593
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.2073326597269416e-06,
+      "loss": 1.0524,
+      "step": 17594
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.206325476838114e-06,
+      "loss": 1.0318,
+      "step": 17595
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.205318421920682e-06,
+      "loss": 1.0163,
+      "step": 17596
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.204311494993616e-06,
+      "loss": 1.0402,
+      "step": 17597
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.203304696075876e-06,
+      "loss": 0.9949,
+      "step": 17598
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.202298025186428e-06,
+      "loss": 1.0444,
+      "step": 17599
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.2012914823442425e-06,
+      "loss": 1.0314,
+      "step": 17600
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.200285067568271e-06,
+      "loss": 1.0485,
+      "step": 17601
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.1992787808774745e-06,
+      "loss": 1.0497,
+      "step": 17602
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.198272622290804e-06,
+      "loss": 1.0031,
+      "step": 17603
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.197266591827225e-06,
+      "loss": 1.0431,
+      "step": 17604
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.196260689505675e-06,
+      "loss": 0.9753,
+      "step": 17605
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.195254915345106e-06,
+      "loss": 0.9709,
+      "step": 17606
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.1942492693644606e-06,
+      "loss": 0.9236,
+      "step": 17607
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.193243751582693e-06,
+      "loss": 1.0494,
+      "step": 17608
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.192238362018732e-06,
+      "loss": 1.0063,
+      "step": 17609
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.191233100691521e-06,
+      "loss": 1.0361,
+      "step": 17610
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.190227967619995e-06,
+      "loss": 0.9788,
+      "step": 17611
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.1892229628230885e-06,
+      "loss": 1.0016,
+      "step": 17612
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.188218086319731e-06,
+      "loss": 0.974,
+      "step": 17613
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.18721333812885e-06,
+      "loss": 1.0527,
+      "step": 17614
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.1862087182693786e-06,
+      "loss": 0.9497,
+      "step": 17615
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.185204226760234e-06,
+      "loss": 0.9618,
+      "step": 17616
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.18419986362034e-06,
+      "loss": 0.9629,
+      "step": 17617
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.1831956288686115e-06,
+      "loss": 1.0348,
+      "step": 17618
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.182191522523976e-06,
+      "loss": 0.9803,
+      "step": 17619
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.181187544605337e-06,
+      "loss": 1.0137,
+      "step": 17620
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.18018369513161e-06,
+      "loss": 1.0214,
+      "step": 17621
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.179179974121705e-06,
+      "loss": 0.9984,
+      "step": 17622
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.178176381594526e-06,
+      "loss": 0.9742,
+      "step": 17623
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.177172917568981e-06,
+      "loss": 0.9871,
+      "step": 17624
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.17616958206397e-06,
+      "loss": 0.9441,
+      "step": 17625
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.175166375098395e-06,
+      "loss": 1.0026,
+      "step": 17626
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.174163296691151e-06,
+      "loss": 0.9431,
+      "step": 17627
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.173160346861134e-06,
+      "loss": 1.0016,
+      "step": 17628
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.172157525627236e-06,
+      "loss": 0.9936,
+      "step": 17629
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.171154833008345e-06,
+      "loss": 1.0004,
+      "step": 17630
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.1701522690233515e-06,
+      "loss": 1.0043,
+      "step": 17631
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.169149833691141e-06,
+      "loss": 0.9198,
+      "step": 17632
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.168147527030592e-06,
+      "loss": 0.9793,
+      "step": 17633
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.16714534906059e-06,
+      "loss": 1.0426,
+      "step": 17634
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.166143299800011e-06,
+      "loss": 0.9849,
+      "step": 17635
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.165141379267729e-06,
+      "loss": 0.9402,
+      "step": 17636
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.164139587482618e-06,
+      "loss": 1.0344,
+      "step": 17637
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.163137924463549e-06,
+      "loss": 1.0137,
+      "step": 17638
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.162136390229389e-06,
+      "loss": 1.031,
+      "step": 17639
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.161134984799008e-06,
+      "loss": 1.0088,
+      "step": 17640
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.16013370819126e-06,
+      "loss": 0.9281,
+      "step": 17641
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.159132560425015e-06,
+      "loss": 1.0295,
+      "step": 17642
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.158131541519128e-06,
+      "loss": 1.039,
+      "step": 17643
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.1571306514924536e-06,
+      "loss": 0.9174,
+      "step": 17644
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.156129890363849e-06,
+      "loss": 0.9317,
+      "step": 17645
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.155129258152163e-06,
+      "loss": 1.0205,
+      "step": 17646
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.154128754876244e-06,
+      "loss": 0.9674,
+      "step": 17647
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.15312838055494e-06,
+      "loss": 1.0115,
+      "step": 17648
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.152128135207095e-06,
+      "loss": 1.0115,
+      "step": 17649
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.151128018851547e-06,
+      "loss": 0.9767,
+      "step": 17650
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.150128031507142e-06,
+      "loss": 1.0054,
+      "step": 17651
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.149128173192706e-06,
+      "loss": 1.0453,
+      "step": 17652
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.148128443927082e-06,
+      "loss": 1.0003,
+      "step": 17653
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.147128843729099e-06,
+      "loss": 0.9798,
+      "step": 17654
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.146129372617589e-06,
+      "loss": 0.9763,
+      "step": 17655
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.1451300306113706e-06,
+      "loss": 1.0213,
+      "step": 17656
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.1441308177292775e-06,
+      "loss": 0.9747,
+      "step": 17657
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.143131733990127e-06,
+      "loss": 0.956,
+      "step": 17658
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.142132779412744e-06,
+      "loss": 0.9481,
+      "step": 17659
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.141133954015935e-06,
+      "loss": 0.93,
+      "step": 17660
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.140135257818522e-06,
+      "loss": 0.9717,
+      "step": 17661
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.139136690839323e-06,
+      "loss": 0.9316,
+      "step": 17662
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.138138253097136e-06,
+      "loss": 0.998,
+      "step": 17663
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.137139944610772e-06,
+      "loss": 1.047,
+      "step": 17664
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.136141765399039e-06,
+      "loss": 1.0534,
+      "step": 17665
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.135143715480743e-06,
+      "loss": 0.9683,
+      "step": 17666
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.1341457948746754e-06,
+      "loss": 1.0213,
+      "step": 17667
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.133148003599636e-06,
+      "loss": 0.9991,
+      "step": 17668
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.132150341674422e-06,
+      "loss": 1.02,
+      "step": 17669
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.131152809117833e-06,
+      "loss": 0.9423,
+      "step": 17670
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.1301554059486425e-06,
+      "loss": 1.0182,
+      "step": 17671
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.129158132185653e-06,
+      "loss": 0.9676,
+      "step": 17672
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.128160987847649e-06,
+      "loss": 1.05,
+      "step": 17673
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.1271639729534076e-06,
+      "loss": 1.0121,
+      "step": 17674
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.126167087521707e-06,
+      "loss": 1.0216,
+      "step": 17675
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.125170331571333e-06,
+      "loss": 0.9863,
+      "step": 17676
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.124173705121063e-06,
+      "loss": 1.0181,
+      "step": 17677
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.1231772081896615e-06,
+      "loss": 0.9823,
+      "step": 17678
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.122180840795901e-06,
+      "loss": 1.0167,
+      "step": 17679
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.121184602958556e-06,
+      "loss": 1.0103,
+      "step": 17680
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.120188494696391e-06,
+      "loss": 1.0814,
+      "step": 17681
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.119192516028164e-06,
+      "loss": 0.925,
+      "step": 17682
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.1181966669726386e-06,
+      "loss": 1.0468,
+      "step": 17683
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.117200947548579e-06,
+      "loss": 1.0506,
+      "step": 17684
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.116205357774734e-06,
+      "loss": 1.0548,
+      "step": 17685
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.11520989766986e-06,
+      "loss": 1.0613,
+      "step": 17686
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.114214567252707e-06,
+      "loss": 0.9871,
+      "step": 17687
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.1132193665420305e-06,
+      "loss": 1.0364,
+      "step": 17688
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.112224295556569e-06,
+      "loss": 0.9982,
+      "step": 17689
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.11122935431507e-06,
+      "loss": 1.0767,
+      "step": 17690
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.1102345428362705e-06,
+      "loss": 0.9672,
+      "step": 17691
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.109239861138919e-06,
+      "loss": 1.0522,
+      "step": 17692
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.108245309241745e-06,
+      "loss": 0.9827,
+      "step": 17693
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.107250887163484e-06,
+      "loss": 0.984,
+      "step": 17694
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.106256594922868e-06,
+      "loss": 0.983,
+      "step": 17695
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.1052624325386276e-06,
+      "loss": 1.0473,
+      "step": 17696
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.104268400029487e-06,
+      "loss": 1.0209,
+      "step": 17697
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.103274497414174e-06,
+      "loss": 0.9628,
+      "step": 17698
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.102280724711407e-06,
+      "loss": 1.048,
+      "step": 17699
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.101287081939908e-06,
+      "loss": 1.0119,
+      "step": 17700
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.100293569118395e-06,
+      "loss": 1.0098,
+      "step": 17701
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.099300186265576e-06,
+      "loss": 1.0189,
+      "step": 17702
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.098306933400175e-06,
+      "loss": 0.9944,
+      "step": 17703
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.097313810540894e-06,
+      "loss": 0.9575,
+      "step": 17704
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.096320817706439e-06,
+      "loss": 1.0148,
+      "step": 17705
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0953279549155195e-06,
+      "loss": 0.9936,
+      "step": 17706
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.094335222186832e-06,
+      "loss": 1.0282,
+      "step": 17707
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.093342619539084e-06,
+      "loss": 0.9815,
+      "step": 17708
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.092350146990967e-06,
+      "loss": 0.9858,
+      "step": 17709
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.091357804561178e-06,
+      "loss": 0.988,
+      "step": 17710
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0903655922684105e-06,
+      "loss": 1.0252,
+      "step": 17711
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.089373510131354e-06,
+      "loss": 0.9983,
+      "step": 17712
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0883815581686964e-06,
+      "loss": 1.0852,
+      "step": 17713
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.087389736399121e-06,
+      "loss": 1.0341,
+      "step": 17714
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.086398044841315e-06,
+      "loss": 1.034,
+      "step": 17715
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.085406483513953e-06,
+      "loss": 1.0167,
+      "step": 17716
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.084415052435718e-06,
+      "loss": 1.0021,
+      "step": 17717
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.083423751625281e-06,
+      "loss": 0.9635,
+      "step": 17718
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0824325811013206e-06,
+      "loss": 0.9919,
+      "step": 17719
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0814415408825e-06,
+      "loss": 1.0053,
+      "step": 17720
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.080450630987495e-06,
+      "loss": 1.0522,
+      "step": 17721
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0794598514349645e-06,
+      "loss": 1.0712,
+      "step": 17722
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.078469202243577e-06,
+      "loss": 0.9954,
+      "step": 17723
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.077478683431993e-06,
+      "loss": 0.9899,
+      "step": 17724
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.076488295018862e-06,
+      "loss": 1.0518,
+      "step": 17725
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.07549803702285e-06,
+      "loss": 0.9616,
+      "step": 17726
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.074507909462606e-06,
+      "loss": 1.0165,
+      "step": 17727
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.073517912356786e-06,
+      "loss": 1.0068,
+      "step": 17728
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.072528045724027e-06,
+      "loss": 0.9756,
+      "step": 17729
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.071538309582985e-06,
+      "loss": 1.0384,
+      "step": 17730
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.070548703952301e-06,
+      "loss": 0.9641,
+      "step": 17731
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.069559228850615e-06,
+      "loss": 0.9557,
+      "step": 17732
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.068569884296567e-06,
+      "loss": 0.9828,
+      "step": 17733
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.067580670308792e-06,
+      "loss": 1.0302,
+      "step": 17734
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.066591586905926e-06,
+      "loss": 1.0302,
+      "step": 17735
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.065602634106593e-06,
+      "loss": 1.0417,
+      "step": 17736
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.06461381192943e-06,
+      "loss": 0.9859,
+      "step": 17737
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.063625120393059e-06,
+      "loss": 0.9703,
+      "step": 17738
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0626365595161105e-06,
+      "loss": 0.9979,
+      "step": 17739
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.061648129317191e-06,
+      "loss": 1.0083,
+      "step": 17740
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.060659829814934e-06,
+      "loss": 0.9956,
+      "step": 17741
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.059671661027951e-06,
+      "loss": 0.9449,
+      "step": 17742
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.058683622974857e-06,
+      "loss": 0.982,
+      "step": 17743
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.057695715674258e-06,
+      "loss": 0.9546,
+      "step": 17744
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.056707939144769e-06,
+      "loss": 0.9484,
+      "step": 17745
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.055720293404999e-06,
+      "loss": 1.0105,
+      "step": 17746
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.054732778473545e-06,
+      "loss": 0.9221,
+      "step": 17747
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.053745394369008e-06,
+      "loss": 0.9955,
+      "step": 17748
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.052758141109994e-06,
+      "loss": 0.9949,
+      "step": 17749
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.051771018715099e-06,
+      "loss": 1.035,
+      "step": 17750
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.050784027202913e-06,
+      "loss": 1.012,
+      "step": 17751
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.049797166592026e-06,
+      "loss": 0.9832,
+      "step": 17752
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.048810436901034e-06,
+      "loss": 0.9877,
+      "step": 17753
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.047823838148524e-06,
+      "loss": 0.9916,
+      "step": 17754
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.046837370353075e-06,
+      "loss": 0.9711,
+      "step": 17755
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.045851033533268e-06,
+      "loss": 0.9689,
+      "step": 17756
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.044864827707692e-06,
+      "loss": 0.9931,
+      "step": 17757
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.043878752894914e-06,
+      "loss": 0.9471,
+      "step": 17758
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0428928091135084e-06,
+      "loss": 1.0273,
+      "step": 17759
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.041906996382055e-06,
+      "loss": 0.9869,
+      "step": 17760
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.040921314719124e-06,
+      "loss": 1.0089,
+      "step": 17761
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0399357641432734e-06,
+      "loss": 0.9754,
+      "step": 17762
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.03895034467307e-06,
+      "loss": 1.0694,
+      "step": 17763
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.037965056327082e-06,
+      "loss": 1.0438,
+      "step": 17764
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.036979899123869e-06,
+      "loss": 1.0675,
+      "step": 17765
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.035994873081981e-06,
+      "loss": 1.0475,
+      "step": 17766
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.035009978219973e-06,
+      "loss": 1.0467,
+      "step": 17767
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.034025214556408e-06,
+      "loss": 0.944,
+      "step": 17768
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.033040582109826e-06,
+      "loss": 0.9868,
+      "step": 17769
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.032056080898777e-06,
+      "loss": 0.9764,
+      "step": 17770
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0310717109418025e-06,
+      "loss": 1.0599,
+      "step": 17771
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0300874722574555e-06,
+      "loss": 1.0292,
+      "step": 17772
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0291033648642664e-06,
+      "loss": 1.0331,
+      "step": 17773
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.028119388780774e-06,
+      "loss": 1.082,
+      "step": 17774
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.027135544025512e-06,
+      "loss": 1.027,
+      "step": 17775
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0261518306170205e-06,
+      "loss": 1.0589,
+      "step": 17776
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.025168248573824e-06,
+      "loss": 0.9862,
+      "step": 17777
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.024184797914449e-06,
+      "loss": 0.9778,
+      "step": 17778
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.023201478657423e-06,
+      "loss": 1.0166,
+      "step": 17779
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.022218290821267e-06,
+      "loss": 0.9775,
+      "step": 17780
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0212352344245024e-06,
+      "loss": 1.012,
+      "step": 17781
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.020252309485646e-06,
+      "loss": 0.9353,
+      "step": 17782
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.019269516023214e-06,
+      "loss": 1.01,
+      "step": 17783
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.01828685405572e-06,
+      "loss": 0.9723,
+      "step": 17784
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.017304323601671e-06,
+      "loss": 1.0838,
+      "step": 17785
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.016321924679577e-06,
+      "loss": 1.0173,
+      "step": 17786
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.015339657307944e-06,
+      "loss": 0.978,
+      "step": 17787
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.014357521505273e-06,
+      "loss": 0.9646,
+      "step": 17788
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.013375517290065e-06,
+      "loss": 0.9787,
+      "step": 17789
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.012393644680819e-06,
+      "loss": 0.9797,
+      "step": 17790
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.011411903696027e-06,
+      "loss": 1.0123,
+      "step": 17791
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.010430294354187e-06,
+      "loss": 1.045,
+      "step": 17792
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.009448816673785e-06,
+      "loss": 0.9709,
+      "step": 17793
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.008467470673312e-06,
+      "loss": 0.9939,
+      "step": 17794
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.007486256371251e-06,
+      "loss": 0.9398,
+      "step": 17795
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.006505173786086e-06,
+      "loss": 0.9494,
+      "step": 17796
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.005524222936297e-06,
+      "loss": 1.048,
+      "step": 17797
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.004543403840365e-06,
+      "loss": 1.0177,
+      "step": 17798
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.003562716516763e-06,
+      "loss": 0.9794,
+      "step": 17799
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0025821609839624e-06,
+      "loss": 1.0069,
+      "step": 17800
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.001601737260436e-06,
+      "loss": 0.9058,
+      "step": 17801
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.000621445364651e-06,
+      "loss": 1.0123,
+      "step": 17802
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.999641285315075e-06,
+      "loss": 0.9147,
+      "step": 17803
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.9986612571301674e-06,
+      "loss": 1.0483,
+      "step": 17804
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.997681360828393e-06,
+      "loss": 0.9745,
+      "step": 17805
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.9967015964282076e-06,
+      "loss": 1.0393,
+      "step": 17806
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.995721963948066e-06,
+      "loss": 0.9908,
+      "step": 17807
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.994742463406427e-06,
+      "loss": 0.9986,
+      "step": 17808
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.99376309482173e-06,
+      "loss": 0.991,
+      "step": 17809
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.992783858212433e-06,
+      "loss": 0.9424,
+      "step": 17810
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.991804753596981e-06,
+      "loss": 1.0337,
+      "step": 17811
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.990825780993817e-06,
+      "loss": 0.9914,
+      "step": 17812
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.989846940421372e-06,
+      "loss": 0.9159,
+      "step": 17813
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.9888682318980974e-06,
+      "loss": 0.979,
+      "step": 17814
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.987889655442423e-06,
+      "loss": 0.9814,
+      "step": 17815
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.986911211072786e-06,
+      "loss": 1.0422,
+      "step": 17816
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9859328988076075e-06,
+      "loss": 0.9987,
+      "step": 17817
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.984954718665326e-06,
+      "loss": 1.035,
+      "step": 17818
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9839766706643655e-06,
+      "loss": 1.0913,
+      "step": 17819
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.982998754823143e-06,
+      "loss": 0.9755,
+      "step": 17820
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.982020971160086e-06,
+      "loss": 0.9917,
+      "step": 17821
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.98104331969361e-06,
+      "loss": 0.9972,
+      "step": 17822
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.980065800442136e-06,
+      "loss": 1.0217,
+      "step": 17823
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.979088413424068e-06,
+      "loss": 1.0377,
+      "step": 17824
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.978111158657823e-06,
+      "loss": 1.0016,
+      "step": 17825
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.977134036161807e-06,
+      "loss": 0.9876,
+      "step": 17826
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.976157045954433e-06,
+      "loss": 0.96,
+      "step": 17827
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.975180188054091e-06,
+      "loss": 1.0092,
+      "step": 17828
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.974203462479193e-06,
+      "loss": 1.024,
+      "step": 17829
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.973226869248136e-06,
+      "loss": 1.0203,
+      "step": 17830
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.972250408379312e-06,
+      "loss": 0.9615,
+      "step": 17831
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.971274079891112e-06,
+      "loss": 0.9353,
+      "step": 17832
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.970297883801935e-06,
+      "loss": 0.997,
+      "step": 17833
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.969321820130169e-06,
+      "loss": 1.0396,
+      "step": 17834
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.968345888894192e-06,
+      "loss": 1.033,
+      "step": 17835
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.967370090112392e-06,
+      "loss": 1.0023,
+      "step": 17836
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.966394423803152e-06,
+      "loss": 0.9764,
+      "step": 17837
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9654188899848504e-06,
+      "loss": 0.9617,
+      "step": 17838
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.964443488675859e-06,
+      "loss": 1.0168,
+      "step": 17839
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.963468219894551e-06,
+      "loss": 1.0022,
+      "step": 17840
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.962493083659306e-06,
+      "loss": 0.993,
+      "step": 17841
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.961518079988482e-06,
+      "loss": 1.0016,
+      "step": 17842
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.960543208900451e-06,
+      "loss": 1.037,
+      "step": 17843
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.959568470413572e-06,
+      "loss": 1.0127,
+      "step": 17844
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.958593864546214e-06,
+      "loss": 1.0869,
+      "step": 17845
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.957619391316728e-06,
+      "loss": 0.9896,
+      "step": 17846
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9566450507434695e-06,
+      "loss": 0.9396,
+      "step": 17847
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.955670842844798e-06,
+      "loss": 0.9868,
+      "step": 17848
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.954696767639066e-06,
+      "loss": 1.012,
+      "step": 17849
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.953722825144613e-06,
+      "loss": 1.0166,
+      "step": 17850
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9527490153797855e-06,
+      "loss": 0.9468,
+      "step": 17851
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.951775338362939e-06,
+      "loss": 0.9812,
+      "step": 17852
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.950801794112402e-06,
+      "loss": 0.9737,
+      "step": 17853
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.949828382646518e-06,
+      "loss": 1.0456,
+      "step": 17854
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.94885510398362e-06,
+      "loss": 1.0384,
+      "step": 17855
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.947881958142049e-06,
+      "loss": 0.9493,
+      "step": 17856
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.946908945140129e-06,
+      "loss": 1.1351,
+      "step": 17857
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.94593606499619e-06,
+      "loss": 1.0397,
+      "step": 17858
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.944963317728555e-06,
+      "loss": 0.9578,
+      "step": 17859
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.943990703355558e-06,
+      "loss": 1.0264,
+      "step": 17860
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.943018221895509e-06,
+      "loss": 1.0695,
+      "step": 17861
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9420458733667315e-06,
+      "loss": 0.9105,
+      "step": 17862
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.94107365778754e-06,
+      "loss": 0.9839,
+      "step": 17863
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.940101575176249e-06,
+      "loss": 1.0139,
+      "step": 17864
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.939129625551168e-06,
+      "loss": 0.961,
+      "step": 17865
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.938157808930606e-06,
+      "loss": 0.9563,
+      "step": 17866
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.937186125332869e-06,
+      "loss": 0.9963,
+      "step": 17867
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9362145747762625e-06,
+      "loss": 1.0269,
+      "step": 17868
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.935243157279084e-06,
+      "loss": 0.9299,
+      "step": 17869
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.934271872859635e-06,
+      "loss": 1.0651,
+      "step": 17870
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.933300721536209e-06,
+      "loss": 0.9814,
+      "step": 17871
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.932329703327101e-06,
+      "loss": 0.9901,
+      "step": 17872
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.931358818250602e-06,
+      "loss": 0.9947,
+      "step": 17873
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.930388066324999e-06,
+      "loss": 0.999,
+      "step": 17874
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9294174475685785e-06,
+      "loss": 0.9701,
+      "step": 17875
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.928446961999625e-06,
+      "loss": 1.0717,
+      "step": 17876
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9274766096364165e-06,
+      "loss": 1.008,
+      "step": 17877
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.926506390497236e-06,
+      "loss": 0.9469,
+      "step": 17878
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.925536304600356e-06,
+      "loss": 0.9975,
+      "step": 17879
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.92456635196405e-06,
+      "loss": 0.9913,
+      "step": 17880
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.923596532606589e-06,
+      "loss": 0.9675,
+      "step": 17881
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9226268465462425e-06,
+      "loss": 1.0642,
+      "step": 17882
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.921657293801277e-06,
+      "loss": 0.9401,
+      "step": 17883
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.920687874389953e-06,
+      "loss": 1.0171,
+      "step": 17884
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.919718588330532e-06,
+      "loss": 1.0106,
+      "step": 17885
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.918749435641274e-06,
+      "loss": 1.0001,
+      "step": 17886
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.917780416340433e-06,
+      "loss": 0.9787,
+      "step": 17887
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.916811530446264e-06,
+      "loss": 0.9588,
+      "step": 17888
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.915842777977017e-06,
+      "loss": 0.9455,
+      "step": 17889
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.914874158950938e-06,
+      "loss": 0.9512,
+      "step": 17890
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.913905673386278e-06,
+      "loss": 1.0094,
+      "step": 17891
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.912937321301278e-06,
+      "loss": 0.9599,
+      "step": 17892
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.911969102714171e-06,
+      "loss": 1.0008,
+      "step": 17893
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.911001017643207e-06,
+      "loss": 1.063,
+      "step": 17894
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.910033066106616e-06,
+      "loss": 1.0178,
+      "step": 17895
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.909065248122633e-06,
+      "loss": 1.0162,
+      "step": 17896
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.908097563709483e-06,
+      "loss": 1.0022,
+      "step": 17897
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9071300128854e-06,
+      "loss": 1.0756,
+      "step": 17898
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.90616259566861e-06,
+      "loss": 0.9577,
+      "step": 17899
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.905195312077337e-06,
+      "loss": 0.9533,
+      "step": 17900
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9042281621297916e-06,
+      "loss": 0.9957,
+      "step": 17901
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.903261145844202e-06,
+      "loss": 1.0073,
+      "step": 17902
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.902294263238785e-06,
+      "loss": 0.9796,
+      "step": 17903
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.901327514331745e-06,
+      "loss": 0.9246,
+      "step": 17904
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.900360899141295e-06,
+      "loss": 0.9891,
+      "step": 17905
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.899394417685646e-06,
+      "loss": 0.9838,
+      "step": 17906
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.898428069983007e-06,
+      "loss": 1.0449,
+      "step": 17907
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.897461856051569e-06,
+      "loss": 1.0241,
+      "step": 17908
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.896495775909543e-06,
+      "loss": 0.9512,
+      "step": 17909
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.895529829575124e-06,
+      "loss": 1.0057,
+      "step": 17910
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.89456401706651e-06,
+      "loss": 0.9732,
+      "step": 17911
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.893598338401884e-06,
+      "loss": 0.9971,
+      "step": 17912
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.892632793599446e-06,
+      "loss": 1.0171,
+      "step": 17913
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8916673826773855e-06,
+      "loss": 0.9638,
+      "step": 17914
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.89070210565388e-06,
+      "loss": 0.9876,
+      "step": 17915
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8897369625471104e-06,
+      "loss": 1.032,
+      "step": 17916
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.888771953375267e-06,
+      "loss": 1.013,
+      "step": 17917
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8878070781565246e-06,
+      "loss": 1.1262,
+      "step": 17918
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.886842336909054e-06,
+      "loss": 0.9009,
+      "step": 17919
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.885877729651027e-06,
+      "loss": 1.0184,
+      "step": 17920
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.88491325640062e-06,
+      "loss": 1.0577,
+      "step": 17921
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.883948917176002e-06,
+      "loss": 0.9667,
+      "step": 17922
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8829847119953296e-06,
+      "loss": 0.961,
+      "step": 17923
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.882020640876766e-06,
+      "loss": 0.9835,
+      "step": 17924
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.881056703838481e-06,
+      "loss": 0.9548,
+      "step": 17925
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.880092900898623e-06,
+      "loss": 0.9436,
+      "step": 17926
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.87912923207535e-06,
+      "loss": 1.0246,
+      "step": 17927
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8781656973868125e-06,
+      "loss": 0.9876,
+      "step": 17928
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.877202296851167e-06,
+      "loss": 0.9634,
+      "step": 17929
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.876239030486554e-06,
+      "loss": 1.0462,
+      "step": 17930
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8752758983111205e-06,
+      "loss": 1.0237,
+      "step": 17931
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.874312900343006e-06,
+      "loss": 1.041,
+      "step": 17932
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.873350036600359e-06,
+      "loss": 1.0847,
+      "step": 17933
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.872387307101309e-06,
+      "loss": 1.0094,
+      "step": 17934
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.87142471186399e-06,
+      "loss": 0.9889,
+      "step": 17935
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.870462250906544e-06,
+      "loss": 1.0113,
+      "step": 17936
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.869499924247089e-06,
+      "loss": 0.9858,
+      "step": 17937
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8685377319037586e-06,
+      "loss": 1.0148,
+      "step": 17938
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8675756738946726e-06,
+      "loss": 0.9342,
+      "step": 17939
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.866613750237963e-06,
+      "loss": 1.0146,
+      "step": 17940
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.865651960951739e-06,
+      "loss": 1.0346,
+      "step": 17941
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.864690306054122e-06,
+      "loss": 0.9109,
+      "step": 17942
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8637287855632255e-06,
+      "loss": 0.9977,
+      "step": 17943
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.862767399497165e-06,
+      "loss": 1.0171,
+      "step": 17944
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.861806147874047e-06,
+      "loss": 0.9848,
+      "step": 17945
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.860845030711977e-06,
+      "loss": 0.9701,
+      "step": 17946
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.859884048029062e-06,
+      "loss": 0.9615,
+      "step": 17947
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8589231998434036e-06,
+      "loss": 0.9445,
+      "step": 17948
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8579624861730975e-06,
+      "loss": 0.9994,
+      "step": 17949
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.857001907036247e-06,
+      "loss": 0.9943,
+      "step": 17950
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8560414624509405e-06,
+      "loss": 1.0377,
+      "step": 17951
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.855081152435273e-06,
+      "loss": 0.983,
+      "step": 17952
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.854120977007333e-06,
+      "loss": 0.9374,
+      "step": 17953
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.853160936185207e-06,
+      "loss": 0.9751,
+      "step": 17954
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.852201029986979e-06,
+      "loss": 0.9541,
+      "step": 17955
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.851241258430731e-06,
+      "loss": 1.0018,
+      "step": 17956
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.850281621534541e-06,
+      "loss": 0.974,
+      "step": 17957
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8493221193164885e-06,
+      "loss": 0.9325,
+      "step": 17958
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.848362751794645e-06,
+      "loss": 1.0634,
+      "step": 17959
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.84740351898708e-06,
+      "loss": 1.074,
+      "step": 17960
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.846444420911866e-06,
+      "loss": 0.9568,
+      "step": 17961
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.84548545758707e-06,
+      "loss": 1.0216,
+      "step": 17962
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.844526629030751e-06,
+      "loss": 1.0525,
+      "step": 17963
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.843567935260975e-06,
+      "loss": 1.0029,
+      "step": 17964
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8426093762957985e-06,
+      "loss": 0.968,
+      "step": 17965
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.841650952153279e-06,
+      "loss": 0.9891,
+      "step": 17966
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.840692662851468e-06,
+      "loss": 0.9593,
+      "step": 17967
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.83973450840842e-06,
+      "loss": 1.046,
+      "step": 17968
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.838776488842181e-06,
+      "loss": 0.9693,
+      "step": 17969
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.837818604170797e-06,
+      "loss": 1.0125,
+      "step": 17970
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8368608544123135e-06,
+      "loss": 0.9747,
+      "step": 17971
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.83590323958477e-06,
+      "loss": 0.954,
+      "step": 17972
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.834945759706205e-06,
+      "loss": 1.0266,
+      "step": 17973
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.833988414794654e-06,
+      "loss": 0.9788,
+      "step": 17974
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8330312048681514e-06,
+      "loss": 0.9477,
+      "step": 17975
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.83207412994473e-06,
+      "loss": 0.9928,
+      "step": 17976
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.831117190042411e-06,
+      "loss": 1.0285,
+      "step": 17977
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.830160385179228e-06,
+      "loss": 1.0318,
+      "step": 17978
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8292037153732e-06,
+      "loss": 0.9961,
+      "step": 17979
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.828247180642353e-06,
+      "loss": 1.0203,
+      "step": 17980
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.827290781004693e-06,
+      "loss": 1.0377,
+      "step": 17981
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.826334516478248e-06,
+      "loss": 1.0272,
+      "step": 17982
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.825378387081025e-06,
+      "loss": 0.9443,
+      "step": 17983
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.824422392831039e-06,
+      "loss": 0.9099,
+      "step": 17984
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.823466533746289e-06,
+      "loss": 0.981,
+      "step": 17985
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.822510809844788e-06,
+      "loss": 0.9624,
+      "step": 17986
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.821555221144541e-06,
+      "loss": 1.0238,
+      "step": 17987
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.82059976766354e-06,
+      "loss": 1.029,
+      "step": 17988
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8196444494197855e-06,
+      "loss": 1.0255,
+      "step": 17989
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.818689266431276e-06,
+      "loss": 0.9152,
+      "step": 17990
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.817734218716005e-06,
+      "loss": 1.0136,
+      "step": 17991
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.816779306291957e-06,
+      "loss": 1.0603,
+      "step": 17992
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.815824529177119e-06,
+      "loss": 1.0458,
+      "step": 17993
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8148698873894835e-06,
+      "loss": 0.999,
+      "step": 17994
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.813915380947031e-06,
+      "loss": 0.9951,
+      "step": 17995
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.812961009867733e-06,
+      "loss": 1.0509,
+      "step": 17996
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.812006774169576e-06,
+      "loss": 1.0142,
+      "step": 17997
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.811052673870535e-06,
+      "loss": 1.0789,
+      "step": 17998
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.810098708988575e-06,
+      "loss": 1.0122,
+      "step": 17999
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.809144879541668e-06,
+      "loss": 0.9767,
+      "step": 18000
+    },
+    {
+      "epoch": 0.76,
+      "eval_text_loss": 0.761838972568512,
+      "eval_text_runtime": 1.3928,
+      "eval_text_samples_per_second": 53.13,
+      "eval_text_steps_per_second": 1.436,
+      "step": 18000
+    },
+    {
+      "epoch": 0.76,
+      "eval_svit_loss": 1.082850456237793,
+      "eval_svit_runtime": 5.1053,
+      "eval_svit_samples_per_second": 133.392,
+      "eval_svit_steps_per_second": 2.155,
+      "step": 18000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.808191185547784e-06,
+      "loss": 1.0064,
+      "step": 18001
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.807237627024892e-06,
+      "loss": 1.0037,
+      "step": 18002
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.806284203990944e-06,
+      "loss": 0.9547,
+      "step": 18003
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8053309164639025e-06,
+      "loss": 0.9824,
+      "step": 18004
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8043777644617265e-06,
+      "loss": 0.9734,
+      "step": 18005
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.803424748002373e-06,
+      "loss": 1.032,
+      "step": 18006
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8024718671037885e-06,
+      "loss": 1.0308,
+      "step": 18007
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8015191217839204e-06,
+      "loss": 0.9728,
+      "step": 18008
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.800566512060727e-06,
+      "loss": 1.0518,
+      "step": 18009
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.79961403795214e-06,
+      "loss": 0.9578,
+      "step": 18010
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.798661699476107e-06,
+      "loss": 1.0321,
+      "step": 18011
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.797709496650562e-06,
+      "loss": 1.0197,
+      "step": 18012
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.796757429493452e-06,
+      "loss": 0.9569,
+      "step": 18013
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.795805498022701e-06,
+      "loss": 0.9457,
+      "step": 18014
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.794853702256243e-06,
+      "loss": 0.9361,
+      "step": 18015
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.793902042212006e-06,
+      "loss": 0.9859,
+      "step": 18016
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.792950517907922e-06,
+      "loss": 1.0062,
+      "step": 18017
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.79199912936191e-06,
+      "loss": 1.0583,
+      "step": 18018
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.7910478765918905e-06,
+      "loss": 0.9789,
+      "step": 18019
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.790096759615782e-06,
+      "loss": 1.128,
+      "step": 18020
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.789145778451503e-06,
+      "loss": 0.9902,
+      "step": 18021
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.788194933116965e-06,
+      "loss": 0.9659,
+      "step": 18022
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.787244223630078e-06,
+      "loss": 0.9175,
+      "step": 18023
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.7862936500087565e-06,
+      "loss": 1.0417,
+      "step": 18024
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.7853432122709e-06,
+      "loss": 1.0691,
+      "step": 18025
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.784392910434413e-06,
+      "loss": 0.9601,
+      "step": 18026
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.783442744517196e-06,
+      "loss": 0.9697,
+      "step": 18027
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.78249271453715e-06,
+      "loss": 0.9779,
+      "step": 18028
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.781542820512168e-06,
+      "loss": 1.0182,
+      "step": 18029
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.780593062460142e-06,
+      "loss": 0.9658,
+      "step": 18030
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.779643440398967e-06,
+      "loss": 1.0107,
+      "step": 18031
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.778693954346525e-06,
+      "loss": 1.0426,
+      "step": 18032
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.777744604320706e-06,
+      "loss": 0.9866,
+      "step": 18033
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.776795390339389e-06,
+      "loss": 0.9673,
+      "step": 18034
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.775846312420457e-06,
+      "loss": 0.9848,
+      "step": 18035
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.774897370581787e-06,
+      "loss": 0.9575,
+      "step": 18036
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.773948564841254e-06,
+      "loss": 0.9465,
+      "step": 18037
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.77299989521673e-06,
+      "loss": 1.0204,
+      "step": 18038
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.772051361726086e-06,
+      "loss": 0.9687,
+      "step": 18039
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.771102964387188e-06,
+      "loss": 1.0163,
+      "step": 18040
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.770154703217901e-06,
+      "loss": 0.9675,
+      "step": 18041
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.7692065782360875e-06,
+      "loss": 0.977,
+      "step": 18042
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.7682585894596096e-06,
+      "loss": 1.0321,
+      "step": 18043
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.76731073690632e-06,
+      "loss": 0.9624,
+      "step": 18044
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.766363020594076e-06,
+      "loss": 1.0217,
+      "step": 18045
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.765415440540729e-06,
+      "loss": 0.9741,
+      "step": 18046
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.764467996764128e-06,
+      "loss": 1.03,
+      "step": 18047
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.7635206892821215e-06,
+      "loss": 0.966,
+      "step": 18048
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.762573518112556e-06,
+      "loss": 0.9789,
+      "step": 18049
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.761626483273262e-06,
+      "loss": 1.0227,
+      "step": 18050
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.760679584782091e-06,
+      "loss": 1.0259,
+      "step": 18051
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.759732822656874e-06,
+      "loss": 0.9342,
+      "step": 18052
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.758786196915447e-06,
+      "loss": 0.9829,
+      "step": 18053
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.757839707575641e-06,
+      "loss": 0.9368,
+      "step": 18054
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.756893354655282e-06,
+      "loss": 1.0955,
+      "step": 18055
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.7559471381722e-06,
+      "loss": 1.051,
+      "step": 18056
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.755001058144218e-06,
+      "loss": 0.9971,
+      "step": 18057
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.754055114589156e-06,
+      "loss": 1.1141,
+      "step": 18058
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.753109307524834e-06,
+      "loss": 1.0804,
+      "step": 18059
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.752163636969068e-06,
+      "loss": 0.9228,
+      "step": 18060
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.751218102939667e-06,
+      "loss": 0.9945,
+      "step": 18061
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.750272705454447e-06,
+      "loss": 0.8941,
+      "step": 18062
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.7493274445312155e-06,
+      "loss": 1.0045,
+      "step": 18063
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.748382320187781e-06,
+      "loss": 1.0018,
+      "step": 18064
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.747437332441937e-06,
+      "loss": 1.026,
+      "step": 18065
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.7464924813114925e-06,
+      "loss": 1.0003,
+      "step": 18066
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.745547766814244e-06,
+      "loss": 1.0221,
+      "step": 18067
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.744603188967989e-06,
+      "loss": 0.96,
+      "step": 18068
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.743658747790512e-06,
+      "loss": 0.9427,
+      "step": 18069
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.742714443299612e-06,
+      "loss": 1.0713,
+      "step": 18070
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.741770275513076e-06,
+      "loss": 1.0178,
+      "step": 18071
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.740826244448684e-06,
+      "loss": 1.0129,
+      "step": 18072
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.739882350124219e-06,
+      "loss": 1.076,
+      "step": 18073
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.738938592557465e-06,
+      "loss": 1.1017,
+      "step": 18074
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.737994971766201e-06,
+      "loss": 1.0244,
+      "step": 18075
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.737051487768197e-06,
+      "loss": 0.96,
+      "step": 18076
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.736108140581223e-06,
+      "loss": 0.9736,
+      "step": 18077
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.735164930223055e-06,
+      "loss": 1.0438,
+      "step": 18078
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.734221856711461e-06,
+      "loss": 1.0152,
+      "step": 18079
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.7332789200642e-06,
+      "loss": 1.0255,
+      "step": 18080
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.732336120299032e-06,
+      "loss": 1.0078,
+      "step": 18081
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.731393457433726e-06,
+      "loss": 0.9149,
+      "step": 18082
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.730450931486032e-06,
+      "loss": 1.0295,
+      "step": 18083
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.729508542473702e-06,
+      "loss": 1.0408,
+      "step": 18084
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.728566290414493e-06,
+      "loss": 1.0664,
+      "step": 18085
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.727624175326156e-06,
+      "loss": 0.9315,
+      "step": 18086
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.726682197226431e-06,
+      "loss": 1.0105,
+      "step": 18087
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.725740356133062e-06,
+      "loss": 0.9417,
+      "step": 18088
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.724798652063795e-06,
+      "loss": 1.0215,
+      "step": 18089
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.72385708503637e-06,
+      "loss": 1.0575,
+      "step": 18090
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.722915655068516e-06,
+      "loss": 0.9904,
+      "step": 18091
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.721974362177968e-06,
+      "loss": 0.9773,
+      "step": 18092
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.721033206382466e-06,
+      "loss": 0.9859,
+      "step": 18093
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.720092187699727e-06,
+      "loss": 1.0264,
+      "step": 18094
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.719151306147483e-06,
+      "loss": 0.9929,
+      "step": 18095
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.7182105617434515e-06,
+      "loss": 1.0219,
+      "step": 18096
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.717269954505365e-06,
+      "loss": 1.0297,
+      "step": 18097
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.716329484450929e-06,
+      "loss": 1.0452,
+      "step": 18098
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.715389151597866e-06,
+      "loss": 0.966,
+      "step": 18099
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.714448955963883e-06,
+      "loss": 0.9626,
+      "step": 18100
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.7135088975667e-06,
+      "loss": 0.9738,
+      "step": 18101
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.712568976424017e-06,
+      "loss": 1.0484,
+      "step": 18102
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.71162919255354e-06,
+      "loss": 0.9278,
+      "step": 18103
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.7106895459729725e-06,
+      "loss": 0.9565,
+      "step": 18104
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.709750036700017e-06,
+      "loss": 0.9996,
+      "step": 18105
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.7088106647523654e-06,
+      "loss": 0.9982,
+      "step": 18106
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.7078714301477185e-06,
+      "loss": 1.0639,
+      "step": 18107
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.7069323329037636e-06,
+      "loss": 0.9551,
+      "step": 18108
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.705993373038193e-06,
+      "loss": 1.0305,
+      "step": 18109
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.7050545505686926e-06,
+      "loss": 0.9782,
+      "step": 18110
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.704115865512945e-06,
+      "loss": 1.056,
+      "step": 18111
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.703177317888641e-06,
+      "loss": 0.9117,
+      "step": 18112
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.7022389077134485e-06,
+      "loss": 1.0448,
+      "step": 18113
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.701300635005052e-06,
+      "loss": 0.9741,
+      "step": 18114
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.700362499781121e-06,
+      "loss": 0.944,
+      "step": 18115
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.699424502059328e-06,
+      "loss": 0.9745,
+      "step": 18116
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.698486641857343e-06,
+      "loss": 1.0272,
+      "step": 18117
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.697548919192832e-06,
+      "loss": 0.9676,
+      "step": 18118
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.696611334083459e-06,
+      "loss": 0.9435,
+      "step": 18119
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.695673886546883e-06,
+      "loss": 0.9733,
+      "step": 18120
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.694736576600766e-06,
+      "loss": 1.0026,
+      "step": 18121
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.693799404262762e-06,
+      "loss": 0.983,
+      "step": 18122
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6928623695505254e-06,
+      "loss": 0.9721,
+      "step": 18123
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.691925472481705e-06,
+      "loss": 0.9652,
+      "step": 18124
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.69098871307395e-06,
+      "loss": 0.9956,
+      "step": 18125
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.690052091344907e-06,
+      "loss": 1.0018,
+      "step": 18126
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6891156073122164e-06,
+      "loss": 1.0169,
+      "step": 18127
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.688179260993523e-06,
+      "loss": 0.9427,
+      "step": 18128
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.687243052406461e-06,
+      "loss": 1.0202,
+      "step": 18129
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.686306981568666e-06,
+      "loss": 0.9426,
+      "step": 18130
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6853710484977725e-06,
+      "loss": 0.9634,
+      "step": 18131
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.684435253211409e-06,
+      "loss": 0.9844,
+      "step": 18132
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6834995957272045e-06,
+      "loss": 1.0032,
+      "step": 18133
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.682564076062777e-06,
+      "loss": 1.0432,
+      "step": 18134
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.681628694235758e-06,
+      "loss": 0.9989,
+      "step": 18135
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.680693450263763e-06,
+      "loss": 1.031,
+      "step": 18136
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6797583441644135e-06,
+      "loss": 0.9854,
+      "step": 18137
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.678823375955314e-06,
+      "loss": 1.0315,
+      "step": 18138
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.677888545654083e-06,
+      "loss": 0.9485,
+      "step": 18139
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.676953853278331e-06,
+      "loss": 0.9425,
+      "step": 18140
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6760192988456626e-06,
+      "loss": 0.9599,
+      "step": 18141
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.675084882373682e-06,
+      "loss": 1.0023,
+      "step": 18142
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.674150603879989e-06,
+      "loss": 1.041,
+      "step": 18143
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.673216463382189e-06,
+      "loss": 1.0314,
+      "step": 18144
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6722824608978656e-06,
+      "loss": 0.9867,
+      "step": 18145
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.671348596444624e-06,
+      "loss": 0.9615,
+      "step": 18146
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.670414870040052e-06,
+      "loss": 0.9995,
+      "step": 18147
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.669481281701739e-06,
+      "loss": 0.9595,
+      "step": 18148
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6685478314472646e-06,
+      "loss": 0.9009,
+      "step": 18149
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.667614519294219e-06,
+      "loss": 0.962,
+      "step": 18150
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.666681345260181e-06,
+      "loss": 1.0651,
+      "step": 18151
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.665748309362731e-06,
+      "loss": 1.0227,
+      "step": 18152
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.664815411619437e-06,
+      "loss": 1.0192,
+      "step": 18153
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.663882652047879e-06,
+      "loss": 1.0413,
+      "step": 18154
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.662950030665628e-06,
+      "loss": 0.975,
+      "step": 18155
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.662017547490244e-06,
+      "loss": 0.9629,
+      "step": 18156
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.661085202539294e-06,
+      "loss": 1.038,
+      "step": 18157
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.660152995830348e-06,
+      "loss": 0.9259,
+      "step": 18158
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.659220927380961e-06,
+      "loss": 0.9862,
+      "step": 18159
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.658288997208688e-06,
+      "loss": 1.002,
+      "step": 18160
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.657357205331084e-06,
+      "loss": 0.9256,
+      "step": 18161
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.656425551765704e-06,
+      "loss": 0.9091,
+      "step": 18162
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6554940365301e-06,
+      "loss": 0.9341,
+      "step": 18163
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.654562659641812e-06,
+      "loss": 0.9643,
+      "step": 18164
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.653631421118383e-06,
+      "loss": 1.0182,
+      "step": 18165
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.652700320977366e-06,
+      "loss": 1.1325,
+      "step": 18166
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.651769359236289e-06,
+      "loss": 0.9609,
+      "step": 18167
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6508385359126925e-06,
+      "loss": 1.0013,
+      "step": 18168
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.649907851024106e-06,
+      "loss": 1.0102,
+      "step": 18169
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.648977304588072e-06,
+      "loss": 0.9132,
+      "step": 18170
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.648046896622108e-06,
+      "loss": 1.0345,
+      "step": 18171
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.64711662714374e-06,
+      "loss": 0.9639,
+      "step": 18172
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.646186496170497e-06,
+      "loss": 0.9627,
+      "step": 18173
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6452565037199014e-06,
+      "loss": 0.9825,
+      "step": 18174
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6443266498094655e-06,
+      "loss": 1.0173,
+      "step": 18175
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.643396934456703e-06,
+      "loss": 1.0888,
+      "step": 18176
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.642467357679137e-06,
+      "loss": 0.9821,
+      "step": 18177
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.641537919494268e-06,
+      "loss": 1.0588,
+      "step": 18178
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.640608619919607e-06,
+      "loss": 0.933,
+      "step": 18179
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.639679458972657e-06,
+      "loss": 1.0439,
+      "step": 18180
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.638750436670928e-06,
+      "loss": 1.0525,
+      "step": 18181
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6378215530319116e-06,
+      "loss": 1.003,
+      "step": 18182
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.636892808073107e-06,
+      "loss": 0.9772,
+      "step": 18183
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6359642018120075e-06,
+      "loss": 1.07,
+      "step": 18184
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.635035734266115e-06,
+      "loss": 1.0176,
+      "step": 18185
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.634107405452906e-06,
+      "loss": 0.9565,
+      "step": 18186
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.633179215389874e-06,
+      "loss": 0.9488,
+      "step": 18187
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6322511640945e-06,
+      "loss": 1.017,
+      "step": 18188
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6313232515842695e-06,
+      "loss": 0.9342,
+      "step": 18189
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.630395477876657e-06,
+      "loss": 0.9752,
+      "step": 18190
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6294678429891425e-06,
+      "loss": 1.0036,
+      "step": 18191
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.628540346939198e-06,
+      "loss": 0.9466,
+      "step": 18192
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.627612989744295e-06,
+      "loss": 0.9676,
+      "step": 18193
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.626685771421901e-06,
+      "loss": 0.9559,
+      "step": 18194
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.625758691989484e-06,
+      "loss": 0.9726,
+      "step": 18195
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.624831751464505e-06,
+      "loss": 1.0677,
+      "step": 18196
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.623904949864425e-06,
+      "loss": 0.9455,
+      "step": 18197
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.622978287206704e-06,
+      "loss": 1.0051,
+      "step": 18198
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6220517635087955e-06,
+      "loss": 0.9801,
+      "step": 18199
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.621125378788152e-06,
+      "loss": 1.007,
+      "step": 18200
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.620199133062227e-06,
+      "loss": 0.9927,
+      "step": 18201
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.619273026348464e-06,
+      "loss": 1.0465,
+      "step": 18202
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.618347058664309e-06,
+      "loss": 0.9933,
+      "step": 18203
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.617421230027204e-06,
+      "loss": 0.9175,
+      "step": 18204
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.616495540454591e-06,
+      "loss": 0.9691,
+      "step": 18205
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.615569989963905e-06,
+      "loss": 0.9639,
+      "step": 18206
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.614644578572582e-06,
+      "loss": 1.0565,
+      "step": 18207
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6137193062980505e-06,
+      "loss": 0.9851,
+      "step": 18208
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.612794173157744e-06,
+      "loss": 1.0474,
+      "step": 18209
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.611869179169086e-06,
+      "loss": 0.9957,
+      "step": 18210
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6109443243495015e-06,
+      "loss": 0.9944,
+      "step": 18211
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.610019608716413e-06,
+      "loss": 0.9674,
+      "step": 18212
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.609095032287238e-06,
+      "loss": 1.0641,
+      "step": 18213
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.608170595079391e-06,
+      "loss": 1.0742,
+      "step": 18214
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.607246297110289e-06,
+      "loss": 1.0251,
+      "step": 18215
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6063221383973405e-06,
+      "loss": 0.96,
+      "step": 18216
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.605398118957957e-06,
+      "loss": 1.0304,
+      "step": 18217
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.604474238809537e-06,
+      "loss": 1.0175,
+      "step": 18218
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.603550497969489e-06,
+      "loss": 1.017,
+      "step": 18219
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.602626896455212e-06,
+      "loss": 0.9164,
+      "step": 18220
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.601703434284108e-06,
+      "loss": 0.9727,
+      "step": 18221
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.600780111473563e-06,
+      "loss": 1.0422,
+      "step": 18222
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.599856928040977e-06,
+      "loss": 0.9811,
+      "step": 18223
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.598933884003737e-06,
+      "loss": 1.0207,
+      "step": 18224
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.598010979379234e-06,
+      "loss": 0.989,
+      "step": 18225
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.597088214184843e-06,
+      "loss": 0.9623,
+      "step": 18226
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.596165588437955e-06,
+      "loss": 1.0391,
+      "step": 18227
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.595243102155951e-06,
+      "loss": 1.0479,
+      "step": 18228
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.594320755356197e-06,
+      "loss": 0.9425,
+      "step": 18229
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5933985480560754e-06,
+      "loss": 1.0269,
+      "step": 18230
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.592476480272956e-06,
+      "loss": 1.0251,
+      "step": 18231
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.591554552024211e-06,
+      "loss": 0.9086,
+      "step": 18232
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.590632763327197e-06,
+      "loss": 0.9586,
+      "step": 18233
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.589711114199287e-06,
+      "loss": 0.9776,
+      "step": 18234
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.588789604657838e-06,
+      "loss": 1.0146,
+      "step": 18235
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.587868234720212e-06,
+      "loss": 0.9346,
+      "step": 18236
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.586947004403756e-06,
+      "loss": 0.9618,
+      "step": 18237
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.586025913725831e-06,
+      "loss": 1.0178,
+      "step": 18238
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.585104962703789e-06,
+      "loss": 0.9201,
+      "step": 18239
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.58418415135497e-06,
+      "loss": 1.0937,
+      "step": 18240
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.583263479696721e-06,
+      "loss": 1.0204,
+      "step": 18241
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5823429477463895e-06,
+      "loss": 1.0134,
+      "step": 18242
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.581422555521315e-06,
+      "loss": 0.9602,
+      "step": 18243
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5805023030388305e-06,
+      "loss": 0.9899,
+      "step": 18244
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.57958219031627e-06,
+      "loss": 0.9708,
+      "step": 18245
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5786622173709695e-06,
+      "loss": 1.0661,
+      "step": 18246
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5777423842202606e-06,
+      "loss": 1.0283,
+      "step": 18247
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.576822690881463e-06,
+      "loss": 0.9866,
+      "step": 18248
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.575903137371901e-06,
+      "loss": 0.9912,
+      "step": 18249
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.574983723708905e-06,
+      "loss": 1.0402,
+      "step": 18250
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.574064449909787e-06,
+      "loss": 0.9933,
+      "step": 18251
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.573145315991863e-06,
+      "loss": 0.9543,
+      "step": 18252
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5722263219724465e-06,
+      "loss": 0.9843,
+      "step": 18253
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.571307467868854e-06,
+      "loss": 0.9934,
+      "step": 18254
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.570388753698387e-06,
+      "loss": 0.9835,
+      "step": 18255
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5694701794783555e-06,
+      "loss": 1.0397,
+      "step": 18256
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5685517452260565e-06,
+      "loss": 0.9871,
+      "step": 18257
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.567633450958801e-06,
+      "loss": 0.9321,
+      "step": 18258
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.566715296693878e-06,
+      "loss": 1.0459,
+      "step": 18259
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.565797282448582e-06,
+      "loss": 1.028,
+      "step": 18260
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.564879408240215e-06,
+      "loss": 0.9196,
+      "step": 18261
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.563961674086057e-06,
+      "loss": 0.9503,
+      "step": 18262
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.563044080003399e-06,
+      "loss": 0.963,
+      "step": 18263
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.562126626009522e-06,
+      "loss": 1.0242,
+      "step": 18264
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.561209312121718e-06,
+      "loss": 1.051,
+      "step": 18265
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.560292138357258e-06,
+      "loss": 1.0052,
+      "step": 18266
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.559375104733418e-06,
+      "loss": 0.9817,
+      "step": 18267
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.558458211267472e-06,
+      "loss": 0.9051,
+      "step": 18268
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5575414579767e-06,
+      "loss": 1.0228,
+      "step": 18269
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.55662484487836e-06,
+      "loss": 0.9602,
+      "step": 18270
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.555708371989724e-06,
+      "loss": 1.0666,
+      "step": 18271
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.554792039328055e-06,
+      "loss": 0.9531,
+      "step": 18272
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.553875846910611e-06,
+      "loss": 0.9348,
+      "step": 18273
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.552959794754653e-06,
+      "loss": 0.958,
+      "step": 18274
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.552043882877434e-06,
+      "loss": 0.9685,
+      "step": 18275
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5511281112962095e-06,
+      "loss": 0.9566,
+      "step": 18276
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.550212480028228e-06,
+      "loss": 1.0542,
+      "step": 18277
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.549296989090738e-06,
+      "loss": 0.9548,
+      "step": 18278
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.548381638500985e-06,
+      "loss": 1.009,
+      "step": 18279
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.54746642827621e-06,
+      "loss": 0.9965,
+      "step": 18280
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.546551358433653e-06,
+      "loss": 1.0018,
+      "step": 18281
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5456364289905506e-06,
+      "loss": 0.9915,
+      "step": 18282
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.544721639964139e-06,
+      "loss": 0.9565,
+      "step": 18283
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.543806991371648e-06,
+      "loss": 0.9381,
+      "step": 18284
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5428924832303075e-06,
+      "loss": 0.9222,
+      "step": 18285
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.541978115557344e-06,
+      "loss": 0.9941,
+      "step": 18286
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.541063888369981e-06,
+      "loss": 0.9654,
+      "step": 18287
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.540149801685441e-06,
+      "loss": 0.9832,
+      "step": 18288
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.53923585552094e-06,
+      "loss": 0.9442,
+      "step": 18289
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.538322049893695e-06,
+      "loss": 0.9612,
+      "step": 18290
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.5374083848209205e-06,
+      "loss": 0.9968,
+      "step": 18291
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.536494860319826e-06,
+      "loss": 1.0646,
+      "step": 18292
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.535581476407619e-06,
+      "loss": 0.9775,
+      "step": 18293
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.534668233101505e-06,
+      "loss": 0.9644,
+      "step": 18294
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.533755130418687e-06,
+      "loss": 0.9903,
+      "step": 18295
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.532842168376365e-06,
+      "loss": 1.0292,
+      "step": 18296
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.531929346991736e-06,
+      "loss": 0.9969,
+      "step": 18297
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.531016666281995e-06,
+      "loss": 1.0216,
+      "step": 18298
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.530104126264332e-06,
+      "loss": 0.9699,
+      "step": 18299
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.529191726955941e-06,
+      "loss": 0.9883,
+      "step": 18300
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.5282794683740065e-06,
+      "loss": 0.9867,
+      "step": 18301
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.527367350535708e-06,
+      "loss": 0.977,
+      "step": 18302
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.5264553734582325e-06,
+      "loss": 0.9422,
+      "step": 18303
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.525543537158758e-06,
+      "loss": 0.9398,
+      "step": 18304
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.524631841654461e-06,
+      "loss": 0.9948,
+      "step": 18305
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.5237202869625095e-06,
+      "loss": 1.0036,
+      "step": 18306
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.522808873100081e-06,
+      "loss": 1.0353,
+      "step": 18307
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.521897600084342e-06,
+      "loss": 0.9936,
+      "step": 18308
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.5209864679324585e-06,
+      "loss": 0.9669,
+      "step": 18309
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.5200754766615875e-06,
+      "loss": 1.0525,
+      "step": 18310
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.519164626288897e-06,
+      "loss": 1.0082,
+      "step": 18311
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.518253916831544e-06,
+      "loss": 0.9392,
+      "step": 18312
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.517343348306678e-06,
+      "loss": 1.0532,
+      "step": 18313
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.51643292073145e-06,
+      "loss": 0.992,
+      "step": 18314
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.515522634123018e-06,
+      "loss": 1.0435,
+      "step": 18315
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.514612488498526e-06,
+      "loss": 1.0089,
+      "step": 18316
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.513702483875111e-06,
+      "loss": 0.988,
+      "step": 18317
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.512792620269923e-06,
+      "loss": 0.9905,
+      "step": 18318
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.511882897700099e-06,
+      "loss": 0.9992,
+      "step": 18319
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.5109733161827786e-06,
+      "loss": 0.9956,
+      "step": 18320
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.510063875735085e-06,
+      "loss": 0.982,
+      "step": 18321
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.509154576374158e-06,
+      "loss": 1.0105,
+      "step": 18322
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.5082454181171265e-06,
+      "loss": 1.0744,
+      "step": 18323
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.5073364009811106e-06,
+      "loss": 1.0069,
+      "step": 18324
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.506427524983233e-06,
+      "loss": 0.9312,
+      "step": 18325
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.505518790140621e-06,
+      "loss": 0.9628,
+      "step": 18326
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.50461019647039e-06,
+      "loss": 0.9762,
+      "step": 18327
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.50370174398965e-06,
+      "loss": 1.0073,
+      "step": 18328
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.502793432715514e-06,
+      "loss": 0.9982,
+      "step": 18329
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.501885262665099e-06,
+      "loss": 1.0032,
+      "step": 18330
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.500977233855509e-06,
+      "loss": 0.9297,
+      "step": 18331
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.500069346303845e-06,
+      "loss": 1.0262,
+      "step": 18332
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.499161600027207e-06,
+      "loss": 1.0354,
+      "step": 18333
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.498253995042704e-06,
+      "loss": 1.0268,
+      "step": 18334
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.497346531367423e-06,
+      "loss": 0.9565,
+      "step": 18335
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.496439209018461e-06,
+      "loss": 1.0157,
+      "step": 18336
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.495532028012906e-06,
+      "loss": 0.957,
+      "step": 18337
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.494624988367855e-06,
+      "loss": 0.9153,
+      "step": 18338
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.493718090100385e-06,
+      "loss": 0.9678,
+      "step": 18339
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.492811333227583e-06,
+      "loss": 1.0075,
+      "step": 18340
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.491904717766525e-06,
+      "loss": 0.9959,
+      "step": 18341
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.490998243734299e-06,
+      "loss": 1.0364,
+      "step": 18342
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.490091911147971e-06,
+      "loss": 0.9334,
+      "step": 18343
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.489185720024616e-06,
+      "loss": 0.9646,
+      "step": 18344
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.488279670381304e-06,
+      "loss": 0.9629,
+      "step": 18345
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4873737622351015e-06,
+      "loss": 0.9664,
+      "step": 18346
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.486467995603073e-06,
+      "loss": 0.9436,
+      "step": 18347
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.485562370502279e-06,
+      "loss": 1.0318,
+      "step": 18348
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.484656886949786e-06,
+      "loss": 1.015,
+      "step": 18349
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.48375154496264e-06,
+      "loss": 1.0484,
+      "step": 18350
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.482846344557901e-06,
+      "loss": 0.9844,
+      "step": 18351
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.481941285752616e-06,
+      "loss": 0.978,
+      "step": 18352
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.481036368563842e-06,
+      "loss": 1.0825,
+      "step": 18353
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4801315930086146e-06,
+      "loss": 0.9932,
+      "step": 18354
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.479226959103982e-06,
+      "loss": 0.8881,
+      "step": 18355
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.478322466866985e-06,
+      "loss": 0.9228,
+      "step": 18356
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.477418116314658e-06,
+      "loss": 0.9458,
+      "step": 18357
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.476513907464041e-06,
+      "loss": 0.9925,
+      "step": 18358
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4756098403321615e-06,
+      "loss": 1.0279,
+      "step": 18359
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.474705914936053e-06,
+      "loss": 1.0107,
+      "step": 18360
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.473802131292741e-06,
+      "loss": 0.9554,
+      "step": 18361
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.47289848941925e-06,
+      "loss": 1.0447,
+      "step": 18362
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.471994989332601e-06,
+      "loss": 0.9541,
+      "step": 18363
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.471091631049816e-06,
+      "loss": 0.9543,
+      "step": 18364
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4701884145879075e-06,
+      "loss": 1.0237,
+      "step": 18365
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.469285339963892e-06,
+      "loss": 0.9354,
+      "step": 18366
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.468382407194779e-06,
+      "loss": 1.0205,
+      "step": 18367
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.467479616297578e-06,
+      "loss": 0.9532,
+      "step": 18368
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4665769672892945e-06,
+      "loss": 1.0319,
+      "step": 18369
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.465674460186931e-06,
+      "loss": 1.0257,
+      "step": 18370
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.46477209500749e-06,
+      "loss": 0.994,
+      "step": 18371
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.463869871767965e-06,
+      "loss": 0.9659,
+      "step": 18372
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.462967790485354e-06,
+      "loss": 0.9775,
+      "step": 18373
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.46206585117665e-06,
+      "loss": 1.0052,
+      "step": 18374
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.461164053858842e-06,
+      "loss": 1.033,
+      "step": 18375
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.460262398548916e-06,
+      "loss": 0.9468,
+      "step": 18376
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.459360885263856e-06,
+      "loss": 1.0412,
+      "step": 18377
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.458459514020646e-06,
+      "loss": 1.0149,
+      "step": 18378
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.457558284836263e-06,
+      "loss": 1.0022,
+      "step": 18379
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.456657197727685e-06,
+      "loss": 0.9834,
+      "step": 18380
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.455756252711883e-06,
+      "loss": 1.0107,
+      "step": 18381
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.454855449805829e-06,
+      "loss": 1.0144,
+      "step": 18382
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.453954789026494e-06,
+      "loss": 1.0485,
+      "step": 18383
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.45305427039084e-06,
+      "loss": 0.993,
+      "step": 18384
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.452153893915836e-06,
+      "loss": 0.8978,
+      "step": 18385
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.451253659618429e-06,
+      "loss": 0.9639,
+      "step": 18386
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.45035356751559e-06,
+      "loss": 0.9565,
+      "step": 18387
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.449453617624268e-06,
+      "loss": 0.9475,
+      "step": 18388
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.44855380996142e-06,
+      "loss": 1.0226,
+      "step": 18389
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.447654144543986e-06,
+      "loss": 1.0473,
+      "step": 18390
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.446754621388922e-06,
+      "loss": 1.039,
+      "step": 18391
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.445855240513167e-06,
+      "loss": 1.0486,
+      "step": 18392
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.44495600193367e-06,
+      "loss": 1.0582,
+      "step": 18393
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.444056905667356e-06,
+      "loss": 1.0073,
+      "step": 18394
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.443157951731173e-06,
+      "loss": 1.0286,
+      "step": 18395
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.442259140142054e-06,
+      "loss": 1.0044,
+      "step": 18396
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.441360470916922e-06,
+      "loss": 0.9484,
+      "step": 18397
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.440461944072709e-06,
+      "loss": 0.9682,
+      "step": 18398
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.439563559626343e-06,
+      "loss": 0.9727,
+      "step": 18399
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.438665317594748e-06,
+      "loss": 0.9678,
+      "step": 18400
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.437767217994837e-06,
+      "loss": 1.0352,
+      "step": 18401
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.436869260843529e-06,
+      "loss": 1.0465,
+      "step": 18402
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.435971446157744e-06,
+      "loss": 1.0029,
+      "step": 18403
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.435073773954394e-06,
+      "loss": 0.9872,
+      "step": 18404
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.434176244250379e-06,
+      "loss": 0.9331,
+      "step": 18405
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.433278857062614e-06,
+      "loss": 1.0219,
+      "step": 18406
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.432381612408005e-06,
+      "loss": 1.0372,
+      "step": 18407
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.431484510303445e-06,
+      "loss": 1.026,
+      "step": 18408
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.430587550765834e-06,
+      "loss": 0.9911,
+      "step": 18409
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.429690733812073e-06,
+      "loss": 0.9691,
+      "step": 18410
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.428794059459056e-06,
+      "loss": 0.9983,
+      "step": 18411
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.427897527723668e-06,
+      "loss": 1.0462,
+      "step": 18412
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.427001138622795e-06,
+      "loss": 0.9511,
+      "step": 18413
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4261048921733294e-06,
+      "loss": 1.0264,
+      "step": 18414
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.425208788392153e-06,
+      "loss": 0.9473,
+      "step": 18415
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4243128272961415e-06,
+      "loss": 0.9921,
+      "step": 18416
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4234170089021694e-06,
+      "loss": 0.985,
+      "step": 18417
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.42252133322712e-06,
+      "loss": 0.9202,
+      "step": 18418
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.421625800287859e-06,
+      "loss": 1.0007,
+      "step": 18419
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.420730410101255e-06,
+      "loss": 1.0196,
+      "step": 18420
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.419835162684174e-06,
+      "loss": 0.9768,
+      "step": 18421
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.418940058053487e-06,
+      "loss": 0.9463,
+      "step": 18422
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.418045096226047e-06,
+      "loss": 1.0334,
+      "step": 18423
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.417150277218715e-06,
+      "loss": 0.919,
+      "step": 18424
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4162556010483425e-06,
+      "loss": 0.9559,
+      "step": 18425
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.415361067731793e-06,
+      "loss": 0.9293,
+      "step": 18426
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.414466677285906e-06,
+      "loss": 0.9558,
+      "step": 18427
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.413572429727533e-06,
+      "loss": 1.0272,
+      "step": 18428
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.412678325073518e-06,
+      "loss": 1.0053,
+      "step": 18429
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4117843633407054e-06,
+      "loss": 0.9813,
+      "step": 18430
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.410890544545931e-06,
+      "loss": 0.985,
+      "step": 18431
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.409996868706036e-06,
+      "loss": 0.8822,
+      "step": 18432
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.40910333583785e-06,
+      "loss": 0.9538,
+      "step": 18433
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.408209945958207e-06,
+      "loss": 0.9838,
+      "step": 18434
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.407316699083936e-06,
+      "loss": 1.0031,
+      "step": 18435
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.406423595231857e-06,
+      "loss": 1.0426,
+      "step": 18436
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.405530634418806e-06,
+      "loss": 1.0105,
+      "step": 18437
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.40463781666159e-06,
+      "loss": 1.0316,
+      "step": 18438
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.403745141977035e-06,
+      "loss": 0.9789,
+      "step": 18439
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.402852610381952e-06,
+      "loss": 0.9901,
+      "step": 18440
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.401960221893157e-06,
+      "loss": 1.0303,
+      "step": 18441
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.401067976527457e-06,
+      "loss": 0.9916,
+      "step": 18442
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.40017587430166e-06,
+      "loss": 1.0344,
+      "step": 18443
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.399283915232571e-06,
+      "loss": 1.0473,
+      "step": 18444
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3983920993369916e-06,
+      "loss": 0.9215,
+      "step": 18445
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.397500426631719e-06,
+      "loss": 1.0635,
+      "step": 18446
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.396608897133552e-06,
+      "loss": 0.9399,
+      "step": 18447
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.395717510859283e-06,
+      "loss": 1.0031,
+      "step": 18448
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3948262678257025e-06,
+      "loss": 0.9235,
+      "step": 18449
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3939351680496e-06,
+      "loss": 0.9764,
+      "step": 18450
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.393044211547759e-06,
+      "loss": 0.9905,
+      "step": 18451
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.392153398336964e-06,
+      "loss": 1.0402,
+      "step": 18452
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.391262728433995e-06,
+      "loss": 1.0179,
+      "step": 18453
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.390372201855628e-06,
+      "loss": 0.9982,
+      "step": 18454
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.389481818618641e-06,
+      "loss": 0.9557,
+      "step": 18455
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3885915787398015e-06,
+      "loss": 0.9438,
+      "step": 18456
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.387701482235882e-06,
+      "loss": 1.017,
+      "step": 18457
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3868115291236514e-06,
+      "loss": 1.0326,
+      "step": 18458
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3859217194198636e-06,
+      "loss": 0.9361,
+      "step": 18459
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.385032053141291e-06,
+      "loss": 1.105,
+      "step": 18460
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.384142530304686e-06,
+      "loss": 1.0382,
+      "step": 18461
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.383253150926808e-06,
+      "loss": 1.0419,
+      "step": 18462
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.382363915024406e-06,
+      "loss": 0.9856,
+      "step": 18463
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.381474822614233e-06,
+      "loss": 0.9453,
+      "step": 18464
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.380585873713038e-06,
+      "loss": 0.9657,
+      "step": 18465
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.379697068337563e-06,
+      "loss": 1.009,
+      "step": 18466
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.378808406504552e-06,
+      "loss": 1.0299,
+      "step": 18467
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.377919888230744e-06,
+      "loss": 1.0421,
+      "step": 18468
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3770315135328795e-06,
+      "loss": 0.8901,
+      "step": 18469
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.376143282427683e-06,
+      "loss": 0.9925,
+      "step": 18470
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.375255194931895e-06,
+      "loss": 1.016,
+      "step": 18471
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3743672510622416e-06,
+      "loss": 1.036,
+      "step": 18472
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.37347945083545e-06,
+      "loss": 1.0378,
+      "step": 18473
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.372591794268239e-06,
+      "loss": 0.9385,
+      "step": 18474
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.371704281377335e-06,
+      "loss": 0.9351,
+      "step": 18475
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.370816912179452e-06,
+      "loss": 0.9846,
+      "step": 18476
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.369929686691311e-06,
+      "loss": 0.9764,
+      "step": 18477
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.369042604929614e-06,
+      "loss": 0.996,
+      "step": 18478
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3681556669110795e-06,
+      "loss": 1.011,
+      "step": 18479
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.367268872652416e-06,
+      "loss": 0.915,
+      "step": 18480
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.36638222217032e-06,
+      "loss": 1.01,
+      "step": 18481
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.365495715481495e-06,
+      "loss": 0.9924,
+      "step": 18482
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.364609352602646e-06,
+      "loss": 1.0301,
+      "step": 18483
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3637231335504674e-06,
+      "loss": 1.0498,
+      "step": 18484
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3628370583416486e-06,
+      "loss": 1.0342,
+      "step": 18485
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.361951126992878e-06,
+      "loss": 0.9668,
+      "step": 18486
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3610653395208535e-06,
+      "loss": 0.8781,
+      "step": 18487
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.360179695942258e-06,
+      "loss": 1.0276,
+      "step": 18488
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.359294196273769e-06,
+      "loss": 1.0132,
+      "step": 18489
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.358408840532066e-06,
+      "loss": 0.9808,
+      "step": 18490
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.357523628733836e-06,
+      "loss": 0.9787,
+      "step": 18491
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.356638560895744e-06,
+      "loss": 0.995,
+      "step": 18492
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3557536370344625e-06,
+      "loss": 0.9503,
+      "step": 18493
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.354868857166665e-06,
+      "loss": 0.9779,
+      "step": 18494
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.35398422130902e-06,
+      "loss": 0.9841,
+      "step": 18495
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.353099729478186e-06,
+      "loss": 1.0073,
+      "step": 18496
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.352215381690821e-06,
+      "loss": 1.0444,
+      "step": 18497
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.351331177963591e-06,
+      "loss": 0.895,
+      "step": 18498
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3504471183131505e-06,
+      "loss": 1.0517,
+      "step": 18499
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.349563202756149e-06,
+      "loss": 1.0018,
+      "step": 18500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.348679431309235e-06,
+      "loss": 0.9581,
+      "step": 18501
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.347795803989064e-06,
+      "loss": 0.9864,
+      "step": 18502
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.346912320812272e-06,
+      "loss": 0.9873,
+      "step": 18503
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.346028981795507e-06,
+      "loss": 1.0032,
+      "step": 18504
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3451457869554015e-06,
+      "loss": 0.9626,
+      "step": 18505
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3442627363086025e-06,
+      "loss": 1.0098,
+      "step": 18506
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.343379829871735e-06,
+      "loss": 1.0419,
+      "step": 18507
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.342497067661435e-06,
+      "loss": 1.0596,
+      "step": 18508
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.341614449694326e-06,
+      "loss": 1.007,
+      "step": 18509
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.340731975987042e-06,
+      "loss": 1.073,
+      "step": 18510
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.339849646556198e-06,
+      "loss": 0.986,
+      "step": 18511
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.338967461418418e-06,
+      "loss": 1.0046,
+      "step": 18512
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.338085420590318e-06,
+      "loss": 1.0055,
+      "step": 18513
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3372035240885135e-06,
+      "loss": 0.9599,
+      "step": 18514
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.336321771929618e-06,
+      "loss": 0.9066,
+      "step": 18515
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.335440164130239e-06,
+      "loss": 1.0526,
+      "step": 18516
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.334558700706985e-06,
+      "loss": 0.9951,
+      "step": 18517
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.333677381676458e-06,
+      "loss": 0.9196,
+      "step": 18518
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.33279620705526e-06,
+      "loss": 0.9435,
+      "step": 18519
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.33191517685999e-06,
+      "loss": 0.9767,
+      "step": 18520
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.331034291107244e-06,
+      "loss": 0.9392,
+      "step": 18521
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.330153549813615e-06,
+      "loss": 1.0188,
+      "step": 18522
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3292729529956935e-06,
+      "loss": 1.0073,
+      "step": 18523
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.328392500670066e-06,
+      "loss": 0.9227,
+      "step": 18524
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.327512192853319e-06,
+      "loss": 1.0805,
+      "step": 18525
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.326632029562034e-06,
+      "loss": 1.1072,
+      "step": 18526
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3257520108127894e-06,
+      "loss": 0.9918,
+      "step": 18527
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.324872136622164e-06,
+      "loss": 0.995,
+      "step": 18528
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.32399240700673e-06,
+      "loss": 0.9712,
+      "step": 18529
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.323112821983061e-06,
+      "loss": 0.991,
+      "step": 18530
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3222333815677236e-06,
+      "loss": 0.904,
+      "step": 18531
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.321354085777284e-06,
+      "loss": 0.987,
+      "step": 18532
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3204749346283065e-06,
+      "loss": 1.0123,
+      "step": 18533
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3195959281373495e-06,
+      "loss": 1.0273,
+      "step": 18534
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.318717066320972e-06,
+      "loss": 0.962,
+      "step": 18535
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3178383491957295e-06,
+      "loss": 0.9774,
+      "step": 18536
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3169597767781735e-06,
+      "loss": 0.9951,
+      "step": 18537
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.316081349084854e-06,
+      "loss": 0.9718,
+      "step": 18538
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.315203066132318e-06,
+      "loss": 0.9919,
+      "step": 18539
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.314324927937109e-06,
+      "loss": 0.9519,
+      "step": 18540
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.313446934515767e-06,
+      "loss": 1.0611,
+      "step": 18541
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3125690858848364e-06,
+      "loss": 0.9684,
+      "step": 18542
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3116913820608416e-06,
+      "loss": 1.0499,
+      "step": 18543
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3108138230603275e-06,
+      "loss": 1.0971,
+      "step": 18544
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.309936408899818e-06,
+      "loss": 1.0241,
+      "step": 18545
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.309059139595848e-06,
+      "loss": 1.025,
+      "step": 18546
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.30818201516493e-06,
+      "loss": 1.0151,
+      "step": 18547
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3073050356235975e-06,
+      "loss": 1.0348,
+      "step": 18548
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.306428200988365e-06,
+      "loss": 0.9089,
+      "step": 18549
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.305551511275751e-06,
+      "loss": 1.0192,
+      "step": 18550
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.304674966502268e-06,
+      "loss": 0.935,
+      "step": 18551
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.303798566684429e-06,
+      "loss": 1.0582,
+      "step": 18552
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.302922311838746e-06,
+      "loss": 0.9563,
+      "step": 18553
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3020462019817136e-06,
+      "loss": 1.0685,
+      "step": 18554
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.301170237129845e-06,
+      "loss": 1.0009,
+      "step": 18555
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.300294417299638e-06,
+      "loss": 1.0058,
+      "step": 18556
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2994187425075916e-06,
+      "loss": 0.8911,
+      "step": 18557
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.298543212770194e-06,
+      "loss": 1.0297,
+      "step": 18558
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.297667828103946e-06,
+      "loss": 0.9708,
+      "step": 18559
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.296792588525335e-06,
+      "loss": 0.9915,
+      "step": 18560
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.295917494050847e-06,
+      "loss": 1.0033,
+      "step": 18561
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2950425446969604e-06,
+      "loss": 0.8998,
+      "step": 18562
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.294167740480165e-06,
+      "loss": 0.9904,
+      "step": 18563
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.293293081416938e-06,
+      "loss": 0.9761,
+      "step": 18564
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.292418567523751e-06,
+      "loss": 1.0016,
+      "step": 18565
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.291544198817075e-06,
+      "loss": 0.9808,
+      "step": 18566
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.290669975313388e-06,
+      "loss": 0.9678,
+      "step": 18567
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.289795897029157e-06,
+      "loss": 1.0066,
+      "step": 18568
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.288921963980841e-06,
+      "loss": 1.0038,
+      "step": 18569
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.288048176184904e-06,
+      "loss": 1.038,
+      "step": 18570
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2871745336578075e-06,
+      "loss": 1.0741,
+      "step": 18571
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.28630103641601e-06,
+      "loss": 0.9944,
+      "step": 18572
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.28542768447596e-06,
+      "loss": 0.9909,
+      "step": 18573
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2845544778541094e-06,
+      "loss": 0.9786,
+      "step": 18574
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.283681416566913e-06,
+      "loss": 1.0347,
+      "step": 18575
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.28280850063081e-06,
+      "loss": 1.0907,
+      "step": 18576
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.281935730062245e-06,
+      "loss": 1.0275,
+      "step": 18577
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.281063104877657e-06,
+      "loss": 0.917,
+      "step": 18578
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.280190625093491e-06,
+      "loss": 0.9663,
+      "step": 18579
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2793182907261706e-06,
+      "loss": 0.9298,
+      "step": 18580
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.278446101792133e-06,
+      "loss": 1.0144,
+      "step": 18581
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.27757405830781e-06,
+      "loss": 0.9962,
+      "step": 18582
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.276702160289628e-06,
+      "loss": 0.8749,
+      "step": 18583
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.275830407754006e-06,
+      "loss": 0.9291,
+      "step": 18584
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.274958800717366e-06,
+      "loss": 0.9385,
+      "step": 18585
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.274087339196132e-06,
+      "loss": 0.9594,
+      "step": 18586
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.273216023206714e-06,
+      "loss": 0.9911,
+      "step": 18587
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.272344852765527e-06,
+      "loss": 1.0363,
+      "step": 18588
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.271473827888977e-06,
+      "loss": 0.9356,
+      "step": 18589
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.270602948593482e-06,
+      "loss": 0.991,
+      "step": 18590
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.269732214895433e-06,
+      "loss": 0.9829,
+      "step": 18591
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.268861626811242e-06,
+      "loss": 0.9601,
+      "step": 18592
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.267991184357301e-06,
+      "loss": 1.0273,
+      "step": 18593
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.267120887550014e-06,
+      "loss": 1.0411,
+      "step": 18594
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.26625073640577e-06,
+      "loss": 0.9713,
+      "step": 18595
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.26538073094096e-06,
+      "loss": 0.9345,
+      "step": 18596
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.264510871171972e-06,
+      "loss": 0.9654,
+      "step": 18597
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2636411571151925e-06,
+      "loss": 0.9448,
+      "step": 18598
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.262771588787004e-06,
+      "loss": 0.9888,
+      "step": 18599
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.261902166203785e-06,
+      "loss": 1.0252,
+      "step": 18600
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.261032889381916e-06,
+      "loss": 1.0391,
+      "step": 18601
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.260163758337768e-06,
+      "loss": 0.9552,
+      "step": 18602
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.259294773087713e-06,
+      "loss": 0.9928,
+      "step": 18603
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.258425933648122e-06,
+      "loss": 1.001,
+      "step": 18604
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.257557240035361e-06,
+      "loss": 1.0123,
+      "step": 18605
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.256688692265792e-06,
+      "loss": 0.9377,
+      "step": 18606
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.255820290355775e-06,
+      "loss": 0.9978,
+      "step": 18607
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.25495203432167e-06,
+      "loss": 0.9075,
+      "step": 18608
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.254083924179831e-06,
+      "loss": 1.0101,
+      "step": 18609
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.253215959946612e-06,
+      "loss": 0.9987,
+      "step": 18610
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.252348141638362e-06,
+      "loss": 0.9375,
+      "step": 18611
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.251480469271427e-06,
+      "loss": 0.9857,
+      "step": 18612
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.25061294286215e-06,
+      "loss": 1.098,
+      "step": 18613
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.249745562426876e-06,
+      "loss": 0.9954,
+      "step": 18614
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2488783279819415e-06,
+      "loss": 0.9747,
+      "step": 18615
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.248011239543683e-06,
+      "loss": 0.9094,
+      "step": 18616
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2471442971284335e-06,
+      "loss": 0.992,
+      "step": 18617
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.246277500752524e-06,
+      "loss": 1.0912,
+      "step": 18618
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.245410850432281e-06,
+      "loss": 1.1049,
+      "step": 18619
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.244544346184031e-06,
+      "loss": 0.9919,
+      "step": 18620
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.243677988024094e-06,
+      "loss": 1.098,
+      "step": 18621
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2428117759687926e-06,
+      "loss": 0.9688,
+      "step": 18622
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.24194571003444e-06,
+      "loss": 0.9703,
+      "step": 18623
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.241079790237355e-06,
+      "loss": 1.0092,
+      "step": 18624
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2402140165938435e-06,
+      "loss": 0.9724,
+      "step": 18625
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.239348389120219e-06,
+      "loss": 0.8967,
+      "step": 18626
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.238482907832778e-06,
+      "loss": 0.8874,
+      "step": 18627
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.237617572747834e-06,
+      "loss": 0.8832,
+      "step": 18628
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.236752383881683e-06,
+      "loss": 1.0481,
+      "step": 18629
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.235887341250625e-06,
+      "loss": 1.0198,
+      "step": 18630
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.235022444870946e-06,
+      "loss": 1.012,
+      "step": 18631
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2341576947589476e-06,
+      "loss": 0.9607,
+      "step": 18632
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2332930909309145e-06,
+      "loss": 0.9706,
+      "step": 18633
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.232428633403137e-06,
+      "loss": 0.938,
+      "step": 18634
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.231564322191891e-06,
+      "loss": 1.1048,
+      "step": 18635
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2307001573134645e-06,
+      "loss": 0.9548,
+      "step": 18636
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.229836138784136e-06,
+      "loss": 0.9797,
+      "step": 18637
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.228972266620172e-06,
+      "loss": 0.9571,
+      "step": 18638
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.228108540837855e-06,
+      "loss": 1.0226,
+      "step": 18639
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2272449614534505e-06,
+      "loss": 1.0114,
+      "step": 18640
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2263815284832296e-06,
+      "loss": 1.0148,
+      "step": 18641
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2255182419434465e-06,
+      "loss": 1.0861,
+      "step": 18642
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.224655101850374e-06,
+      "loss": 1.0405,
+      "step": 18643
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.223792108220265e-06,
+      "loss": 1.0418,
+      "step": 18644
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2229292610693805e-06,
+      "loss": 1.0361,
+      "step": 18645
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.222066560413966e-06,
+      "loss": 1.0661,
+      "step": 18646
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.221204006270277e-06,
+      "loss": 0.9855,
+      "step": 18647
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2203415986545654e-06,
+      "loss": 0.9972,
+      "step": 18648
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.219479337583069e-06,
+      "loss": 1.1094,
+      "step": 18649
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.218617223072029e-06,
+      "loss": 0.9023,
+      "step": 18650
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.217755255137692e-06,
+      "loss": 1.0703,
+      "step": 18651
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.216893433796294e-06,
+      "loss": 0.9912,
+      "step": 18652
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2160317590640624e-06,
+      "loss": 0.9922,
+      "step": 18653
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.215170230957231e-06,
+      "loss": 0.9787,
+      "step": 18654
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.214308849492031e-06,
+      "loss": 1.0338,
+      "step": 18655
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.213447614684692e-06,
+      "loss": 1.0231,
+      "step": 18656
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2125865265514275e-06,
+      "loss": 0.9273,
+      "step": 18657
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.21172558510846e-06,
+      "loss": 0.926,
+      "step": 18658
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.210864790372014e-06,
+      "loss": 0.8927,
+      "step": 18659
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.210004142358295e-06,
+      "loss": 1.0213,
+      "step": 18660
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2091436410835194e-06,
+      "loss": 0.9836,
+      "step": 18661
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.208283286563894e-06,
+      "loss": 1.0182,
+      "step": 18662
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.207423078815632e-06,
+      "loss": 0.9815,
+      "step": 18663
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2065630178549295e-06,
+      "loss": 1.0496,
+      "step": 18664
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.205703103697989e-06,
+      "loss": 0.9411,
+      "step": 18665
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.204843336361008e-06,
+      "loss": 1.0481,
+      "step": 18666
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.203983715860188e-06,
+      "loss": 0.9873,
+      "step": 18667
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.203124242211713e-06,
+      "loss": 0.9855,
+      "step": 18668
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2022649154317745e-06,
+      "loss": 0.9458,
+      "step": 18669
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.201405735536567e-06,
+      "loss": 1.0135,
+      "step": 18670
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.200546702542266e-06,
+      "loss": 1.022,
+      "step": 18671
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.199687816465058e-06,
+      "loss": 0.9762,
+      "step": 18672
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.198829077321117e-06,
+      "loss": 0.9487,
+      "step": 18673
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.197970485126625e-06,
+      "loss": 0.9433,
+      "step": 18674
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.197112039897751e-06,
+      "loss": 1.0241,
+      "step": 18675
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.196253741650665e-06,
+      "loss": 0.9857,
+      "step": 18676
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.195395590401536e-06,
+      "loss": 1.0055,
+      "step": 18677
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.194537586166532e-06,
+      "loss": 0.9419,
+      "step": 18678
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1936797289618105e-06,
+      "loss": 0.9113,
+      "step": 18679
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.192822018803532e-06,
+      "loss": 0.9456,
+      "step": 18680
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.191964455707854e-06,
+      "loss": 1.1011,
+      "step": 18681
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.191107039690928e-06,
+      "loss": 0.9938,
+      "step": 18682
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.190249770768908e-06,
+      "loss": 0.8863,
+      "step": 18683
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.18939264895794e-06,
+      "loss": 1.0215,
+      "step": 18684
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.188535674274171e-06,
+      "loss": 1.0218,
+      "step": 18685
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.187678846733743e-06,
+      "loss": 1.0421,
+      "step": 18686
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1868221663527964e-06,
+      "loss": 1.0121,
+      "step": 18687
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.185965633147468e-06,
+      "loss": 0.9855,
+      "step": 18688
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.185109247133891e-06,
+      "loss": 1.0232,
+      "step": 18689
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.184253008328199e-06,
+      "loss": 1.036,
+      "step": 18690
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1833969167465205e-06,
+      "loss": 0.8948,
+      "step": 18691
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.182540972404981e-06,
+      "loss": 0.9833,
+      "step": 18692
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.181685175319703e-06,
+      "loss": 1.0308,
+      "step": 18693
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.180829525506808e-06,
+      "loss": 0.9622,
+      "step": 18694
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.179974022982414e-06,
+      "loss": 0.9353,
+      "step": 18695
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.179118667762635e-06,
+      "loss": 1.0371,
+      "step": 18696
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.178263459863585e-06,
+      "loss": 0.9585,
+      "step": 18697
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1774083993013715e-06,
+      "loss": 1.0214,
+      "step": 18698
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.176553486092101e-06,
+      "loss": 0.8941,
+      "step": 18699
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.17569872025188e-06,
+      "loss": 0.9788,
+      "step": 18700
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.174844101796807e-06,
+      "loss": 1.0507,
+      "step": 18701
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.173989630742981e-06,
+      "loss": 1.0507,
+      "step": 18702
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.173135307106498e-06,
+      "loss": 0.9477,
+      "step": 18703
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1722811309034485e-06,
+      "loss": 0.9409,
+      "step": 18704
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.171427102149926e-06,
+      "loss": 1.017,
+      "step": 18705
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.170573220862016e-06,
+      "loss": 1.0071,
+      "step": 18706
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1697194870558035e-06,
+      "loss": 1.0257,
+      "step": 18707
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.16886590074737e-06,
+      "loss": 0.9486,
+      "step": 18708
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.168012461952793e-06,
+      "loss": 1.0122,
+      "step": 18709
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.167159170688152e-06,
+      "loss": 0.931,
+      "step": 18710
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.166306026969513e-06,
+      "loss": 0.9961,
+      "step": 18711
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.165453030812955e-06,
+      "loss": 1.0055,
+      "step": 18712
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.164600182234542e-06,
+      "loss": 1.0095,
+      "step": 18713
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.163747481250342e-06,
+      "loss": 0.9446,
+      "step": 18714
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.162894927876408e-06,
+      "loss": 0.9777,
+      "step": 18715
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.16204252212881e-06,
+      "loss": 1.0432,
+      "step": 18716
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1611902640235995e-06,
+      "loss": 0.9962,
+      "step": 18717
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.160338153576834e-06,
+      "loss": 0.9882,
+      "step": 18718
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1594861908045575e-06,
+      "loss": 0.9459,
+      "step": 18719
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.158634375722825e-06,
+      "loss": 1.0371,
+      "step": 18720
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.157782708347682e-06,
+      "loss": 0.9722,
+      "step": 18721
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.156931188695166e-06,
+      "loss": 1.0158,
+      "step": 18722
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.156079816781318e-06,
+      "loss": 0.8903,
+      "step": 18723
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.155228592622179e-06,
+      "loss": 0.9797,
+      "step": 18724
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.154377516233784e-06,
+      "loss": 0.9516,
+      "step": 18725
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.153526587632158e-06,
+      "loss": 1.0069,
+      "step": 18726
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.152675806833335e-06,
+      "loss": 1.0082,
+      "step": 18727
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.151825173853341e-06,
+      "loss": 1.0449,
+      "step": 18728
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1509746887082e-06,
+      "loss": 0.9561,
+      "step": 18729
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.150124351413926e-06,
+      "loss": 1.0218,
+      "step": 18730
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1492741619865435e-06,
+      "loss": 1.0184,
+      "step": 18731
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1484241204420675e-06,
+      "loss": 0.9262,
+      "step": 18732
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.147574226796505e-06,
+      "loss": 0.977,
+      "step": 18733
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.146724481065865e-06,
+      "loss": 0.9932,
+      "step": 18734
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.14587488326616e-06,
+      "loss": 0.9965,
+      "step": 18735
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.145025433413395e-06,
+      "loss": 1.0134,
+      "step": 18736
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.144176131523562e-06,
+      "loss": 0.9997,
+      "step": 18737
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.143326977612662e-06,
+      "loss": 1.0202,
+      "step": 18738
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.142477971696695e-06,
+      "loss": 0.9921,
+      "step": 18739
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1416291137916534e-06,
+      "loss": 0.9678,
+      "step": 18740
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.140780403913522e-06,
+      "loss": 0.9812,
+      "step": 18741
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.139931842078288e-06,
+      "loss": 0.9463,
+      "step": 18742
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.139083428301942e-06,
+      "loss": 1.0356,
+      "step": 18743
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.138235162600459e-06,
+      "loss": 0.9683,
+      "step": 18744
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.137387044989821e-06,
+      "loss": 1.0126,
+      "step": 18745
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.136539075486001e-06,
+      "loss": 0.998,
+      "step": 18746
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.135691254104977e-06,
+      "loss": 1.0312,
+      "step": 18747
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.134843580862716e-06,
+      "loss": 1.0241,
+      "step": 18748
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.133996055775184e-06,
+      "loss": 1.021,
+      "step": 18749
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.133148678858344e-06,
+      "loss": 1.0247,
+      "step": 18750
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.132301450128168e-06,
+      "loss": 0.9723,
+      "step": 18751
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.131454369600606e-06,
+      "loss": 1.0319,
+      "step": 18752
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.130607437291616e-06,
+      "loss": 1.0493,
+      "step": 18753
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.129760653217153e-06,
+      "loss": 0.9817,
+      "step": 18754
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.128914017393166e-06,
+      "loss": 0.9892,
+      "step": 18755
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.128067529835605e-06,
+      "loss": 0.9475,
+      "step": 18756
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1272211905604115e-06,
+      "loss": 1.0527,
+      "step": 18757
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.126374999583536e-06,
+      "loss": 1.0512,
+      "step": 18758
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.12552895692091e-06,
+      "loss": 1.0756,
+      "step": 18759
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1246830625884705e-06,
+      "loss": 1.0734,
+      "step": 18760
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.123837316602153e-06,
+      "loss": 1.0481,
+      "step": 18761
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.122991718977896e-06,
+      "loss": 0.9329,
+      "step": 18762
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.122146269731616e-06,
+      "loss": 0.9613,
+      "step": 18763
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.121300968879246e-06,
+      "loss": 0.9683,
+      "step": 18764
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.120455816436707e-06,
+      "loss": 0.9243,
+      "step": 18765
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.119610812419918e-06,
+      "loss": 1.0174,
+      "step": 18766
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.118765956844797e-06,
+      "loss": 0.921,
+      "step": 18767
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.1179212497272586e-06,
+      "loss": 1.0141,
+      "step": 18768
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.1170766910832136e-06,
+      "loss": 0.9767,
+      "step": 18769
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.1162322809285724e-06,
+      "loss": 0.9428,
+      "step": 18770
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.115388019279239e-06,
+      "loss": 0.9398,
+      "step": 18771
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.114543906151119e-06,
+      "loss": 0.9724,
+      "step": 18772
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.113699941560112e-06,
+      "loss": 0.9639,
+      "step": 18773
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.112856125522114e-06,
+      "loss": 0.9676,
+      "step": 18774
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.112012458053021e-06,
+      "loss": 0.9588,
+      "step": 18775
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.111168939168725e-06,
+      "loss": 0.9513,
+      "step": 18776
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.110325568885116e-06,
+      "loss": 0.993,
+      "step": 18777
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.1094823472180795e-06,
+      "loss": 1.0678,
+      "step": 18778
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.108639274183499e-06,
+      "loss": 1.0988,
+      "step": 18779
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.1077963497972554e-06,
+      "loss": 0.9788,
+      "step": 18780
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.106953574075228e-06,
+      "loss": 0.9935,
+      "step": 18781
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.106110947033291e-06,
+      "loss": 0.9927,
+      "step": 18782
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.105268468687316e-06,
+      "loss": 0.9255,
+      "step": 18783
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.104426139053174e-06,
+      "loss": 0.9879,
+      "step": 18784
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.103583958146731e-06,
+      "loss": 0.9939,
+      "step": 18785
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.102741925983851e-06,
+      "loss": 1.013,
+      "step": 18786
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.1019000425803976e-06,
+      "loss": 0.9304,
+      "step": 18787
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.101058307952225e-06,
+      "loss": 0.9667,
+      "step": 18788
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.100216722115192e-06,
+      "loss": 0.9967,
+      "step": 18789
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.09937528508515e-06,
+      "loss": 0.9123,
+      "step": 18790
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.09853399687795e-06,
+      "loss": 1.022,
+      "step": 18791
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.097692857509439e-06,
+      "loss": 1.0285,
+      "step": 18792
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0968518669954604e-06,
+      "loss": 1.0221,
+      "step": 18793
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.09601102535186e-06,
+      "loss": 0.9759,
+      "step": 18794
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.095170332594467e-06,
+      "loss": 0.9186,
+      "step": 18795
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.094329788739127e-06,
+      "loss": 1.0328,
+      "step": 18796
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.093489393801669e-06,
+      "loss": 0.958,
+      "step": 18797
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.092649147797927e-06,
+      "loss": 0.9997,
+      "step": 18798
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.09180905074372e-06,
+      "loss": 1.0299,
+      "step": 18799
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.090969102654883e-06,
+      "loss": 0.9994,
+      "step": 18800
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.090129303547231e-06,
+      "loss": 1.0778,
+      "step": 18801
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.08928965343659e-06,
+      "loss": 0.9326,
+      "step": 18802
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.088450152338765e-06,
+      "loss": 0.9741,
+      "step": 18803
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.087610800269582e-06,
+      "loss": 0.9516,
+      "step": 18804
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0867715972448465e-06,
+      "loss": 1.0399,
+      "step": 18805
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.085932543280365e-06,
+      "loss": 0.9294,
+      "step": 18806
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.085093638391939e-06,
+      "loss": 1.0752,
+      "step": 18807
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.084254882595382e-06,
+      "loss": 0.9721,
+      "step": 18808
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.083416275906487e-06,
+      "loss": 1.0271,
+      "step": 18809
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.082577818341049e-06,
+      "loss": 0.9559,
+      "step": 18810
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.081739509914861e-06,
+      "loss": 0.979,
+      "step": 18811
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.080901350643719e-06,
+      "loss": 1.0399,
+      "step": 18812
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.080063340543412e-06,
+      "loss": 1.0401,
+      "step": 18813
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.079225479629718e-06,
+      "loss": 0.9465,
+      "step": 18814
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.078387767918426e-06,
+      "loss": 1.033,
+      "step": 18815
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.077550205425317e-06,
+      "loss": 0.9871,
+      "step": 18816
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.076712792166163e-06,
+      "loss": 0.9788,
+      "step": 18817
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.075875528156738e-06,
+      "loss": 0.8573,
+      "step": 18818
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.075038413412817e-06,
+      "loss": 0.9351,
+      "step": 18819
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.074201447950172e-06,
+      "loss": 0.9398,
+      "step": 18820
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.07336463178456e-06,
+      "loss": 0.9975,
+      "step": 18821
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.072527964931746e-06,
+      "loss": 0.9861,
+      "step": 18822
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.071691447407496e-06,
+      "loss": 0.9598,
+      "step": 18823
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.070855079227564e-06,
+      "loss": 1.0119,
+      "step": 18824
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0700188604077034e-06,
+      "loss": 0.9824,
+      "step": 18825
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.069182790963663e-06,
+      "loss": 0.9909,
+      "step": 18826
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.068346870911202e-06,
+      "loss": 0.9756,
+      "step": 18827
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.067511100266056e-06,
+      "loss": 1.0605,
+      "step": 18828
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.066675479043972e-06,
+      "loss": 1.0846,
+      "step": 18829
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.06584000726069e-06,
+      "loss": 0.9871,
+      "step": 18830
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.065004684931951e-06,
+      "loss": 1.0113,
+      "step": 18831
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.064169512073486e-06,
+      "loss": 0.9783,
+      "step": 18832
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0633344887010285e-06,
+      "loss": 0.998,
+      "step": 18833
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.062499614830304e-06,
+      "loss": 1.0484,
+      "step": 18834
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.061664890477049e-06,
+      "loss": 0.9771,
+      "step": 18835
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.060830315656976e-06,
+      "loss": 0.9769,
+      "step": 18836
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.059995890385811e-06,
+      "loss": 1.0688,
+      "step": 18837
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.05916161467927e-06,
+      "loss": 1.0185,
+      "step": 18838
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.05832748855307e-06,
+      "loss": 1.0242,
+      "step": 18839
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.057493512022922e-06,
+      "loss": 0.9476,
+      "step": 18840
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.056659685104537e-06,
+      "loss": 0.9801,
+      "step": 18841
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.05582600781362e-06,
+      "loss": 0.9839,
+      "step": 18842
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.054992480165876e-06,
+      "loss": 1.0695,
+      "step": 18843
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.054159102177005e-06,
+      "loss": 0.9938,
+      "step": 18844
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.053325873862703e-06,
+      "loss": 1.0312,
+      "step": 18845
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0524927952386735e-06,
+      "loss": 1.1191,
+      "step": 18846
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.051659866320601e-06,
+      "loss": 0.8907,
+      "step": 18847
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0508270871241775e-06,
+      "loss": 0.9562,
+      "step": 18848
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.049994457665091e-06,
+      "loss": 1.059,
+      "step": 18849
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.049161977959025e-06,
+      "loss": 0.9966,
+      "step": 18850
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.048329648021661e-06,
+      "loss": 0.9868,
+      "step": 18851
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.047497467868676e-06,
+      "loss": 0.8678,
+      "step": 18852
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.046665437515748e-06,
+      "loss": 0.9577,
+      "step": 18853
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.045833556978549e-06,
+      "loss": 1.0537,
+      "step": 18854
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.045001826272748e-06,
+      "loss": 0.9869,
+      "step": 18855
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.044170245414013e-06,
+      "loss": 1.009,
+      "step": 18856
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.043338814418009e-06,
+      "loss": 0.99,
+      "step": 18857
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.042507533300395e-06,
+      "loss": 0.9793,
+      "step": 18858
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.041676402076833e-06,
+      "loss": 1.0105,
+      "step": 18859
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.040845420762978e-06,
+      "loss": 0.9507,
+      "step": 18860
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0400145893744815e-06,
+      "loss": 0.9755,
+      "step": 18861
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.039183907926995e-06,
+      "loss": 0.9556,
+      "step": 18862
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.038353376436165e-06,
+      "loss": 0.9325,
+      "step": 18863
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.037522994917637e-06,
+      "loss": 0.9301,
+      "step": 18864
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0366927633870525e-06,
+      "loss": 0.9786,
+      "step": 18865
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.035862681860051e-06,
+      "loss": 0.9765,
+      "step": 18866
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.03503275035227e-06,
+      "loss": 1.111,
+      "step": 18867
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.034202968879336e-06,
+      "loss": 0.9991,
+      "step": 18868
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.033373337456887e-06,
+      "loss": 0.9695,
+      "step": 18869
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0325438561005485e-06,
+      "loss": 1.031,
+      "step": 18870
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.031714524825944e-06,
+      "loss": 1.009,
+      "step": 18871
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.030885343648695e-06,
+      "loss": 0.9955,
+      "step": 18872
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0300563125844244e-06,
+      "loss": 1.0419,
+      "step": 18873
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.029227431648745e-06,
+      "loss": 0.9432,
+      "step": 18874
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.02839870085727e-06,
+      "loss": 0.9649,
+      "step": 18875
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.027570120225613e-06,
+      "loss": 1.0231,
+      "step": 18876
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.02674168976938e-06,
+      "loss": 1.005,
+      "step": 18877
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.025913409504179e-06,
+      "loss": 0.9883,
+      "step": 18878
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.025085279445603e-06,
+      "loss": 0.9688,
+      "step": 18879
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.024257299609262e-06,
+      "loss": 0.9496,
+      "step": 18880
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.023429470010748e-06,
+      "loss": 1.0833,
+      "step": 18881
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.022601790665656e-06,
+      "loss": 0.9676,
+      "step": 18882
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.021774261589573e-06,
+      "loss": 1.0652,
+      "step": 18883
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.020946882798092e-06,
+      "loss": 1.0389,
+      "step": 18884
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0201196543067955e-06,
+      "loss": 0.8752,
+      "step": 18885
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.01929257613127e-06,
+      "loss": 0.9265,
+      "step": 18886
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.018465648287086e-06,
+      "loss": 1.0131,
+      "step": 18887
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.017638870789827e-06,
+      "loss": 0.9736,
+      "step": 18888
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.01681224365507e-06,
+      "loss": 0.9886,
+      "step": 18889
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.015985766898378e-06,
+      "loss": 0.9149,
+      "step": 18890
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.015159440535322e-06,
+      "loss": 0.9969,
+      "step": 18891
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.014333264581468e-06,
+      "loss": 1.0206,
+      "step": 18892
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.013507239052383e-06,
+      "loss": 1.0305,
+      "step": 18893
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.012681363963621e-06,
+      "loss": 0.9597,
+      "step": 18894
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.011855639330736e-06,
+      "loss": 1.0497,
+      "step": 18895
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.01103006516929e-06,
+      "loss": 0.9548,
+      "step": 18896
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0102046414948345e-06,
+      "loss": 0.9559,
+      "step": 18897
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.00937936832291e-06,
+      "loss": 0.9956,
+      "step": 18898
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.008554245669064e-06,
+      "loss": 1.0188,
+      "step": 18899
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.007729273548846e-06,
+      "loss": 1.0311,
+      "step": 18900
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0069044519777884e-06,
+      "loss": 0.9589,
+      "step": 18901
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0060797809714295e-06,
+      "loss": 0.9269,
+      "step": 18902
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.005255260545307e-06,
+      "loss": 0.9199,
+      "step": 18903
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.004430890714954e-06,
+      "loss": 1.0099,
+      "step": 18904
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.003606671495891e-06,
+      "loss": 0.9973,
+      "step": 18905
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.002782602903646e-06,
+      "loss": 1.0175,
+      "step": 18906
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.001958684953747e-06,
+      "loss": 0.988,
+      "step": 18907
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.001134917661713e-06,
+      "loss": 0.9825,
+      "step": 18908
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.000311301043056e-06,
+      "loss": 0.951,
+      "step": 18909
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.999487835113292e-06,
+      "loss": 0.9977,
+      "step": 18910
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.998664519887938e-06,
+      "loss": 1.0032,
+      "step": 18911
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.997841355382497e-06,
+      "loss": 0.9576,
+      "step": 18912
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9970183416124755e-06,
+      "loss": 1.0731,
+      "step": 18913
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.996195478593375e-06,
+      "loss": 1.0698,
+      "step": 18914
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.995372766340704e-06,
+      "loss": 1.0267,
+      "step": 18915
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.99455020486995e-06,
+      "loss": 0.9885,
+      "step": 18916
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.993727794196612e-06,
+      "loss": 0.9426,
+      "step": 18917
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.992905534336178e-06,
+      "loss": 0.9842,
+      "step": 18918
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.992083425304146e-06,
+      "loss": 0.988,
+      "step": 18919
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.991261467115991e-06,
+      "loss": 0.9979,
+      "step": 18920
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.990439659787202e-06,
+      "loss": 1.0132,
+      "step": 18921
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.989618003333258e-06,
+      "loss": 1.0471,
+      "step": 18922
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9887964977696355e-06,
+      "loss": 0.9665,
+      "step": 18923
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.987975143111811e-06,
+      "loss": 0.9378,
+      "step": 18924
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.987153939375254e-06,
+      "loss": 1.1181,
+      "step": 18925
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.986332886575435e-06,
+      "loss": 1.0437,
+      "step": 18926
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.98551198472782e-06,
+      "loss": 0.9564,
+      "step": 18927
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9846912338478715e-06,
+      "loss": 0.9847,
+      "step": 18928
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.983870633951051e-06,
+      "loss": 1.0496,
+      "step": 18929
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.983050185052815e-06,
+      "loss": 0.9732,
+      "step": 18930
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9822298871686174e-06,
+      "loss": 0.9946,
+      "step": 18931
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.981409740313912e-06,
+      "loss": 0.9587,
+      "step": 18932
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.980589744504145e-06,
+      "loss": 1.0176,
+      "step": 18933
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.979769899754766e-06,
+      "loss": 0.9487,
+      "step": 18934
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.978950206081217e-06,
+      "loss": 1.008,
+      "step": 18935
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.978130663498937e-06,
+      "loss": 0.9715,
+      "step": 18936
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.977311272023364e-06,
+      "loss": 1.0189,
+      "step": 18937
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.976492031669935e-06,
+      "loss": 1.0079,
+      "step": 18938
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.97567294245408e-06,
+      "loss": 1.0105,
+      "step": 18939
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.974854004391228e-06,
+      "loss": 0.9595,
+      "step": 18940
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.974035217496804e-06,
+      "loss": 1.0501,
+      "step": 18941
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.973216581786235e-06,
+      "loss": 0.97,
+      "step": 18942
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.972398097274939e-06,
+      "loss": 0.982,
+      "step": 18943
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.971579763978333e-06,
+      "loss": 0.9897,
+      "step": 18944
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.970761581911834e-06,
+      "loss": 1.0292,
+      "step": 18945
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.969943551090853e-06,
+      "loss": 0.9809,
+      "step": 18946
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.969125671530798e-06,
+      "loss": 0.9545,
+      "step": 18947
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.968307943247077e-06,
+      "loss": 1.0193,
+      "step": 18948
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.967490366255093e-06,
+      "loss": 0.9417,
+      "step": 18949
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.966672940570246e-06,
+      "loss": 0.9729,
+      "step": 18950
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.965855666207935e-06,
+      "loss": 0.9513,
+      "step": 18951
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.965038543183551e-06,
+      "loss": 1.0008,
+      "step": 18952
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.96422157151249e-06,
+      "loss": 0.916,
+      "step": 18953
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.96340475121014e-06,
+      "loss": 0.9905,
+      "step": 18954
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.962588082291891e-06,
+      "loss": 1.0454,
+      "step": 18955
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.961771564773117e-06,
+      "loss": 1.0248,
+      "step": 18956
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.960955198669206e-06,
+      "loss": 0.9362,
+      "step": 18957
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9601389839955365e-06,
+      "loss": 0.954,
+      "step": 18958
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.959322920767478e-06,
+      "loss": 0.9158,
+      "step": 18959
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.958507009000409e-06,
+      "loss": 0.9828,
+      "step": 18960
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.957691248709695e-06,
+      "loss": 1.028,
+      "step": 18961
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.956875639910704e-06,
+      "loss": 0.9512,
+      "step": 18962
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.956060182618792e-06,
+      "loss": 1.0077,
+      "step": 18963
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.95524487684933e-06,
+      "loss": 1.0175,
+      "step": 18964
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9544297226176715e-06,
+      "loss": 1.0641,
+      "step": 18965
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.953614719939173e-06,
+      "loss": 0.8636,
+      "step": 18966
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.952799868829181e-06,
+      "loss": 0.961,
+      "step": 18967
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.951985169303051e-06,
+      "loss": 0.9923,
+      "step": 18968
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.951170621376128e-06,
+      "loss": 1.0766,
+      "step": 18969
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.950356225063757e-06,
+      "loss": 0.9763,
+      "step": 18970
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.949541980381271e-06,
+      "loss": 0.9672,
+      "step": 18971
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9487278873440165e-06,
+      "loss": 0.9384,
+      "step": 18972
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.947913945967327e-06,
+      "loss": 0.9747,
+      "step": 18973
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.94710015626653e-06,
+      "loss": 0.9842,
+      "step": 18974
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.946286518256956e-06,
+      "loss": 0.9693,
+      "step": 18975
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.945473031953934e-06,
+      "loss": 0.8789,
+      "step": 18976
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9446596973727906e-06,
+      "loss": 0.9997,
+      "step": 18977
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.943846514528838e-06,
+      "loss": 0.9267,
+      "step": 18978
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.943033483437397e-06,
+      "loss": 1.0011,
+      "step": 18979
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.942220604113785e-06,
+      "loss": 0.9624,
+      "step": 18980
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.941407876573316e-06,
+      "loss": 0.942,
+      "step": 18981
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.940595300831294e-06,
+      "loss": 1.1036,
+      "step": 18982
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.939782876903024e-06,
+      "loss": 0.9999,
+      "step": 18983
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.938970604803819e-06,
+      "loss": 0.9084,
+      "step": 18984
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.938158484548969e-06,
+      "loss": 0.9665,
+      "step": 18985
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.937346516153778e-06,
+      "loss": 0.9981,
+      "step": 18986
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9365346996335376e-06,
+      "loss": 1.0787,
+      "step": 18987
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.935723035003545e-06,
+      "loss": 0.9755,
+      "step": 18988
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.934911522279084e-06,
+      "loss": 1.0039,
+      "step": 18989
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.934100161475441e-06,
+      "loss": 1.0425,
+      "step": 18990
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.933288952607902e-06,
+      "loss": 1.058,
+      "step": 18991
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9324778956917515e-06,
+      "loss": 1.0055,
+      "step": 18992
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.931666990742259e-06,
+      "loss": 0.9838,
+      "step": 18993
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.930856237774701e-06,
+      "loss": 1.0462,
+      "step": 18994
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.930045636804357e-06,
+      "loss": 0.9436,
+      "step": 18995
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.929235187846488e-06,
+      "loss": 1.0098,
+      "step": 18996
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.928424890916362e-06,
+      "loss": 1.0373,
+      "step": 18997
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.927614746029243e-06,
+      "loss": 1.0682,
+      "step": 18998
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.926804753200396e-06,
+      "loss": 0.9361,
+      "step": 18999
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.925994912445073e-06,
+      "loss": 1.0265,
+      "step": 19000
+    },
+    {
+      "epoch": 0.81,
+      "eval_text_loss": 0.7590646147727966,
+      "eval_text_runtime": 1.4196,
+      "eval_text_samples_per_second": 52.129,
+      "eval_text_steps_per_second": 1.409,
+      "step": 19000
+    },
+    {
+      "epoch": 0.81,
+      "eval_svit_loss": 1.0802959203720093,
+      "eval_svit_runtime": 5.0349,
+      "eval_svit_samples_per_second": 135.255,
+      "eval_svit_steps_per_second": 2.185,
+      "step": 19000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.92518522377853e-06,
+      "loss": 1.0284,
+      "step": 19001
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.9243756872160185e-06,
+      "loss": 1.0308,
+      "step": 19002
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.923566302772792e-06,
+      "loss": 0.9495,
+      "step": 19003
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.922757070464091e-06,
+      "loss": 1.0086,
+      "step": 19004
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.9219479903051626e-06,
+      "loss": 0.9935,
+      "step": 19005
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.921139062311246e-06,
+      "loss": 1.0257,
+      "step": 19006
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.920330286497579e-06,
+      "loss": 0.966,
+      "step": 19007
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.919521662879395e-06,
+      "loss": 1.0334,
+      "step": 19008
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.918713191471929e-06,
+      "loss": 0.9862,
+      "step": 19009
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.917904872290407e-06,
+      "loss": 0.9797,
+      "step": 19010
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.917096705350057e-06,
+      "loss": 1.0432,
+      "step": 19011
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.916288690666102e-06,
+      "loss": 0.9711,
+      "step": 19012
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.915480828253763e-06,
+      "loss": 0.9452,
+      "step": 19013
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.914673118128255e-06,
+      "loss": 0.9301,
+      "step": 19014
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.913865560304796e-06,
+      "loss": 1.0317,
+      "step": 19015
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.9130581547985965e-06,
+      "loss": 0.9665,
+      "step": 19016
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.912250901624864e-06,
+      "loss": 1.0741,
+      "step": 19017
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.911443800798808e-06,
+      "loss": 0.9922,
+      "step": 19018
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.910636852335627e-06,
+      "loss": 0.9884,
+      "step": 19019
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.909830056250526e-06,
+      "loss": 0.9163,
+      "step": 19020
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.9090234125587e-06,
+      "loss": 1.0002,
+      "step": 19021
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.908216921275346e-06,
+      "loss": 1.1089,
+      "step": 19022
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.907410582415651e-06,
+      "loss": 0.992,
+      "step": 19023
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.906604395994809e-06,
+      "loss": 0.9205,
+      "step": 19024
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.905798362028002e-06,
+      "loss": 1.0552,
+      "step": 19025
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.904992480530416e-06,
+      "loss": 0.9823,
+      "step": 19026
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.90418675151723e-06,
+      "loss": 0.9871,
+      "step": 19027
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.9033811750036225e-06,
+      "loss": 1.0618,
+      "step": 19028
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.902575751004766e-06,
+      "loss": 0.9575,
+      "step": 19029
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.901770479535834e-06,
+      "loss": 0.9921,
+      "step": 19030
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.900965360611993e-06,
+      "loss": 0.9951,
+      "step": 19031
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.900160394248412e-06,
+      "loss": 0.9758,
+      "step": 19032
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.899355580460253e-06,
+      "loss": 1.0113,
+      "step": 19033
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.898550919262674e-06,
+      "loss": 0.969,
+      "step": 19034
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.897746410670838e-06,
+      "loss": 0.9849,
+      "step": 19035
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.89694205469989e-06,
+      "loss": 0.9769,
+      "step": 19036
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.896137851364987e-06,
+      "loss": 1.0256,
+      "step": 19037
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.89533380068128e-06,
+      "loss": 0.9636,
+      "step": 19038
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.894529902663914e-06,
+      "loss": 1.0415,
+      "step": 19039
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.893726157328025e-06,
+      "loss": 0.9359,
+      "step": 19040
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.892922564688761e-06,
+      "loss": 0.9368,
+      "step": 19041
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.892119124761256e-06,
+      "loss": 1.0374,
+      "step": 19042
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.891315837560649e-06,
+      "loss": 0.972,
+      "step": 19043
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.89051270310206e-06,
+      "loss": 0.9548,
+      "step": 19044
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.889709721400627e-06,
+      "loss": 1.0141,
+      "step": 19045
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8889068924714765e-06,
+      "loss": 0.9519,
+      "step": 19046
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.888104216329724e-06,
+      "loss": 1.0087,
+      "step": 19047
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.887301692990494e-06,
+      "loss": 0.953,
+      "step": 19048
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.886499322468903e-06,
+      "loss": 0.9963,
+      "step": 19049
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.885697104780069e-06,
+      "loss": 0.9275,
+      "step": 19050
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.884895039939092e-06,
+      "loss": 1.0119,
+      "step": 19051
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.884093127961092e-06,
+      "loss": 0.9866,
+      "step": 19052
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.883291368861169e-06,
+      "loss": 1.0463,
+      "step": 19053
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.882489762654429e-06,
+      "loss": 1.0083,
+      "step": 19054
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.881688309355966e-06,
+      "loss": 1.0166,
+      "step": 19055
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.88088700898088e-06,
+      "loss": 0.9032,
+      "step": 19056
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.880085861544271e-06,
+      "loss": 1.0422,
+      "step": 19057
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.879284867061219e-06,
+      "loss": 1.0681,
+      "step": 19058
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.878484025546817e-06,
+      "loss": 1.0031,
+      "step": 19059
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.877683337016153e-06,
+      "loss": 0.9596,
+      "step": 19060
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.87688280148431e-06,
+      "loss": 0.9535,
+      "step": 19061
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.876082418966362e-06,
+      "loss": 1.0031,
+      "step": 19062
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.875282189477385e-06,
+      "loss": 0.9891,
+      "step": 19063
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8744821130324606e-06,
+      "loss": 1.0655,
+      "step": 19064
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8736821896466575e-06,
+      "loss": 1.0151,
+      "step": 19065
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8728824193350385e-06,
+      "loss": 0.9557,
+      "step": 19066
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8720828021126705e-06,
+      "loss": 0.9326,
+      "step": 19067
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.871283337994621e-06,
+      "loss": 0.9727,
+      "step": 19068
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.870484026995944e-06,
+      "loss": 1.0093,
+      "step": 19069
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.869684869131697e-06,
+      "loss": 1.0414,
+      "step": 19070
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.868885864416932e-06,
+      "loss": 0.9625,
+      "step": 19071
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.868087012866707e-06,
+      "loss": 0.9667,
+      "step": 19072
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.867288314496063e-06,
+      "loss": 0.9686,
+      "step": 19073
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.866489769320045e-06,
+      "loss": 0.9368,
+      "step": 19074
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.865691377353697e-06,
+      "loss": 0.9953,
+      "step": 19075
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8648931386120604e-06,
+      "loss": 1.0183,
+      "step": 19076
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.864095053110167e-06,
+      "loss": 0.9285,
+      "step": 19077
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.863297120863051e-06,
+      "loss": 0.979,
+      "step": 19078
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.86249934188575e-06,
+      "loss": 0.9803,
+      "step": 19079
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.861701716193282e-06,
+      "loss": 1.0192,
+      "step": 19080
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.860904243800675e-06,
+      "loss": 1.0399,
+      "step": 19081
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.86010692472295e-06,
+      "loss": 1.0142,
+      "step": 19082
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.859309758975132e-06,
+      "loss": 0.9535,
+      "step": 19083
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.85851274657223e-06,
+      "loss": 1.0439,
+      "step": 19084
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.85771588752926e-06,
+      "loss": 0.9591,
+      "step": 19085
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8569191818612295e-06,
+      "loss": 0.9457,
+      "step": 19086
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8561226295831534e-06,
+      "loss": 0.9763,
+      "step": 19087
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.855326230710029e-06,
+      "loss": 1.0007,
+      "step": 19088
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.854529985256861e-06,
+      "loss": 0.97,
+      "step": 19089
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.853733893238646e-06,
+      "loss": 1.0711,
+      "step": 19090
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.852937954670383e-06,
+      "loss": 1.0781,
+      "step": 19091
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.852142169567061e-06,
+      "loss": 0.9186,
+      "step": 19092
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.851346537943675e-06,
+      "loss": 1.0129,
+      "step": 19093
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.850551059815208e-06,
+      "loss": 1.0584,
+      "step": 19094
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8497557351966455e-06,
+      "loss": 1.0158,
+      "step": 19095
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.848960564102969e-06,
+      "loss": 1.0735,
+      "step": 19096
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.848165546549158e-06,
+      "loss": 1.1433,
+      "step": 19097
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.847370682550187e-06,
+      "loss": 0.9251,
+      "step": 19098
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.846575972121029e-06,
+      "loss": 1.0222,
+      "step": 19099
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.845781415276655e-06,
+      "loss": 1.0026,
+      "step": 19100
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.84498701203203e-06,
+      "loss": 1.0362,
+      "step": 19101
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.844192762402119e-06,
+      "loss": 0.9567,
+      "step": 19102
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.843398666401883e-06,
+      "loss": 0.9834,
+      "step": 19103
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.842604724046281e-06,
+      "loss": 0.9788,
+      "step": 19104
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8418109353502665e-06,
+      "loss": 1.0501,
+      "step": 19105
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8410173003287934e-06,
+      "loss": 1.0027,
+      "step": 19106
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.840223818996813e-06,
+      "loss": 1.1172,
+      "step": 19107
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.839430491369267e-06,
+      "loss": 1.038,
+      "step": 19108
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.838637317461105e-06,
+      "loss": 0.9811,
+      "step": 19109
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.837844297287264e-06,
+      "loss": 0.994,
+      "step": 19110
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.837051430862682e-06,
+      "loss": 1.042,
+      "step": 19111
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8362587182022964e-06,
+      "loss": 0.9535,
+      "step": 19112
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.835466159321039e-06,
+      "loss": 0.9974,
+      "step": 19113
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.834673754233836e-06,
+      "loss": 0.9224,
+      "step": 19114
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.833881502955619e-06,
+      "loss": 1.0537,
+      "step": 19115
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8330894055013065e-06,
+      "loss": 0.9246,
+      "step": 19116
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8322974618858225e-06,
+      "loss": 1.0176,
+      "step": 19117
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8315056721240835e-06,
+      "loss": 0.9174,
+      "step": 19118
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.830714036231006e-06,
+      "loss": 0.9641,
+      "step": 19119
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.829922554221496e-06,
+      "loss": 1.0208,
+      "step": 19120
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.829131226110468e-06,
+      "loss": 1.0146,
+      "step": 19121
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.828340051912828e-06,
+      "loss": 0.965,
+      "step": 19122
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.827549031643481e-06,
+      "loss": 0.968,
+      "step": 19123
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.82675816531732e-06,
+      "loss": 1.1,
+      "step": 19124
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.825967452949247e-06,
+      "loss": 1.0519,
+      "step": 19125
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.825176894554158e-06,
+      "loss": 0.9716,
+      "step": 19126
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.824386490146945e-06,
+      "loss": 0.9922,
+      "step": 19127
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.82359623974249e-06,
+      "loss": 1.0102,
+      "step": 19128
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.822806143355686e-06,
+      "loss": 0.9693,
+      "step": 19129
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.822016201001416e-06,
+      "loss": 1.0647,
+      "step": 19130
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.821226412694556e-06,
+      "loss": 0.9612,
+      "step": 19131
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.82043677844998e-06,
+      "loss": 0.9986,
+      "step": 19132
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.819647298282572e-06,
+      "loss": 1.016,
+      "step": 19133
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.818857972207198e-06,
+      "loss": 1.0136,
+      "step": 19134
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.818068800238724e-06,
+      "loss": 1.0315,
+      "step": 19135
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.81727978239202e-06,
+      "loss": 1.0378,
+      "step": 19136
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8164909186819455e-06,
+      "loss": 0.9805,
+      "step": 19137
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.815702209123365e-06,
+      "loss": 0.9718,
+      "step": 19138
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.814913653731127e-06,
+      "loss": 0.9943,
+      "step": 19139
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.814125252520091e-06,
+      "loss": 0.9744,
+      "step": 19140
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.81333700550511e-06,
+      "loss": 1.0063,
+      "step": 19141
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8125489127010272e-06,
+      "loss": 1.0261,
+      "step": 19142
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8117609741226882e-06,
+      "loss": 0.9991,
+      "step": 19143
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8109731897849383e-06,
+      "loss": 0.968,
+      "step": 19144
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8101855597026174e-06,
+      "loss": 0.9905,
+      "step": 19145
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.809398083890559e-06,
+      "loss": 0.9711,
+      "step": 19146
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.808610762363594e-06,
+      "loss": 1.0166,
+      "step": 19147
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.80782359513656e-06,
+      "loss": 1.0344,
+      "step": 19148
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.807036582224283e-06,
+      "loss": 1.0611,
+      "step": 19149
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.806249723641583e-06,
+      "loss": 1.1312,
+      "step": 19150
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8054630194032843e-06,
+      "loss": 1.0403,
+      "step": 19151
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.804676469524211e-06,
+      "loss": 1.0036,
+      "step": 19152
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8038900740191733e-06,
+      "loss": 1.0408,
+      "step": 19153
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.803103832902986e-06,
+      "loss": 1.0389,
+      "step": 19154
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8023177461904573e-06,
+      "loss": 1.0607,
+      "step": 19155
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.801531813896401e-06,
+      "loss": 0.9763,
+      "step": 19156
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8007460360356156e-06,
+      "loss": 1.0291,
+      "step": 19157
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7999604126229042e-06,
+      "loss": 1.04,
+      "step": 19158
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7991749436730632e-06,
+      "loss": 1.0327,
+      "step": 19159
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.798389629200896e-06,
+      "loss": 0.9535,
+      "step": 19160
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.797604469221189e-06,
+      "loss": 0.9993,
+      "step": 19161
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.796819463748732e-06,
+      "loss": 0.9571,
+      "step": 19162
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.796034612798314e-06,
+      "loss": 0.9895,
+      "step": 19163
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.795249916384719e-06,
+      "loss": 1.0133,
+      "step": 19164
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.794465374522728e-06,
+      "loss": 0.9316,
+      "step": 19165
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7936809872271166e-06,
+      "loss": 1.0144,
+      "step": 19166
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.792896754512667e-06,
+      "loss": 0.9077,
+      "step": 19167
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.792112676394145e-06,
+      "loss": 0.9555,
+      "step": 19168
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.791328752886323e-06,
+      "loss": 1.0231,
+      "step": 19169
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.790544984003964e-06,
+      "loss": 0.9797,
+      "step": 19170
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7897613697618422e-06,
+      "loss": 1.0012,
+      "step": 19171
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7889779101747062e-06,
+      "loss": 0.9806,
+      "step": 19172
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7881946052573204e-06,
+      "loss": 1.0397,
+      "step": 19173
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.787411455024438e-06,
+      "loss": 1.0268,
+      "step": 19174
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.786628459490811e-06,
+      "loss": 1.0511,
+      "step": 19175
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.785845618671188e-06,
+      "loss": 0.975,
+      "step": 19176
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.785062932580318e-06,
+      "loss": 1.0317,
+      "step": 19177
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7842804012329416e-06,
+      "loss": 0.9536,
+      "step": 19178
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7834980246438004e-06,
+      "loss": 0.9927,
+      "step": 19179
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7827158028276314e-06,
+      "loss": 0.96,
+      "step": 19180
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7819337357991695e-06,
+      "loss": 1.0256,
+      "step": 19181
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7811518235731463e-06,
+      "loss": 0.9935,
+      "step": 19182
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7803700661642895e-06,
+      "loss": 0.9953,
+      "step": 19183
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.779588463587327e-06,
+      "loss": 0.9859,
+      "step": 19184
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.778807015856981e-06,
+      "loss": 1.0152,
+      "step": 19185
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7780257229879716e-06,
+      "loss": 1.0299,
+      "step": 19186
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7772445849950147e-06,
+      "loss": 0.9473,
+      "step": 19187
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.776463601892825e-06,
+      "loss": 1.0169,
+      "step": 19188
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7756827736961147e-06,
+      "loss": 1.0339,
+      "step": 19189
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.774902100419591e-06,
+      "loss": 0.9665,
+      "step": 19190
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.77412158207796e-06,
+      "loss": 0.9974,
+      "step": 19191
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.773341218685924e-06,
+      "loss": 1.0117,
+      "step": 19192
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7725610102581836e-06,
+      "loss": 0.9853,
+      "step": 19193
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.771780956809433e-06,
+      "loss": 1.0366,
+      "step": 19194
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.771001058354369e-06,
+      "loss": 0.9839,
+      "step": 19195
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7702213149076796e-06,
+      "loss": 0.9955,
+      "step": 19196
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7694417264840546e-06,
+      "loss": 0.9955,
+      "step": 19197
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.768662293098178e-06,
+      "loss": 0.9734,
+      "step": 19198
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.767883014764732e-06,
+      "loss": 1.0264,
+      "step": 19199
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7671038914983964e-06,
+      "loss": 0.946,
+      "step": 19200
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7663249233138466e-06,
+      "loss": 0.9954,
+      "step": 19201
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7655461102257567e-06,
+      "loss": 0.9633,
+      "step": 19202
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.764767452248799e-06,
+      "loss": 0.9203,
+      "step": 19203
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7639889493976336e-06,
+      "loss": 1.0257,
+      "step": 19204
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7632106016869323e-06,
+      "loss": 0.9551,
+      "step": 19205
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.762432409131355e-06,
+      "loss": 1.0483,
+      "step": 19206
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7616543717455623e-06,
+      "loss": 1.0077,
+      "step": 19207
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.760876489544202e-06,
+      "loss": 1.0339,
+      "step": 19208
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7600987625419352e-06,
+      "loss": 1.0474,
+      "step": 19209
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7593211907534093e-06,
+      "loss": 1.003,
+      "step": 19210
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.758543774193273e-06,
+      "loss": 0.9617,
+      "step": 19211
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7577665128761644e-06,
+      "loss": 0.9729,
+      "step": 19212
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.75698940681673e-06,
+      "loss": 1.0346,
+      "step": 19213
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7562124560296114e-06,
+      "loss": 0.959,
+      "step": 19214
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7554356605294344e-06,
+      "loss": 1.0581,
+      "step": 19215
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7546590203308356e-06,
+      "loss": 1.0637,
+      "step": 19216
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7538825354484455e-06,
+      "loss": 0.9662,
+      "step": 19217
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.753106205896895e-06,
+      "loss": 0.9873,
+      "step": 19218
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7523300316907984e-06,
+      "loss": 0.9969,
+      "step": 19219
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7515540128447793e-06,
+      "loss": 0.9911,
+      "step": 19220
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7507781493734598e-06,
+      "loss": 0.9682,
+      "step": 19221
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7500024412914542e-06,
+      "loss": 0.9601,
+      "step": 19222
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7492268886133676e-06,
+      "loss": 0.9912,
+      "step": 19223
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.748451491353815e-06,
+      "loss": 0.9361,
+      "step": 19224
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7476762495274045e-06,
+      "loss": 0.9953,
+      "step": 19225
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.746901163148732e-06,
+      "loss": 1.03,
+      "step": 19226
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7461262322324003e-06,
+      "loss": 1.0464,
+      "step": 19227
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.745351456793008e-06,
+      "loss": 0.9219,
+      "step": 19228
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7445768368451533e-06,
+      "loss": 1.0129,
+      "step": 19229
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7438023724034217e-06,
+      "loss": 0.9257,
+      "step": 19230
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.743028063482399e-06,
+      "loss": 1.0357,
+      "step": 19231
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7422539100966783e-06,
+      "loss": 0.9893,
+      "step": 19232
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7414799122608417e-06,
+      "loss": 0.9752,
+      "step": 19233
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.740706069989464e-06,
+      "loss": 1.0628,
+      "step": 19234
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7399323832971203e-06,
+      "loss": 1.0508,
+      "step": 19235
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7391588521983945e-06,
+      "loss": 0.9675,
+      "step": 19236
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.738385476707849e-06,
+      "loss": 0.9993,
+      "step": 19237
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7376122568400534e-06,
+      "loss": 0.9537,
+      "step": 19238
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.736839192609571e-06,
+      "loss": 0.9562,
+      "step": 19239
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7360662840309707e-06,
+      "loss": 0.957,
+      "step": 19240
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.735293531118805e-06,
+      "loss": 1.041,
+      "step": 19241
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.734520933887632e-06,
+      "loss": 0.9908,
+      "step": 19242
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.733748492352003e-06,
+      "loss": 0.965,
+      "step": 19243
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7329762065264754e-06,
+      "loss": 0.9474,
+      "step": 19244
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7322040764255896e-06,
+      "loss": 0.9461,
+      "step": 19245
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7314321020638923e-06,
+      "loss": 1.0053,
+      "step": 19246
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7306602834559262e-06,
+      "loss": 1.0107,
+      "step": 19247
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7298886206162278e-06,
+      "loss": 0.9754,
+      "step": 19248
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.729117113559334e-06,
+      "loss": 1.0582,
+      "step": 19249
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.728345762299778e-06,
+      "loss": 0.9613,
+      "step": 19250
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.727574566852088e-06,
+      "loss": 1.0051,
+      "step": 19251
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7268035272307926e-06,
+      "loss": 1.0035,
+      "step": 19252
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7260326434504148e-06,
+      "loss": 0.9467,
+      "step": 19253
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7252619155254736e-06,
+      "loss": 0.9667,
+      "step": 19254
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7244913434704947e-06,
+      "loss": 0.96,
+      "step": 19255
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.723720927299984e-06,
+      "loss": 0.9695,
+      "step": 19256
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.722950667028459e-06,
+      "loss": 1.0229,
+      "step": 19257
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.722180562670428e-06,
+      "loss": 1.0153,
+      "step": 19258
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.721410614240395e-06,
+      "loss": 0.9972,
+      "step": 19259
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7206408217528668e-06,
+      "loss": 1.0077,
+      "step": 19260
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7198711852223425e-06,
+      "loss": 0.9755,
+      "step": 19261
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7191017046633195e-06,
+      "loss": 1.0535,
+      "step": 19262
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7183323800902918e-06,
+      "loss": 0.9306,
+      "step": 19263
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7175632115177517e-06,
+      "loss": 0.9334,
+      "step": 19264
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.716794198960189e-06,
+      "loss": 1.0172,
+      "step": 19265
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.716025342432087e-06,
+      "loss": 0.9486,
+      "step": 19266
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.715256641947931e-06,
+      "loss": 0.9497,
+      "step": 19267
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.714488097522199e-06,
+      "loss": 0.974,
+      "step": 19268
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7137197091693693e-06,
+      "loss": 1.0402,
+      "step": 19269
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7129514769039146e-06,
+      "loss": 0.9939,
+      "step": 19270
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.712183400740307e-06,
+      "loss": 0.9856,
+      "step": 19271
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7114154806930135e-06,
+      "loss": 1.0079,
+      "step": 19272
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.710647716776501e-06,
+      "loss": 1.0624,
+      "step": 19273
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7098801090052317e-06,
+      "loss": 1.0362,
+      "step": 19274
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.709112657393663e-06,
+      "loss": 0.9299,
+      "step": 19275
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7083453619562547e-06,
+      "loss": 0.9593,
+      "step": 19276
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7075782227074532e-06,
+      "loss": 0.9898,
+      "step": 19277
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.706811239661716e-06,
+      "loss": 1.0015,
+      "step": 19278
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7060444128334897e-06,
+      "loss": 0.9815,
+      "step": 19279
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7052777422372157e-06,
+      "loss": 0.9507,
+      "step": 19280
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7045112278873386e-06,
+      "loss": 1.0383,
+      "step": 19281
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.703744869798295e-06,
+      "loss": 1.0498,
+      "step": 19282
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7029786679845227e-06,
+      "loss": 1.0541,
+      "step": 19283
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7022126224604528e-06,
+      "loss": 0.947,
+      "step": 19284
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7014467332405157e-06,
+      "loss": 1.0617,
+      "step": 19285
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.700681000339138e-06,
+      "loss": 0.9736,
+      "step": 19286
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.699915423770748e-06,
+      "loss": 1.0071,
+      "step": 19287
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6991500035497575e-06,
+      "loss": 1.0107,
+      "step": 19288
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6983847396905917e-06,
+      "loss": 0.9721,
+      "step": 19289
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6976196322076654e-06,
+      "loss": 0.9802,
+      "step": 19290
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.696854681115391e-06,
+      "loss": 1.0048,
+      "step": 19291
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6960898864281718e-06,
+      "loss": 0.9871,
+      "step": 19292
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6953252481604197e-06,
+      "loss": 1.1058,
+      "step": 19293
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6945607663265383e-06,
+      "loss": 0.9953,
+      "step": 19294
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.693796440940929e-06,
+      "loss": 0.9991,
+      "step": 19295
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6930322720179815e-06,
+      "loss": 1.0599,
+      "step": 19296
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6922682595720983e-06,
+      "loss": 1.1129,
+      "step": 19297
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.691504403617672e-06,
+      "loss": 0.9361,
+      "step": 19298
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6907407041690834e-06,
+      "loss": 0.9728,
+      "step": 19299
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6899771612407224e-06,
+      "loss": 0.9897,
+      "step": 19300
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6892137748469728e-06,
+      "loss": 0.9617,
+      "step": 19301
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.688450545002216e-06,
+      "loss": 0.9893,
+      "step": 19302
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6876874717208236e-06,
+      "loss": 1.008,
+      "step": 19303
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.68692455501717e-06,
+      "loss": 1.0233,
+      "step": 19304
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6861617949056308e-06,
+      "loss": 0.9624,
+      "step": 19305
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.685399191400573e-06,
+      "loss": 0.9708,
+      "step": 19306
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.684636744516359e-06,
+      "loss": 0.9849,
+      "step": 19307
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6838744542673493e-06,
+      "loss": 1.0243,
+      "step": 19308
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.68311232066791e-06,
+      "loss": 0.9321,
+      "step": 19309
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.682350343732391e-06,
+      "loss": 0.9909,
+      "step": 19310
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.681588523475145e-06,
+      "loss": 0.9535,
+      "step": 19311
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.680826859910528e-06,
+      "loss": 0.8723,
+      "step": 19312
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.680065353052885e-06,
+      "loss": 0.9864,
+      "step": 19313
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6793040029165595e-06,
+      "loss": 0.9623,
+      "step": 19314
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6785428095158894e-06,
+      "loss": 0.9337,
+      "step": 19315
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.677781772865219e-06,
+      "loss": 0.9849,
+      "step": 19316
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6770208929788843e-06,
+      "loss": 0.9667,
+      "step": 19317
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6762601698712133e-06,
+      "loss": 0.9894,
+      "step": 19318
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.675499603556534e-06,
+      "loss": 0.9599,
+      "step": 19319
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.674739194049183e-06,
+      "loss": 1.0109,
+      "step": 19320
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.673978941363474e-06,
+      "loss": 0.9329,
+      "step": 19321
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.673218845513732e-06,
+      "loss": 1.0316,
+      "step": 19322
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6724589065142716e-06,
+      "loss": 0.9466,
+      "step": 19323
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6716991243794154e-06,
+      "loss": 0.9439,
+      "step": 19324
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.670939499123467e-06,
+      "loss": 1.0464,
+      "step": 19325
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6701800307607396e-06,
+      "loss": 1.0618,
+      "step": 19326
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6694207193055344e-06,
+      "loss": 0.9374,
+      "step": 19327
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6686615647721637e-06,
+      "loss": 1.0018,
+      "step": 19328
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6679025671749205e-06,
+      "loss": 0.9601,
+      "step": 19329
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6671437265281018e-06,
+      "loss": 0.983,
+      "step": 19330
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6663850428460046e-06,
+      "loss": 0.9441,
+      "step": 19331
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6656265161429183e-06,
+      "loss": 1.0464,
+      "step": 19332
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6648681464331317e-06,
+      "loss": 1.0231,
+      "step": 19333
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6641099337309307e-06,
+      "loss": 1.0866,
+      "step": 19334
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6633518780505963e-06,
+      "loss": 1.0675,
+      "step": 19335
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6625939794064095e-06,
+      "loss": 1.0581,
+      "step": 19336
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.661836237812645e-06,
+      "loss": 1.0015,
+      "step": 19337
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6610786532835773e-06,
+      "loss": 1.052,
+      "step": 19338
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6603212258334768e-06,
+      "loss": 0.9988,
+      "step": 19339
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6595639554766105e-06,
+      "loss": 0.9281,
+      "step": 19340
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.658806842227245e-06,
+      "loss": 1.0012,
+      "step": 19341
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6580498860996385e-06,
+      "loss": 0.9408,
+      "step": 19342
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6572930871080527e-06,
+      "loss": 0.9761,
+      "step": 19343
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.656536445266742e-06,
+      "loss": 0.9147,
+      "step": 19344
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6557799605899592e-06,
+      "loss": 0.8929,
+      "step": 19345
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.655023633091954e-06,
+      "loss": 0.9905,
+      "step": 19346
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.654267462786974e-06,
+      "loss": 1.0353,
+      "step": 19347
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6535114496892622e-06,
+      "loss": 1.0594,
+      "step": 19348
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6527555938130603e-06,
+      "loss": 1.0078,
+      "step": 19349
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6519998951726042e-06,
+      "loss": 0.895,
+      "step": 19350
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6512443537821317e-06,
+      "loss": 1.0644,
+      "step": 19351
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6504889696558734e-06,
+      "loss": 1.0137,
+      "step": 19352
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6497337428080573e-06,
+      "loss": 1.0842,
+      "step": 19353
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6489786732529123e-06,
+      "loss": 0.9379,
+      "step": 19354
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.648223761004658e-06,
+      "loss": 1.0493,
+      "step": 19355
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6474690060775177e-06,
+      "loss": 0.9768,
+      "step": 19356
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6467144084857077e-06,
+      "loss": 0.9888,
+      "step": 19357
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6459599682434415e-06,
+      "loss": 0.9568,
+      "step": 19358
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.64520568536493e-06,
+      "loss": 1.0239,
+      "step": 19359
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.644451559864385e-06,
+      "loss": 0.9986,
+      "step": 19360
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6436975917560046e-06,
+      "loss": 1.0192,
+      "step": 19361
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.642943781053998e-06,
+      "loss": 1.0086,
+      "step": 19362
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6421901277725626e-06,
+      "loss": 0.9749,
+      "step": 19363
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6414366319258968e-06,
+      "loss": 0.9168,
+      "step": 19364
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.640683293528187e-06,
+      "loss": 0.9743,
+      "step": 19365
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.639930112593631e-06,
+      "loss": 0.966,
+      "step": 19366
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.639177089136413e-06,
+      "loss": 0.9548,
+      "step": 19367
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6384242231707204e-06,
+      "loss": 0.9885,
+      "step": 19368
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6376715147107315e-06,
+      "loss": 0.9933,
+      "step": 19369
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6369189637706273e-06,
+      "loss": 1.0184,
+      "step": 19370
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6361665703645834e-06,
+      "loss": 1.0225,
+      "step": 19371
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6354143345067684e-06,
+      "loss": 0.964,
+      "step": 19372
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6346622562113567e-06,
+      "loss": 1.0157,
+      "step": 19373
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6339103354925145e-06,
+      "loss": 0.9936,
+      "step": 19374
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6331585723644074e-06,
+      "loss": 0.9663,
+      "step": 19375
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.632406966841188e-06,
+      "loss": 0.9546,
+      "step": 19376
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.631655518937022e-06,
+      "loss": 0.9399,
+      "step": 19377
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6309042286660626e-06,
+      "loss": 1.0175,
+      "step": 19378
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.630153096042464e-06,
+      "loss": 1.021,
+      "step": 19379
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.629402121080368e-06,
+      "loss": 1.0084,
+      "step": 19380
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6286513037939267e-06,
+      "loss": 1.0578,
+      "step": 19381
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6279006441972844e-06,
+      "loss": 0.9965,
+      "step": 19382
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6271501423045756e-06,
+      "loss": 1.0196,
+      "step": 19383
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6263997981299374e-06,
+      "loss": 0.9831,
+      "step": 19384
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.625649611687509e-06,
+      "loss": 0.9424,
+      "step": 19385
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6248995829914218e-06,
+      "loss": 0.9749,
+      "step": 19386
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.624149712055798e-06,
+      "loss": 1.0434,
+      "step": 19387
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6233999988947646e-06,
+      "loss": 1.0076,
+      "step": 19388
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6226504435224474e-06,
+      "loss": 0.9858,
+      "step": 19389
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6219010459529657e-06,
+      "loss": 1.0577,
+      "step": 19390
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6211518062004317e-06,
+      "loss": 1.092,
+      "step": 19391
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.620402724278958e-06,
+      "loss": 1.009,
+      "step": 19392
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6196538002026628e-06,
+      "loss": 1.0012,
+      "step": 19393
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.618905033985645e-06,
+      "loss": 0.9904,
+      "step": 19394
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6181564256420134e-06,
+      "loss": 1.0529,
+      "step": 19395
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.617407975185865e-06,
+      "loss": 1.0671,
+      "step": 19396
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.616659682631306e-06,
+      "loss": 0.9248,
+      "step": 19397
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6159115479924258e-06,
+      "loss": 0.9791,
+      "step": 19398
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.615163571283316e-06,
+      "loss": 0.9849,
+      "step": 19399
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6144157525180716e-06,
+      "loss": 0.9202,
+      "step": 19400
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.613668091710777e-06,
+      "loss": 0.9917,
+      "step": 19401
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.612920588875514e-06,
+      "loss": 1.0525,
+      "step": 19402
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6121732440263603e-06,
+      "loss": 0.9269,
+      "step": 19403
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6114260571774032e-06,
+      "loss": 0.9593,
+      "step": 19404
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6106790283427084e-06,
+      "loss": 0.9605,
+      "step": 19405
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.609932157536351e-06,
+      "loss": 0.9842,
+      "step": 19406
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6091854447723977e-06,
+      "loss": 0.9677,
+      "step": 19407
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6084388900649194e-06,
+      "loss": 0.9521,
+      "step": 19408
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6076924934279735e-06,
+      "loss": 1.0,
+      "step": 19409
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6069462548756207e-06,
+      "loss": 1.0602,
+      "step": 19410
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6062001744219167e-06,
+      "loss": 0.9512,
+      "step": 19411
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6054542520809222e-06,
+      "loss": 0.9102,
+      "step": 19412
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.604708487866682e-06,
+      "loss": 0.9846,
+      "step": 19413
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.603962881793243e-06,
+      "loss": 0.9724,
+      "step": 19414
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.603217433874653e-06,
+      "loss": 1.0182,
+      "step": 19415
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.602472144124952e-06,
+      "loss": 0.9699,
+      "step": 19416
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6017270125581806e-06,
+      "loss": 0.8857,
+      "step": 19417
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6009820391883747e-06,
+      "loss": 1.0084,
+      "step": 19418
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.600237224029565e-06,
+      "loss": 0.9569,
+      "step": 19419
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.599492567095784e-06,
+      "loss": 0.9886,
+      "step": 19420
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.598748068401057e-06,
+      "loss": 0.9923,
+      "step": 19421
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5980037279594094e-06,
+      "loss": 1.0005,
+      "step": 19422
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5972595457848615e-06,
+      "loss": 0.9611,
+      "step": 19423
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5965155218914307e-06,
+      "loss": 1.0059,
+      "step": 19424
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5957716562931336e-06,
+      "loss": 0.9116,
+      "step": 19425
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.59502794900398e-06,
+      "loss": 1.0589,
+      "step": 19426
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.594284400037982e-06,
+      "loss": 0.9966,
+      "step": 19427
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5935410094091427e-06,
+      "loss": 1.0075,
+      "step": 19428
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.592797777131467e-06,
+      "loss": 0.9635,
+      "step": 19429
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.592054703218956e-06,
+      "loss": 1.0109,
+      "step": 19430
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.591311787685604e-06,
+      "loss": 1.087,
+      "step": 19431
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5905690305454067e-06,
+      "loss": 1.0463,
+      "step": 19432
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.589826431812357e-06,
+      "loss": 0.9958,
+      "step": 19433
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.58908399150044e-06,
+      "loss": 0.9878,
+      "step": 19434
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.588341709623643e-06,
+      "loss": 1.005,
+      "step": 19435
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.587599586195947e-06,
+      "loss": 0.9719,
+      "step": 19436
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.586857621231332e-06,
+      "loss": 0.9876,
+      "step": 19437
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.586115814743775e-06,
+      "loss": 0.9217,
+      "step": 19438
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5853741667472488e-06,
+      "loss": 1.058,
+      "step": 19439
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.584632677255723e-06,
+      "loss": 1.0326,
+      "step": 19440
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.583891346283164e-06,
+      "loss": 1.0598,
+      "step": 19441
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5831501738435383e-06,
+      "loss": 0.9391,
+      "step": 19442
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5824091599508073e-06,
+      "loss": 0.9934,
+      "step": 19443
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5816683046189297e-06,
+      "loss": 1.0101,
+      "step": 19444
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.580927607861855e-06,
+      "loss": 0.9754,
+      "step": 19445
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.580187069693544e-06,
+      "loss": 0.9462,
+      "step": 19446
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5794466901279402e-06,
+      "loss": 1.0055,
+      "step": 19447
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5787064691789965e-06,
+      "loss": 1.0076,
+      "step": 19448
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.577966406860646e-06,
+      "loss": 1.0046,
+      "step": 19449
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.577226503186837e-06,
+      "loss": 1.0304,
+      "step": 19450
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5764867581715072e-06,
+      "loss": 1.0374,
+      "step": 19451
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.575747171828589e-06,
+      "loss": 0.996,
+      "step": 19452
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.57500774417201e-06,
+      "loss": 0.9771,
+      "step": 19453
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5742684752157053e-06,
+      "loss": 0.9732,
+      "step": 19454
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.573529364973599e-06,
+      "loss": 1.0054,
+      "step": 19455
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5727904134596083e-06,
+      "loss": 0.9725,
+      "step": 19456
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.572051620687659e-06,
+      "loss": 0.9356,
+      "step": 19457
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5713129866716644e-06,
+      "loss": 0.9655,
+      "step": 19458
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5705745114255423e-06,
+      "loss": 1.0151,
+      "step": 19459
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5698361949631945e-06,
+      "loss": 0.964,
+      "step": 19460
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5690980372985353e-06,
+      "loss": 0.964,
+      "step": 19461
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5683600384454687e-06,
+      "loss": 0.9016,
+      "step": 19462
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5676221984178987e-06,
+      "loss": 1.0874,
+      "step": 19463
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5668845172297142e-06,
+      "loss": 0.9663,
+      "step": 19464
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5661469948948205e-06,
+      "loss": 0.901,
+      "step": 19465
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5654096314271093e-06,
+      "loss": 1.0287,
+      "step": 19466
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5646724268404647e-06,
+      "loss": 0.9481,
+      "step": 19467
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5639353811487744e-06,
+      "loss": 1.0203,
+      "step": 19468
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.563198494365926e-06,
+      "loss": 0.8859,
+      "step": 19469
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5624617665058004e-06,
+      "loss": 0.9532,
+      "step": 19470
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.561725197582271e-06,
+      "loss": 0.9447,
+      "step": 19471
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.560988787609212e-06,
+      "loss": 0.9971,
+      "step": 19472
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5602525366005004e-06,
+      "loss": 1.023,
+      "step": 19473
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5595164445700027e-06,
+      "loss": 0.9583,
+      "step": 19474
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5587805115315825e-06,
+      "loss": 0.9613,
+      "step": 19475
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5580447374991002e-06,
+      "loss": 1.0032,
+      "step": 19476
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5573091224864244e-06,
+      "loss": 0.9961,
+      "step": 19477
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.556573666507404e-06,
+      "loss": 0.9737,
+      "step": 19478
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.555838369575895e-06,
+      "loss": 1.0017,
+      "step": 19479
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5551032317057458e-06,
+      "loss": 1.0826,
+      "step": 19480
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5543682529108105e-06,
+      "loss": 1.0041,
+      "step": 19481
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5536334332049275e-06,
+      "loss": 0.898,
+      "step": 19482
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5528987726019403e-06,
+      "loss": 1.0327,
+      "step": 19483
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.552164271115686e-06,
+      "loss": 1.0001,
+      "step": 19484
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.551429928760005e-06,
+      "loss": 0.9786,
+      "step": 19485
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.550695745548727e-06,
+      "loss": 1.0471,
+      "step": 19486
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5499617214956784e-06,
+      "loss": 0.9878,
+      "step": 19487
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5492278566146944e-06,
+      "loss": 0.9788,
+      "step": 19488
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5484941509195916e-06,
+      "loss": 1.0712,
+      "step": 19489
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5477606044241924e-06,
+      "loss": 1.0008,
+      "step": 19490
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5470272171423126e-06,
+      "loss": 0.9832,
+      "step": 19491
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5462939890877737e-06,
+      "loss": 0.9491,
+      "step": 19492
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5455609202743803e-06,
+      "loss": 1.0178,
+      "step": 19493
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.544828010715944e-06,
+      "loss": 1.0009,
+      "step": 19494
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.544095260426267e-06,
+      "loss": 0.9888,
+      "step": 19495
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5433626694191604e-06,
+      "loss": 0.9351,
+      "step": 19496
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.542630237708417e-06,
+      "loss": 1.0296,
+      "step": 19497
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5418979653078347e-06,
+      "loss": 1.0258,
+      "step": 19498
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5411658522312077e-06,
+      "loss": 1.0464,
+      "step": 19499
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.540433898492326e-06,
+      "loss": 1.0333,
+      "step": 19500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5397021041049785e-06,
+      "loss": 1.016,
+      "step": 19501
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.538970469082949e-06,
+      "loss": 0.9634,
+      "step": 19502
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.53823899344002e-06,
+      "loss": 0.9881,
+      "step": 19503
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.53750767718997e-06,
+      "loss": 1.0223,
+      "step": 19504
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5367765203465734e-06,
+      "loss": 0.9768,
+      "step": 19505
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.536045522923605e-06,
+      "loss": 1.0783,
+      "step": 19506
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5353146849348323e-06,
+      "loss": 1.0005,
+      "step": 19507
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.534584006394023e-06,
+      "loss": 0.978,
+      "step": 19508
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5338534873149426e-06,
+      "loss": 1.0043,
+      "step": 19509
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.533123127711349e-06,
+      "loss": 0.9852,
+      "step": 19510
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5323929275970004e-06,
+      "loss": 0.9624,
+      "step": 19511
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.531662886985652e-06,
+      "loss": 0.983,
+      "step": 19512
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5309330058910563e-06,
+      "loss": 0.9629,
+      "step": 19513
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.530203284326961e-06,
+      "loss": 0.9992,
+      "step": 19514
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.529473722307111e-06,
+      "loss": 0.9896,
+      "step": 19515
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.52874431984525e-06,
+      "loss": 1.029,
+      "step": 19516
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.528015076955118e-06,
+      "loss": 0.9829,
+      "step": 19517
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5272859936504514e-06,
+      "loss": 0.9982,
+      "step": 19518
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5265570699449824e-06,
+      "loss": 0.9748,
+      "step": 19519
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.525828305852443e-06,
+      "loss": 0.9962,
+      "step": 19520
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.525099701386561e-06,
+      "loss": 0.9872,
+      "step": 19521
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5243712565610602e-06,
+      "loss": 1.0072,
+      "step": 19522
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.523642971389663e-06,
+      "loss": 0.9341,
+      "step": 19523
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5229148458860864e-06,
+      "loss": 1.0599,
+      "step": 19524
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.522186880064048e-06,
+      "loss": 0.9594,
+      "step": 19525
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.521459073937259e-06,
+      "loss": 0.9476,
+      "step": 19526
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.520731427519431e-06,
+      "loss": 0.996,
+      "step": 19527
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5200039408242694e-06,
+      "loss": 0.9722,
+      "step": 19528
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5192766138654734e-06,
+      "loss": 1.0043,
+      "step": 19529
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5185494466567507e-06,
+      "loss": 1.0185,
+      "step": 19530
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5178224392117944e-06,
+      "loss": 0.9781,
+      "step": 19531
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.517095591544304e-06,
+      "loss": 1.0067,
+      "step": 19532
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.516368903667962e-06,
+      "loss": 0.9462,
+      "step": 19533
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5156423755964644e-06,
+      "loss": 0.944,
+      "step": 19534
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.514916007343494e-06,
+      "loss": 1.0612,
+      "step": 19535
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.514189798922737e-06,
+      "loss": 0.984,
+      "step": 19536
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.513463750347864e-06,
+      "loss": 0.9813,
+      "step": 19537
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5127378616325603e-06,
+      "loss": 0.9708,
+      "step": 19538
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.512012132790499e-06,
+      "loss": 0.9468,
+      "step": 19539
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5112865638353445e-06,
+      "loss": 1.0416,
+      "step": 19540
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5105611547807664e-06,
+      "loss": 0.9001,
+      "step": 19541
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.509835905640431e-06,
+      "loss": 1.0353,
+      "step": 19542
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5091108164280018e-06,
+      "loss": 0.968,
+      "step": 19543
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.50838588715713e-06,
+      "loss": 0.9782,
+      "step": 19544
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.507661117841477e-06,
+      "loss": 1.0214,
+      "step": 19545
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.506936508494694e-06,
+      "loss": 1.006,
+      "step": 19546
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.506212059130432e-06,
+      "loss": 0.9906,
+      "step": 19547
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5054877697623302e-06,
+      "loss": 0.9506,
+      "step": 19548
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.504763640404039e-06,
+      "loss": 0.9009,
+      "step": 19549
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5040396710691995e-06,
+      "loss": 1.0073,
+      "step": 19550
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5033158617714435e-06,
+      "loss": 1.0226,
+      "step": 19551
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.502592212524405e-06,
+      "loss": 1.004,
+      "step": 19552
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5018687233417215e-06,
+      "loss": 0.8938,
+      "step": 19553
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.501145394237019e-06,
+      "loss": 0.9735,
+      "step": 19554
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.50042222522392e-06,
+      "loss": 0.9604,
+      "step": 19555
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.499699216316046e-06,
+      "loss": 0.978,
+      "step": 19556
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4989763675270207e-06,
+      "loss": 1.0055,
+      "step": 19557
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.49825367887046e-06,
+      "loss": 0.9623,
+      "step": 19558
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4975311503599743e-06,
+      "loss": 0.9788,
+      "step": 19559
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4968087820091713e-06,
+      "loss": 1.0364,
+      "step": 19560
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4960865738316667e-06,
+      "loss": 0.9911,
+      "step": 19561
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4953645258410567e-06,
+      "loss": 0.9601,
+      "step": 19562
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.494642638050945e-06,
+      "loss": 1.0215,
+      "step": 19563
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4939209104749275e-06,
+      "loss": 1.0252,
+      "step": 19564
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4931993431266057e-06,
+      "loss": 0.9324,
+      "step": 19565
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.492477936019566e-06,
+      "loss": 0.9761,
+      "step": 19566
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4917566891673984e-06,
+      "loss": 1.0065,
+      "step": 19567
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.491035602583688e-06,
+      "loss": 0.9887,
+      "step": 19568
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.490314676282024e-06,
+      "loss": 0.9747,
+      "step": 19569
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4895939102759783e-06,
+      "loss": 1.0297,
+      "step": 19570
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4888733045791324e-06,
+      "loss": 0.9839,
+      "step": 19571
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4881528592050575e-06,
+      "loss": 1.0122,
+      "step": 19572
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4874325741673276e-06,
+      "loss": 1.0107,
+      "step": 19573
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.486712449479508e-06,
+      "loss": 1.0328,
+      "step": 19574
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.485992485155163e-06,
+      "loss": 0.984,
+      "step": 19575
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.485272681207861e-06,
+      "loss": 1.0105,
+      "step": 19576
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4845530376511526e-06,
+      "loss": 1.0535,
+      "step": 19577
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4838335544985983e-06,
+      "loss": 0.9913,
+      "step": 19578
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.483114231763746e-06,
+      "loss": 1.0343,
+      "step": 19579
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4823950694601537e-06,
+      "loss": 1.0082,
+      "step": 19580
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.481676067601362e-06,
+      "loss": 0.9915,
+      "step": 19581
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.480957226200916e-06,
+      "loss": 1.0448,
+      "step": 19582
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4802385452723555e-06,
+      "loss": 1.0517,
+      "step": 19583
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.479520024829221e-06,
+      "loss": 1.0125,
+      "step": 19584
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4788016648850433e-06,
+      "loss": 0.9591,
+      "step": 19585
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.478083465453357e-06,
+      "loss": 0.9332,
+      "step": 19586
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.47736542654769e-06,
+      "loss": 1.0238,
+      "step": 19587
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.476647548181568e-06,
+      "loss": 1.0159,
+      "step": 19588
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.475929830368512e-06,
+      "loss": 1.006,
+      "step": 19589
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.475212273122043e-06,
+      "loss": 0.9916,
+      "step": 19590
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4744948764556775e-06,
+      "loss": 0.9379,
+      "step": 19591
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4737776403829295e-06,
+      "loss": 0.9703,
+      "step": 19592
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.473060564917308e-06,
+      "loss": 0.9939,
+      "step": 19593
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4723436500723217e-06,
+      "loss": 0.9813,
+      "step": 19594
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.471626895861474e-06,
+      "loss": 0.9612,
+      "step": 19595
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.470910302298267e-06,
+      "loss": 0.9323,
+      "step": 19596
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.470193869396199e-06,
+      "loss": 1.0164,
+      "step": 19597
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.469477597168765e-06,
+      "loss": 0.9696,
+      "step": 19598
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4687614856294574e-06,
+      "loss": 0.9852,
+      "step": 19599
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4680455347917656e-06,
+      "loss": 1.0167,
+      "step": 19600
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4673297446691757e-06,
+      "loss": 1.1137,
+      "step": 19601
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.466614115275171e-06,
+      "loss": 0.9945,
+      "step": 19602
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.465898646623233e-06,
+      "loss": 0.9875,
+      "step": 19603
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4651833387268362e-06,
+      "loss": 0.8804,
+      "step": 19604
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.464468191599456e-06,
+      "loss": 0.9676,
+      "step": 19605
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.463753205254564e-06,
+      "loss": 0.9918,
+      "step": 19606
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.463038379705629e-06,
+      "loss": 0.9988,
+      "step": 19607
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.462323714966114e-06,
+      "loss": 1.012,
+      "step": 19608
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4616092110494825e-06,
+      "loss": 0.9911,
+      "step": 19609
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4608948679691923e-06,
+      "loss": 0.9586,
+      "step": 19610
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4601806857387013e-06,
+      "loss": 0.97,
+      "step": 19611
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4594666643714628e-06,
+      "loss": 0.9732,
+      "step": 19612
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.458752803880921e-06,
+      "loss": 1.0161,
+      "step": 19613
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4580391042805286e-06,
+      "loss": 0.9547,
+      "step": 19614
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.457325565583728e-06,
+      "loss": 0.9717,
+      "step": 19615
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.456612187803963e-06,
+      "loss": 1.0112,
+      "step": 19616
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4558989709546634e-06,
+      "loss": 1.001,
+      "step": 19617
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4551859150492696e-06,
+      "loss": 0.9327,
+      "step": 19618
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4544730201012133e-06,
+      "loss": 0.9509,
+      "step": 19619
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.453760286123925e-06,
+      "loss": 0.8953,
+      "step": 19620
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.453047713130823e-06,
+      "loss": 0.9842,
+      "step": 19621
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.452335301135337e-06,
+      "loss": 0.9907,
+      "step": 19622
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4516230501508867e-06,
+      "loss": 1.0828,
+      "step": 19623
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.450910960190884e-06,
+      "loss": 0.9582,
+      "step": 19624
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.450199031268742e-06,
+      "loss": 0.8778,
+      "step": 19625
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.449487263397876e-06,
+      "loss": 1.0042,
+      "step": 19626
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4487756565916946e-06,
+      "loss": 1.0344,
+      "step": 19627
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4480642108635955e-06,
+      "loss": 1.0153,
+      "step": 19628
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.447352926226981e-06,
+      "loss": 0.9776,
+      "step": 19629
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4466418026952556e-06,
+      "loss": 1.0475,
+      "step": 19630
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4459308402818116e-06,
+      "loss": 1.0126,
+      "step": 19631
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.445220039000037e-06,
+      "loss": 1.0828,
+      "step": 19632
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4445093988633275e-06,
+      "loss": 1.0288,
+      "step": 19633
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.443798919885069e-06,
+      "loss": 0.9865,
+      "step": 19634
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4430886020786395e-06,
+      "loss": 0.9609,
+      "step": 19635
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4423784454574185e-06,
+      "loss": 1.0003,
+      "step": 19636
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.44166845003479e-06,
+      "loss": 1.0074,
+      "step": 19637
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.440958615824127e-06,
+      "loss": 0.9543,
+      "step": 19638
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4402489428387955e-06,
+      "loss": 1.0297,
+      "step": 19639
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.439539431092164e-06,
+      "loss": 0.9541,
+      "step": 19640
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.438830080597602e-06,
+      "loss": 1.017,
+      "step": 19641
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4381208913684706e-06,
+      "loss": 0.9737,
+      "step": 19642
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.437411863418124e-06,
+      "loss": 1.0702,
+      "step": 19643
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.436702996759921e-06,
+      "loss": 1.0038,
+      "step": 19644
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.435994291407218e-06,
+      "loss": 1.0371,
+      "step": 19645
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4352857473733594e-06,
+      "loss": 1.0488,
+      "step": 19646
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.434577364671693e-06,
+      "loss": 0.9831,
+      "step": 19647
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.433869143315563e-06,
+      "loss": 0.995,
+      "step": 19648
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.433161083318313e-06,
+      "loss": 0.9447,
+      "step": 19649
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4324531846932763e-06,
+      "loss": 0.9945,
+      "step": 19650
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4317454474537893e-06,
+      "loss": 1.0174,
+      "step": 19651
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4310378716131812e-06,
+      "loss": 1.0495,
+      "step": 19652
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4303304571847877e-06,
+      "loss": 0.946,
+      "step": 19653
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.429623204181926e-06,
+      "loss": 0.9887,
+      "step": 19654
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4289161126179227e-06,
+      "loss": 0.9984,
+      "step": 19655
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4282091825060964e-06,
+      "loss": 1.0185,
+      "step": 19656
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4275024138597626e-06,
+      "loss": 0.9092,
+      "step": 19657
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.426795806692235e-06,
+      "loss": 0.9697,
+      "step": 19658
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4260893610168246e-06,
+      "loss": 0.9216,
+      "step": 19659
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4253830768468376e-06,
+      "loss": 0.9486,
+      "step": 19660
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4246769541955804e-06,
+      "loss": 0.9941,
+      "step": 19661
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4239709930763514e-06,
+      "loss": 0.962,
+      "step": 19662
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4232651935024474e-06,
+      "loss": 1.0733,
+      "step": 19663
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4225595554871706e-06,
+      "loss": 0.9459,
+      "step": 19664
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4218540790438053e-06,
+      "loss": 0.9563,
+      "step": 19665
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.421148764185644e-06,
+      "loss": 0.9499,
+      "step": 19666
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4204436109259712e-06,
+      "loss": 0.952,
+      "step": 19667
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4197386192780717e-06,
+      "loss": 0.9954,
+      "step": 19668
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.419033789255223e-06,
+      "loss": 0.9628,
+      "step": 19669
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.418329120870703e-06,
+      "loss": 0.9736,
+      "step": 19670
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.417624614137786e-06,
+      "loss": 1.0259,
+      "step": 19671
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4169202690697413e-06,
+      "loss": 0.9401,
+      "step": 19672
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4162160856798378e-06,
+      "loss": 1.0666,
+      "step": 19673
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4155120639813393e-06,
+      "loss": 1.0346,
+      "step": 19674
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4148082039875075e-06,
+      "loss": 0.9441,
+      "step": 19675
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.414104505711602e-06,
+      "loss": 0.9421,
+      "step": 19676
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.413400969166877e-06,
+      "loss": 0.9641,
+      "step": 19677
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4126975943665843e-06,
+      "loss": 0.956,
+      "step": 19678
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4119943813239752e-06,
+      "loss": 0.9709,
+      "step": 19679
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4112913300522945e-06,
+      "loss": 1.0372,
+      "step": 19680
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.410588440564786e-06,
+      "loss": 0.9327,
+      "step": 19681
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.409885712874691e-06,
+      "loss": 0.9645,
+      "step": 19682
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.409183146995244e-06,
+      "loss": 1.0467,
+      "step": 19683
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4084807429396817e-06,
+      "loss": 0.9767,
+      "step": 19684
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4077785007212343e-06,
+      "loss": 0.9681,
+      "step": 19685
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4070764203531284e-06,
+      "loss": 0.9133,
+      "step": 19686
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4063745018485905e-06,
+      "loss": 0.9653,
+      "step": 19687
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.405672745220844e-06,
+      "loss": 1.0734,
+      "step": 19688
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4049711504831063e-06,
+      "loss": 0.9712,
+      "step": 19689
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.40426971764859e-06,
+      "loss": 0.996,
+      "step": 19690
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.403568446730512e-06,
+      "loss": 0.9648,
+      "step": 19691
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.402867337742082e-06,
+      "loss": 1.0003,
+      "step": 19692
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4021663906965042e-06,
+      "loss": 1.0273,
+      "step": 19693
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4014656056069855e-06,
+      "loss": 0.9323,
+      "step": 19694
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.400764982486723e-06,
+      "loss": 0.962,
+      "step": 19695
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4000645213489193e-06,
+      "loss": 1.0503,
+      "step": 19696
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3993642222067604e-06,
+      "loss": 1.0593,
+      "step": 19697
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3986640850734443e-06,
+      "loss": 1.0199,
+      "step": 19698
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3979641099621588e-06,
+      "loss": 0.9307,
+      "step": 19699
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.39726429688609e-06,
+      "loss": 0.9514,
+      "step": 19700
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.396564645858414e-06,
+      "loss": 1.0449,
+      "step": 19701
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3958651568923164e-06,
+      "loss": 0.9385,
+      "step": 19702
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3951658300009716e-06,
+      "loss": 1.031,
+      "step": 19703
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.394466665197556e-06,
+      "loss": 1.0541,
+      "step": 19704
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.39376766249523e-06,
+      "loss": 0.9468,
+      "step": 19705
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.39306882190717e-06,
+      "loss": 0.959,
+      "step": 19706
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3923701434465385e-06,
+      "loss": 1.1193,
+      "step": 19707
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.391671627126493e-06,
+      "loss": 0.9233,
+      "step": 19708
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3909732729601896e-06,
+      "loss": 0.9894,
+      "step": 19709
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.390275080960789e-06,
+      "loss": 0.9398,
+      "step": 19710
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.389577051141443e-06,
+      "loss": 0.9701,
+      "step": 19711
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.388879183515296e-06,
+      "loss": 1.0255,
+      "step": 19712
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3881814780954915e-06,
+      "loss": 0.9479,
+      "step": 19713
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3874839348951797e-06,
+      "loss": 0.9588,
+      "step": 19714
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3867865539274983e-06,
+      "loss": 0.8982,
+      "step": 19715
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.386089335205578e-06,
+      "loss": 0.9916,
+      "step": 19716
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3853922787425556e-06,
+      "loss": 1.014,
+      "step": 19717
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.384695384551565e-06,
+      "loss": 0.9531,
+      "step": 19718
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3839986526457283e-06,
+      "loss": 0.9652,
+      "step": 19719
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3833020830381693e-06,
+      "loss": 0.9471,
+      "step": 19720
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3826056757420144e-06,
+      "loss": 0.9399,
+      "step": 19721
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3819094307703806e-06,
+      "loss": 0.9081,
+      "step": 19722
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3812133481363796e-06,
+      "loss": 0.9642,
+      "step": 19723
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.380517427853122e-06,
+      "loss": 0.9994,
+      "step": 19724
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.379821669933723e-06,
+      "loss": 1.0094,
+      "step": 19725
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3791260743912868e-06,
+      "loss": 0.9512,
+      "step": 19726
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3784306412389126e-06,
+      "loss": 0.8938,
+      "step": 19727
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3777353704897003e-06,
+      "loss": 0.9464,
+      "step": 19728
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3770402621567525e-06,
+      "loss": 0.9031,
+      "step": 19729
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.376345316253157e-06,
+      "loss": 0.9873,
+      "step": 19730
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3756505327920066e-06,
+      "loss": 1.0246,
+      "step": 19731
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3749559117863864e-06,
+      "loss": 0.9827,
+      "step": 19732
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3742614532493867e-06,
+      "loss": 0.9519,
+      "step": 19733
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3735671571940836e-06,
+      "loss": 1.0194,
+      "step": 19734
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3728730236335565e-06,
+      "loss": 0.9929,
+      "step": 19735
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3721790525808794e-06,
+      "loss": 0.9408,
+      "step": 19736
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.371485244049131e-06,
+      "loss": 0.9779,
+      "step": 19737
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3707915980513728e-06,
+      "loss": 1.0342,
+      "step": 19738
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.370098114600675e-06,
+      "loss": 1.0,
+      "step": 19739
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3694047937100984e-06,
+      "loss": 0.9149,
+      "step": 19740
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3687116353927052e-06,
+      "loss": 1.0113,
+      "step": 19741
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3680186396615515e-06,
+      "loss": 1.0516,
+      "step": 19742
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.36732580652969e-06,
+      "loss": 1.031,
+      "step": 19743
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3666331360101725e-06,
+      "loss": 1.0036,
+      "step": 19744
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3659406281160467e-06,
+      "loss": 1.0029,
+      "step": 19745
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3652482828603576e-06,
+      "loss": 0.9463,
+      "step": 19746
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3645561002561457e-06,
+      "loss": 0.895,
+      "step": 19747
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3638640803164518e-06,
+      "loss": 0.9776,
+      "step": 19748
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3631722230543083e-06,
+      "loss": 0.9646,
+      "step": 19749
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.362480528482749e-06,
+      "loss": 0.9564,
+      "step": 19750
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3617889966148045e-06,
+      "loss": 1.0871,
+      "step": 19751
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.361097627463499e-06,
+      "loss": 0.9847,
+      "step": 19752
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3604064210418573e-06,
+      "loss": 1.0268,
+      "step": 19753
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.359715377362899e-06,
+      "loss": 0.951,
+      "step": 19754
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.35902449643964e-06,
+      "loss": 0.9565,
+      "step": 19755
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.358333778285097e-06,
+      "loss": 0.977,
+      "step": 19756
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3576432229122783e-06,
+      "loss": 1.022,
+      "step": 19757
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3569528303341926e-06,
+      "loss": 1.0046,
+      "step": 19758
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3562626005638455e-06,
+      "loss": 1.0117,
+      "step": 19759
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3555725336142377e-06,
+      "loss": 1.0215,
+      "step": 19760
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.35488262949837e-06,
+      "loss": 1.015,
+      "step": 19761
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3541928882292343e-06,
+      "loss": 0.9084,
+      "step": 19762
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3535033098198256e-06,
+      "loss": 0.9814,
+      "step": 19763
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3528138942831336e-06,
+      "loss": 0.9067,
+      "step": 19764
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3521246416321445e-06,
+      "loss": 1.0058,
+      "step": 19765
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3514355518798416e-06,
+      "loss": 1.0404,
+      "step": 19766
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3507466250392032e-06,
+      "loss": 0.9625,
+      "step": 19767
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3500578611232092e-06,
+      "loss": 1.0087,
+      "step": 19768
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.349369260144833e-06,
+      "loss": 0.9393,
+      "step": 19769
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3486808221170455e-06,
+      "loss": 1.0076,
+      "step": 19770
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3479925470528143e-06,
+      "loss": 0.9795,
+      "step": 19771
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.347304434965104e-06,
+      "loss": 0.9749,
+      "step": 19772
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3466164858668796e-06,
+      "loss": 0.9262,
+      "step": 19773
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3459286997710927e-06,
+      "loss": 1.0303,
+      "step": 19774
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.345241076690707e-06,
+      "loss": 0.9939,
+      "step": 19775
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.344553616638671e-06,
+      "loss": 0.9708,
+      "step": 19776
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3438663196279375e-06,
+      "loss": 0.9332,
+      "step": 19777
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.343179185671447e-06,
+      "loss": 1.0657,
+      "step": 19778
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3424922147821474e-06,
+      "loss": 0.9392,
+      "step": 19779
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.341805406972981e-06,
+      "loss": 0.9218,
+      "step": 19780
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3411187622568777e-06,
+      "loss": 1.01,
+      "step": 19781
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3404322806467796e-06,
+      "loss": 0.9826,
+      "step": 19782
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3397459621556127e-06,
+      "loss": 1.0807,
+      "step": 19783
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.339059806796311e-06,
+      "loss": 0.9531,
+      "step": 19784
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3383738145817903e-06,
+      "loss": 1.0159,
+      "step": 19785
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.33768798552498e-06,
+      "loss": 1.0058,
+      "step": 19786
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.337002319638796e-06,
+      "loss": 1.0934,
+      "step": 19787
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3363168169361575e-06,
+      "loss": 0.9733,
+      "step": 19788
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3356314774299705e-06,
+      "loss": 0.9914,
+      "step": 19789
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3349463011331495e-06,
+      "loss": 0.9757,
+      "step": 19790
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3342612880586034e-06,
+      "loss": 1.0427,
+      "step": 19791
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3335764382192308e-06,
+      "loss": 1.1019,
+      "step": 19792
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3328917516279302e-06,
+      "loss": 1.024,
+      "step": 19793
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.332207228297605e-06,
+      "loss": 0.9553,
+      "step": 19794
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3315228682411496e-06,
+      "loss": 1.0277,
+      "step": 19795
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3308386714714494e-06,
+      "loss": 1.0077,
+      "step": 19796
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.330154638001394e-06,
+      "loss": 1.0279,
+      "step": 19797
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3294707678438722e-06,
+      "loss": 0.9663,
+      "step": 19798
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.328787061011766e-06,
+      "loss": 1.0007,
+      "step": 19799
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.32810351751795e-06,
+      "loss": 0.9699,
+      "step": 19800
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3274201373752993e-06,
+      "loss": 1.0094,
+      "step": 19801
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.326736920596695e-06,
+      "loss": 0.9879,
+      "step": 19802
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.326053867194997e-06,
+      "loss": 1.0733,
+      "step": 19803
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.325370977183078e-06,
+      "loss": 1.0672,
+      "step": 19804
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3246882505737967e-06,
+      "loss": 1.0045,
+      "step": 19805
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.324005687380021e-06,
+      "loss": 0.9846,
+      "step": 19806
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3233232876146006e-06,
+      "loss": 1.0681,
+      "step": 19807
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.322641051290391e-06,
+      "loss": 0.9682,
+      "step": 19808
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3219589784202467e-06,
+      "loss": 1.0298,
+      "step": 19809
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3212770690170173e-06,
+      "loss": 0.9845,
+      "step": 19810
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.320595323093542e-06,
+      "loss": 1.014,
+      "step": 19811
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.319913740662664e-06,
+      "loss": 1.0648,
+      "step": 19812
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.319232321737227e-06,
+      "loss": 0.9956,
+      "step": 19813
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3185510663300617e-06,
+      "loss": 1.0082,
+      "step": 19814
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3178699744540026e-06,
+      "loss": 0.9671,
+      "step": 19815
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.317189046121877e-06,
+      "loss": 0.9583,
+      "step": 19816
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.316508281346518e-06,
+      "loss": 0.9425,
+      "step": 19817
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.315827680140743e-06,
+      "loss": 1.0129,
+      "step": 19818
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3151472425173735e-06,
+      "loss": 0.9956,
+      "step": 19819
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3144669684892253e-06,
+      "loss": 0.9877,
+      "step": 19820
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3137868580691195e-06,
+      "loss": 1.0035,
+      "step": 19821
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3131069112698603e-06,
+      "loss": 0.9825,
+      "step": 19822
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.312427128104258e-06,
+      "loss": 1.01,
+      "step": 19823
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.311747508585117e-06,
+      "loss": 0.938,
+      "step": 19824
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3110680527252394e-06,
+      "loss": 1.023,
+      "step": 19825
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3103887605374255e-06,
+      "loss": 0.9712,
+      "step": 19826
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3097096320344684e-06,
+      "loss": 0.9932,
+      "step": 19827
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.309030667229163e-06,
+      "loss": 0.9854,
+      "step": 19828
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3083518661342975e-06,
+      "loss": 0.9606,
+      "step": 19829
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3076732287626604e-06,
+      "loss": 0.9801,
+      "step": 19830
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3069947551270325e-06,
+      "loss": 1.0449,
+      "step": 19831
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3063164452401942e-06,
+      "loss": 0.927,
+      "step": 19832
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.305638299114925e-06,
+      "loss": 1.005,
+      "step": 19833
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3049603167639968e-06,
+      "loss": 1.0027,
+      "step": 19834
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3042824982001824e-06,
+      "loss": 1.0404,
+      "step": 19835
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.303604843436248e-06,
+      "loss": 1.0449,
+      "step": 19836
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3029273524849605e-06,
+      "loss": 0.9669,
+      "step": 19837
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3022500253590797e-06,
+      "loss": 0.9893,
+      "step": 19838
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.301572862071365e-06,
+      "loss": 1.0477,
+      "step": 19839
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3008958626345734e-06,
+      "loss": 0.9816,
+      "step": 19840
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3002190270614555e-06,
+      "loss": 0.9199,
+      "step": 19841
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.299542355364762e-06,
+      "loss": 0.9428,
+      "step": 19842
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2988658475572377e-06,
+      "loss": 0.9889,
+      "step": 19843
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2981895036516277e-06,
+      "loss": 0.9341,
+      "step": 19844
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2975133236606716e-06,
+      "loss": 0.9632,
+      "step": 19845
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.296837307597106e-06,
+      "loss": 0.9684,
+      "step": 19846
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2961614554736654e-06,
+      "loss": 0.9795,
+      "step": 19847
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.295485767303081e-06,
+      "loss": 1.018,
+      "step": 19848
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.294810243098079e-06,
+      "loss": 0.9073,
+      "step": 19849
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.294134882871386e-06,
+      "loss": 0.9038,
+      "step": 19850
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.293459686635723e-06,
+      "loss": 0.9021,
+      "step": 19851
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.292784654403809e-06,
+      "loss": 0.9876,
+      "step": 19852
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.29210978618836e-06,
+      "loss": 0.9606,
+      "step": 19853
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2914350820020836e-06,
+      "loss": 0.9342,
+      "step": 19854
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.290760541857696e-06,
+      "loss": 0.9618,
+      "step": 19855
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2900861657678995e-06,
+      "loss": 0.9323,
+      "step": 19856
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2894119537454e-06,
+      "loss": 0.9791,
+      "step": 19857
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2887379058028916e-06,
+      "loss": 1.0169,
+      "step": 19858
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.288064021953077e-06,
+      "loss": 0.9102,
+      "step": 19859
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2873903022086486e-06,
+      "loss": 0.963,
+      "step": 19860
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2867167465822994e-06,
+      "loss": 0.9668,
+      "step": 19861
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2860433550867093e-06,
+      "loss": 0.9529,
+      "step": 19862
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2853701277345708e-06,
+      "loss": 0.9438,
+      "step": 19863
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2846970645385635e-06,
+      "loss": 1.0262,
+      "step": 19864
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2840241655113637e-06,
+      "loss": 1.0645,
+      "step": 19865
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2833514306656467e-06,
+      "loss": 1.0024,
+      "step": 19866
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.282678860014087e-06,
+      "loss": 1.0213,
+      "step": 19867
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.282006453569354e-06,
+      "loss": 0.9838,
+      "step": 19868
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2813342113441096e-06,
+      "loss": 1.0147,
+      "step": 19869
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.28066213335102e-06,
+      "loss": 1.0204,
+      "step": 19870
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.279990219602746e-06,
+      "loss": 0.9909,
+      "step": 19871
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2793184701119443e-06,
+      "loss": 0.9339,
+      "step": 19872
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.278646884891264e-06,
+      "loss": 0.9995,
+      "step": 19873
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.277975463953359e-06,
+      "loss": 0.9147,
+      "step": 19874
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2773042073108785e-06,
+      "loss": 0.9752,
+      "step": 19875
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2766331149764644e-06,
+      "loss": 0.9953,
+      "step": 19876
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.275962186962755e-06,
+      "loss": 1.0148,
+      "step": 19877
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.275291423282394e-06,
+      "loss": 1.016,
+      "step": 19878
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.274620823948016e-06,
+      "loss": 0.931,
+      "step": 19879
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.27395038897225e-06,
+      "loss": 0.9809,
+      "step": 19880
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2732801183677247e-06,
+      "loss": 0.9542,
+      "step": 19881
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2726100121470675e-06,
+      "loss": 1.0198,
+      "step": 19882
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2719400703229045e-06,
+      "loss": 1.0192,
+      "step": 19883
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.271270292907849e-06,
+      "loss": 1.0356,
+      "step": 19884
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2706006799145187e-06,
+      "loss": 0.9775,
+      "step": 19885
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.269931231355532e-06,
+      "loss": 0.9949,
+      "step": 19886
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2692619472434943e-06,
+      "loss": 1.0399,
+      "step": 19887
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.268592827591014e-06,
+      "loss": 0.9573,
+      "step": 19888
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.267923872410693e-06,
+      "loss": 0.9904,
+      "step": 19889
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.26725508171514e-06,
+      "loss": 0.9817,
+      "step": 19890
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2665864555169437e-06,
+      "loss": 0.987,
+      "step": 19891
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2659179938287035e-06,
+      "loss": 1.0152,
+      "step": 19892
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2652496966630086e-06,
+      "loss": 0.95,
+      "step": 19893
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2645815640324546e-06,
+      "loss": 0.9685,
+      "step": 19894
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2639135959496185e-06,
+      "loss": 0.987,
+      "step": 19895
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2632457924270834e-06,
+      "loss": 0.9936,
+      "step": 19896
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2625781534774375e-06,
+      "loss": 0.9801,
+      "step": 19897
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2619106791132474e-06,
+      "loss": 0.987,
+      "step": 19898
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.261243369347089e-06,
+      "loss": 0.9363,
+      "step": 19899
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.260576224191532e-06,
+      "loss": 0.9953,
+      "step": 19900
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.259909243659149e-06,
+      "loss": 1.0216,
+      "step": 19901
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2592424277624947e-06,
+      "loss": 1.0614,
+      "step": 19902
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.258575776514136e-06,
+      "loss": 1.0356,
+      "step": 19903
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.257909289926625e-06,
+      "loss": 1.0387,
+      "step": 19904
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.257242968012525e-06,
+      "loss": 1.0066,
+      "step": 19905
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.256576810784381e-06,
+      "loss": 1.0618,
+      "step": 19906
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.255910818254742e-06,
+      "loss": 0.9926,
+      "step": 19907
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.255244990436153e-06,
+      "loss": 0.9862,
+      "step": 19908
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2545793273411576e-06,
+      "loss": 0.9733,
+      "step": 19909
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.253913828982294e-06,
+      "loss": 0.9653,
+      "step": 19910
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.253248495372098e-06,
+      "loss": 0.9666,
+      "step": 19911
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2525833265231024e-06,
+      "loss": 1.0199,
+      "step": 19912
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.251918322447837e-06,
+      "loss": 0.9651,
+      "step": 19913
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2512534831588285e-06,
+      "loss": 0.9343,
+      "step": 19914
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2505888086686003e-06,
+      "loss": 0.9868,
+      "step": 19915
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.249924298989672e-06,
+      "loss": 1.0556,
+      "step": 19916
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.249259954134562e-06,
+      "loss": 0.9689,
+      "step": 19917
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.248595774115784e-06,
+      "loss": 0.9106,
+      "step": 19918
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2479317589458496e-06,
+      "loss": 0.9799,
+      "step": 19919
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2472679086372664e-06,
+      "loss": 0.9798,
+      "step": 19920
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2466042232025384e-06,
+      "loss": 1.0335,
+      "step": 19921
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.245940702654168e-06,
+      "loss": 0.965,
+      "step": 19922
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.245277347004654e-06,
+      "loss": 0.925,
+      "step": 19923
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2446141562664926e-06,
+      "loss": 0.9528,
+      "step": 19924
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2439511304521743e-06,
+      "loss": 0.9427,
+      "step": 19925
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.243288269574191e-06,
+      "loss": 1.0027,
+      "step": 19926
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2426255736450253e-06,
+      "loss": 0.842,
+      "step": 19927
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2419630426771643e-06,
+      "loss": 1.0015,
+      "step": 19928
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2413006766830856e-06,
+      "loss": 0.9505,
+      "step": 19929
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2406384756752664e-06,
+      "loss": 0.9401,
+      "step": 19930
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2399764396661805e-06,
+      "loss": 1.0438,
+      "step": 19931
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2393145686682996e-06,
+      "loss": 0.9263,
+      "step": 19932
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2386528626940896e-06,
+      "loss": 0.9679,
+      "step": 19933
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2379913217560154e-06,
+      "loss": 1.0378,
+      "step": 19934
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.23732994586654e-06,
+      "loss": 0.9562,
+      "step": 19935
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2366687350381203e-06,
+      "loss": 1.0398,
+      "step": 19936
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2360076892832125e-06,
+      "loss": 0.8929,
+      "step": 19937
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2353468086142638e-06,
+      "loss": 1.0015,
+      "step": 19938
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2346860930437297e-06,
+      "loss": 0.9234,
+      "step": 19939
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.234025542584053e-06,
+      "loss": 0.9912,
+      "step": 19940
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.233365157247678e-06,
+      "loss": 0.9476,
+      "step": 19941
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.232704937047039e-06,
+      "loss": 1.0467,
+      "step": 19942
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2320448819945787e-06,
+      "loss": 0.9824,
+      "step": 19943
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2313849921027276e-06,
+      "loss": 1.0286,
+      "step": 19944
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.23072526738392e-06,
+      "loss": 1.1023,
+      "step": 19945
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2300657078505727e-06,
+      "loss": 0.9877,
+      "step": 19946
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.22940631351512e-06,
+      "loss": 0.9536,
+      "step": 19947
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.228747084389982e-06,
+      "loss": 1.021,
+      "step": 19948
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.228088020487571e-06,
+      "loss": 0.9638,
+      "step": 19949
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2274291218203026e-06,
+      "loss": 0.9679,
+      "step": 19950
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2267703884005923e-06,
+      "loss": 1.0178,
+      "step": 19951
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2261118202408483e-06,
+      "loss": 0.9298,
+      "step": 19952
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2254534173534725e-06,
+      "loss": 1.044,
+      "step": 19953
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2247951797508663e-06,
+      "loss": 0.9562,
+      "step": 19954
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2241371074454332e-06,
+      "loss": 0.9905,
+      "step": 19955
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.22347920044957e-06,
+      "loss": 1.0274,
+      "step": 19956
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2228214587756624e-06,
+      "loss": 0.9867,
+      "step": 19957
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.222163882436107e-06,
+      "loss": 1.0167,
+      "step": 19958
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.22150647144329e-06,
+      "loss": 0.9819,
+      "step": 19959
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2208492258095914e-06,
+      "loss": 0.9616,
+      "step": 19960
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2201921455473907e-06,
+      "loss": 1.0292,
+      "step": 19961
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.219535230669071e-06,
+      "loss": 0.9477,
+      "step": 19962
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2188784811870045e-06,
+      "loss": 1.0009,
+      "step": 19963
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2182218971135592e-06,
+      "loss": 1.0131,
+      "step": 19964
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2175654784611025e-06,
+      "loss": 1.0046,
+      "step": 19965
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2169092252420045e-06,
+      "loss": 0.988,
+      "step": 19966
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.216253137468626e-06,
+      "loss": 0.878,
+      "step": 19967
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.215597215153322e-06,
+      "loss": 1.0443,
+      "step": 19968
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2149414583084472e-06,
+      "loss": 0.9249,
+      "step": 19969
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.214285866946362e-06,
+      "loss": 0.9442,
+      "step": 19970
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2136304410794066e-06,
+      "loss": 1.0422,
+      "step": 19971
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.212975180719932e-06,
+      "loss": 1.0525,
+      "step": 19972
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2123200858802773e-06,
+      "loss": 0.9974,
+      "step": 19973
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.21166515657279e-06,
+      "loss": 1.066,
+      "step": 19974
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2110103928097993e-06,
+      "loss": 0.9388,
+      "step": 19975
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2103557946036424e-06,
+      "loss": 1.0338,
+      "step": 19976
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2097013619666477e-06,
+      "loss": 1.0516,
+      "step": 19977
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2090470949111466e-06,
+      "loss": 1.0057,
+      "step": 19978
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2083929934494598e-06,
+      "loss": 0.9051,
+      "step": 19979
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2077390575939093e-06,
+      "loss": 0.9852,
+      "step": 19980
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.207085287356814e-06,
+      "loss": 1.0189,
+      "step": 19981
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2064316827504884e-06,
+      "loss": 1.0655,
+      "step": 19982
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.205778243787244e-06,
+      "loss": 1.057,
+      "step": 19983
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2051249704793874e-06,
+      "loss": 1.0503,
+      "step": 19984
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2044718628392315e-06,
+      "loss": 1.0192,
+      "step": 19985
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2038189208790717e-06,
+      "loss": 1.0098,
+      "step": 19986
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2031661446112106e-06,
+      "loss": 0.9706,
+      "step": 19987
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.20251353404794e-06,
+      "loss": 0.9432,
+      "step": 19988
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2018610892015612e-06,
+      "loss": 1.058,
+      "step": 19989
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2012088100843576e-06,
+      "loss": 1.0305,
+      "step": 19990
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.200556696708618e-06,
+      "loss": 0.9409,
+      "step": 19991
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1999047490866255e-06,
+      "loss": 1.0418,
+      "step": 19992
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1992529672306617e-06,
+      "loss": 0.9665,
+      "step": 19993
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1986013511530047e-06,
+      "loss": 0.9883,
+      "step": 19994
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1979499008659265e-06,
+      "loss": 0.9677,
+      "step": 19995
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.197298616381701e-06,
+      "loss": 0.9657,
+      "step": 19996
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.196647497712595e-06,
+      "loss": 1.0172,
+      "step": 19997
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1959965448708733e-06,
+      "loss": 0.9619,
+      "step": 19998
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1953457578687987e-06,
+      "loss": 0.9864,
+      "step": 19999
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.194695136718629e-06,
+      "loss": 0.9599,
+      "step": 20000
+    },
+    {
+      "epoch": 0.85,
+      "eval_text_loss": 0.757544219493866,
+      "eval_text_runtime": 1.4184,
+      "eval_text_samples_per_second": 52.171,
+      "eval_text_steps_per_second": 1.41,
+      "step": 20000
+    },
+    {
+      "epoch": 0.85,
+      "eval_svit_loss": 1.0779048204421997,
+      "eval_svit_runtime": 4.9544,
+      "eval_svit_samples_per_second": 137.454,
+      "eval_svit_steps_per_second": 2.22,
+      "step": 20000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.19404468143262e-06,
+      "loss": 0.9356,
+      "step": 20001
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1933943920230264e-06,
+      "loss": 0.974,
+      "step": 20002
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.192744268502094e-06,
+      "loss": 0.9646,
+      "step": 20003
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1920943108820715e-06,
+      "loss": 0.9382,
+      "step": 20004
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.191444519175201e-06,
+      "loss": 0.9299,
+      "step": 20005
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.190794893393723e-06,
+      "loss": 0.8685,
+      "step": 20006
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1901454335498753e-06,
+      "loss": 0.9531,
+      "step": 20007
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.189496139655889e-06,
+      "loss": 0.9963,
+      "step": 20008
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1888470117239983e-06,
+      "loss": 1.0295,
+      "step": 20009
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.188198049766428e-06,
+      "loss": 0.8827,
+      "step": 20010
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1875492537954023e-06,
+      "loss": 0.9591,
+      "step": 20011
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.186900623823145e-06,
+      "loss": 0.9852,
+      "step": 20012
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.186252159861872e-06,
+      "loss": 1.0169,
+      "step": 20013
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1856038619237995e-06,
+      "loss": 0.9619,
+      "step": 20014
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1849557300211387e-06,
+      "loss": 0.9498,
+      "step": 20015
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1843077641660993e-06,
+      "loss": 0.9882,
+      "step": 20016
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.183659964370886e-06,
+      "loss": 0.9661,
+      "step": 20017
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1830123306477014e-06,
+      "loss": 0.9407,
+      "step": 20018
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1823648630087447e-06,
+      "loss": 0.9549,
+      "step": 20019
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1817175614662137e-06,
+      "loss": 1.0109,
+      "step": 20020
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1810704260323016e-06,
+      "loss": 0.9571,
+      "step": 20021
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1804234567191916e-06,
+      "loss": 1.1068,
+      "step": 20022
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1797766535390797e-06,
+      "loss": 1.0217,
+      "step": 20023
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.179130016504146e-06,
+      "loss": 0.9084,
+      "step": 20024
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.178483545626573e-06,
+      "loss": 0.9723,
+      "step": 20025
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1778372409185326e-06,
+      "loss": 1.0342,
+      "step": 20026
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1771911023922033e-06,
+      "loss": 0.9321,
+      "step": 20027
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1765451300597573e-06,
+      "loss": 1.0397,
+      "step": 20028
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1758993239333627e-06,
+      "loss": 0.9674,
+      "step": 20029
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1752536840251794e-06,
+      "loss": 0.9618,
+      "step": 20030
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.174608210347374e-06,
+      "loss": 1.0294,
+      "step": 20031
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1739629029121057e-06,
+      "loss": 1.0464,
+      "step": 20032
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1733177617315274e-06,
+      "loss": 1.0053,
+      "step": 20033
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1726727868177903e-06,
+      "loss": 1.0022,
+      "step": 20034
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1720279781830467e-06,
+      "loss": 0.9247,
+      "step": 20035
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1713833358394453e-06,
+      "loss": 0.953,
+      "step": 20036
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.170738859799123e-06,
+      "loss": 0.9934,
+      "step": 20037
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.170094550074221e-06,
+      "loss": 0.9373,
+      "step": 20038
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.169450406676878e-06,
+      "loss": 0.9671,
+      "step": 20039
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1688064296192315e-06,
+      "loss": 0.9156,
+      "step": 20040
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.168162618913403e-06,
+      "loss": 0.9555,
+      "step": 20041
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1675189745715242e-06,
+      "loss": 1.0659,
+      "step": 20042
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1668754966057227e-06,
+      "loss": 1.0359,
+      "step": 20043
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.166232185028115e-06,
+      "loss": 0.9018,
+      "step": 20044
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1655890398508187e-06,
+      "loss": 0.9816,
+      "step": 20045
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.164946061085952e-06,
+      "loss": 1.0087,
+      "step": 20046
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1643032487456273e-06,
+      "loss": 0.9816,
+      "step": 20047
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1636606028419494e-06,
+      "loss": 1.05,
+      "step": 20048
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1630181233870227e-06,
+      "loss": 0.9827,
+      "step": 20049
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1623758103929535e-06,
+      "loss": 0.9867,
+      "step": 20050
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.161733663871842e-06,
+      "loss": 0.9601,
+      "step": 20051
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.16109168383578e-06,
+      "loss": 1.035,
+      "step": 20052
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.16044987029686e-06,
+      "loss": 0.9293,
+      "step": 20053
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.159808223267178e-06,
+      "loss": 1.0577,
+      "step": 20054
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1591667427588136e-06,
+      "loss": 0.9734,
+      "step": 20055
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1585254287838543e-06,
+      "loss": 0.9471,
+      "step": 20056
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1578842813543756e-06,
+      "loss": 0.9687,
+      "step": 20057
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.157243300482463e-06,
+      "loss": 1.0637,
+      "step": 20058
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.156602486180185e-06,
+      "loss": 0.9057,
+      "step": 20059
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1559618384596126e-06,
+      "loss": 0.9424,
+      "step": 20060
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1553213573328115e-06,
+      "loss": 0.9309,
+      "step": 20061
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.154681042811855e-06,
+      "loss": 0.9726,
+      "step": 20062
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1540408949087953e-06,
+      "loss": 1.1088,
+      "step": 20063
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.153400913635695e-06,
+      "loss": 0.9713,
+      "step": 20064
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.152761099004608e-06,
+      "loss": 1.0313,
+      "step": 20065
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.152121451027587e-06,
+      "loss": 1.0103,
+      "step": 20066
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.151481969716681e-06,
+      "loss": 0.9515,
+      "step": 20067
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1508426550839365e-06,
+      "loss": 0.892,
+      "step": 20068
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1502035071413953e-06,
+      "loss": 0.9615,
+      "step": 20069
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1495645259010968e-06,
+      "loss": 0.9674,
+      "step": 20070
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1489257113750774e-06,
+      "loss": 0.928,
+      "step": 20071
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1482870635753693e-06,
+      "loss": 1.0436,
+      "step": 20072
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.147648582514007e-06,
+      "loss": 0.9976,
+      "step": 20073
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1470102682030136e-06,
+      "loss": 0.9318,
+      "step": 20074
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.146372120654413e-06,
+      "loss": 0.9809,
+      "step": 20075
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.145734139880227e-06,
+      "loss": 1.023,
+      "step": 20076
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.145096325892472e-06,
+      "loss": 1.0766,
+      "step": 20077
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.144458678703163e-06,
+      "loss": 0.9969,
+      "step": 20078
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.143821198324312e-06,
+      "loss": 0.9531,
+      "step": 20079
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1431838847679262e-06,
+      "loss": 0.9553,
+      "step": 20080
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.142546738046012e-06,
+      "loss": 0.9434,
+      "step": 20081
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1419097581705683e-06,
+      "loss": 1.039,
+      "step": 20082
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1412729451535975e-06,
+      "loss": 0.9808,
+      "step": 20083
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.140636299007092e-06,
+      "loss": 1.1096,
+      "step": 20084
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.139999819743046e-06,
+      "loss": 0.9469,
+      "step": 20085
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1393635073734473e-06,
+      "loss": 1.0437,
+      "step": 20086
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1387273619102844e-06,
+      "loss": 0.9888,
+      "step": 20087
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.138091383365538e-06,
+      "loss": 1.0316,
+      "step": 20088
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1374555717511894e-06,
+      "loss": 0.998,
+      "step": 20089
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1368199270792134e-06,
+      "loss": 0.8881,
+      "step": 20090
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.136184449361586e-06,
+      "loss": 0.9262,
+      "step": 20091
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1355491386102754e-06,
+      "loss": 0.9607,
+      "step": 20092
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1349139948372514e-06,
+      "loss": 0.9974,
+      "step": 20093
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.134279018054475e-06,
+      "loss": 0.9935,
+      "step": 20094
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1336442082739095e-06,
+      "loss": 1.0949,
+      "step": 20095
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.133009565507511e-06,
+      "loss": 0.973,
+      "step": 20096
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.132375089767237e-06,
+      "loss": 1.0694,
+      "step": 20097
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1317407810650373e-06,
+      "loss": 1.1136,
+      "step": 20098
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1311066394128566e-06,
+      "loss": 1.0164,
+      "step": 20099
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.130472664822646e-06,
+      "loss": 0.9793,
+      "step": 20100
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1298388573063446e-06,
+      "loss": 0.9708,
+      "step": 20101
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1292052168758913e-06,
+      "loss": 1.0192,
+      "step": 20102
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.128571743543224e-06,
+      "loss": 0.9313,
+      "step": 20103
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.127938437320273e-06,
+      "loss": 0.9177,
+      "step": 20104
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.127305298218971e-06,
+      "loss": 0.9814,
+      "step": 20105
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.126672326251238e-06,
+      "loss": 0.9887,
+      "step": 20106
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.126039521429003e-06,
+      "loss": 0.8959,
+      "step": 20107
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.125406883764184e-06,
+      "loss": 0.9517,
+      "step": 20108
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1247744132686996e-06,
+      "loss": 0.9733,
+      "step": 20109
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1241421099544592e-06,
+      "loss": 0.9581,
+      "step": 20110
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1235099738333774e-06,
+      "loss": 0.9257,
+      "step": 20111
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1228780049173617e-06,
+      "loss": 1.0418,
+      "step": 20112
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1222462032183155e-06,
+      "loss": 1.0422,
+      "step": 20113
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.121614568748137e-06,
+      "loss": 1.0534,
+      "step": 20114
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1209831015187288e-06,
+      "loss": 0.9921,
+      "step": 20115
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1203518015419857e-06,
+      "loss": 1.033,
+      "step": 20116
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1197206688297945e-06,
+      "loss": 1.0198,
+      "step": 20117
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.119089703394046e-06,
+      "loss": 0.9881,
+      "step": 20118
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1184589052466274e-06,
+      "loss": 0.9726,
+      "step": 20119
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1178282743994225e-06,
+      "loss": 0.9842,
+      "step": 20120
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1171978108643055e-06,
+      "loss": 1.0013,
+      "step": 20121
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1165675146531527e-06,
+      "loss": 0.9371,
+      "step": 20122
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.115937385777841e-06,
+      "loss": 1.0087,
+      "step": 20123
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.11530742425024e-06,
+      "loss": 0.9674,
+      "step": 20124
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1146776300822126e-06,
+      "loss": 0.9937,
+      "step": 20125
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1140480032856203e-06,
+      "loss": 0.9921,
+      "step": 20126
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.113418543872331e-06,
+      "loss": 1.0009,
+      "step": 20127
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.112789251854196e-06,
+      "loss": 0.9923,
+      "step": 20128
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1121601272430704e-06,
+      "loss": 0.9308,
+      "step": 20129
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1115311700508025e-06,
+      "loss": 1.0044,
+      "step": 20130
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.110902380289246e-06,
+      "loss": 1.0772,
+      "step": 20131
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1102737579702403e-06,
+      "loss": 0.9195,
+      "step": 20132
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1096453031056262e-06,
+      "loss": 0.9633,
+      "step": 20133
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.109017015707245e-06,
+      "loss": 0.9926,
+      "step": 20134
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1083888957869324e-06,
+      "loss": 1.0081,
+      "step": 20135
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1077609433565174e-06,
+      "loss": 1.0212,
+      "step": 20136
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1071331584278254e-06,
+      "loss": 0.9846,
+      "step": 20137
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1065055410126914e-06,
+      "loss": 0.9873,
+      "step": 20138
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1058780911229295e-06,
+      "loss": 0.9643,
+      "step": 20139
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1052508087703617e-06,
+      "loss": 1.008,
+      "step": 20140
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.104623693966802e-06,
+      "loss": 0.9609,
+      "step": 20141
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1039967467240684e-06,
+      "loss": 0.9981,
+      "step": 20142
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.103369967053967e-06,
+      "loss": 0.9708,
+      "step": 20143
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.102743354968303e-06,
+      "loss": 1.0569,
+      "step": 20144
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1021169104788794e-06,
+      "loss": 0.9077,
+      "step": 20145
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.101490633597504e-06,
+      "loss": 0.991,
+      "step": 20146
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.100864524335965e-06,
+      "loss": 1.0091,
+      "step": 20147
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1002385827060604e-06,
+      "loss": 1.0051,
+      "step": 20148
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0996128087195797e-06,
+      "loss": 0.9375,
+      "step": 20149
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0989872023883115e-06,
+      "loss": 1.0058,
+      "step": 20150
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0983617637240405e-06,
+      "loss": 1.0434,
+      "step": 20151
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.097736492738547e-06,
+      "loss": 1.0867,
+      "step": 20152
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.097111389443609e-06,
+      "loss": 0.8901,
+      "step": 20153
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.096486453851002e-06,
+      "loss": 0.9783,
+      "step": 20154
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.095861685972498e-06,
+      "loss": 1.0425,
+      "step": 20155
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0952370858198654e-06,
+      "loss": 1.0522,
+      "step": 20156
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0946126534048708e-06,
+      "loss": 0.9974,
+      "step": 20157
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0939883887392744e-06,
+      "loss": 1.0124,
+      "step": 20158
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0933642918348366e-06,
+      "loss": 0.9446,
+      "step": 20159
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0927403627033128e-06,
+      "loss": 1.0025,
+      "step": 20160
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0921166013564564e-06,
+      "loss": 1.0439,
+      "step": 20161
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0914930078060168e-06,
+      "loss": 1.0063,
+      "step": 20162
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0908695820637412e-06,
+      "loss": 1.0094,
+      "step": 20163
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.090246324141372e-06,
+      "loss": 1.0439,
+      "step": 20164
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0896232340506507e-06,
+      "loss": 1.0039,
+      "step": 20165
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.089000311803313e-06,
+      "loss": 1.0846,
+      "step": 20166
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.088377557411093e-06,
+      "loss": 0.9051,
+      "step": 20167
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0877549708857227e-06,
+      "loss": 1.0771,
+      "step": 20168
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0871325522389285e-06,
+      "loss": 0.9979,
+      "step": 20169
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0865103014824352e-06,
+      "loss": 0.9803,
+      "step": 20170
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0858882186279638e-06,
+      "loss": 0.9472,
+      "step": 20171
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0852663036872323e-06,
+      "loss": 0.8981,
+      "step": 20172
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0846445566719554e-06,
+      "loss": 0.9429,
+      "step": 20173
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0840229775938457e-06,
+      "loss": 1.049,
+      "step": 20174
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.083401566464611e-06,
+      "loss": 0.9015,
+      "step": 20175
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.082780323295958e-06,
+      "loss": 0.9587,
+      "step": 20176
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.082159248099587e-06,
+      "loss": 1.0536,
+      "step": 20177
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.081538340887198e-06,
+      "loss": 0.9221,
+      "step": 20178
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.080917601670488e-06,
+      "loss": 0.9922,
+      "step": 20179
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.080297030461148e-06,
+      "loss": 0.9933,
+      "step": 20180
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.079676627270869e-06,
+      "loss": 1.0003,
+      "step": 20181
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0790563921113383e-06,
+      "loss": 1.0758,
+      "step": 20182
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0784363249942342e-06,
+      "loss": 1.0481,
+      "step": 20183
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0778164259312417e-06,
+      "loss": 0.9824,
+      "step": 20184
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0771966949340372e-06,
+      "loss": 1.0206,
+      "step": 20185
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0765771320142958e-06,
+      "loss": 0.9431,
+      "step": 20186
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0759577371836814e-06,
+      "loss": 0.9795,
+      "step": 20187
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0753385104538684e-06,
+      "loss": 0.9761,
+      "step": 20188
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.07471945183652e-06,
+      "loss": 1.0395,
+      "step": 20189
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.074100561343292e-06,
+      "loss": 1.0464,
+      "step": 20190
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0734818389858492e-06,
+      "loss": 0.9926,
+      "step": 20191
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0728632847758424e-06,
+      "loss": 1.0326,
+      "step": 20192
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0722448987249264e-06,
+      "loss": 0.9502,
+      "step": 20193
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.071626680844745e-06,
+      "loss": 0.9335,
+      "step": 20194
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.071008631146947e-06,
+      "loss": 1.0053,
+      "step": 20195
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0703907496431742e-06,
+      "loss": 1.06,
+      "step": 20196
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0697730363450667e-06,
+      "loss": 0.946,
+      "step": 20197
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.069155491264253e-06,
+      "loss": 1.0623,
+      "step": 20198
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.068538114412375e-06,
+      "loss": 0.9669,
+      "step": 20199
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0679209058010596e-06,
+      "loss": 0.9474,
+      "step": 20200
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0673038654419295e-06,
+      "loss": 0.9788,
+      "step": 20201
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0666869933466084e-06,
+      "loss": 0.9483,
+      "step": 20202
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0660702895267205e-06,
+      "loss": 0.9539,
+      "step": 20203
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0654537539938813e-06,
+      "loss": 1.0097,
+      "step": 20204
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0648373867597e-06,
+      "loss": 0.9574,
+      "step": 20205
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.06422118783579e-06,
+      "loss": 1.0007,
+      "step": 20206
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.063605157233759e-06,
+      "loss": 1.0326,
+      "step": 20207
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.062989294965213e-06,
+      "loss": 0.9434,
+      "step": 20208
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0623736010417494e-06,
+      "loss": 0.9074,
+      "step": 20209
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0617580754749645e-06,
+      "loss": 0.9895,
+      "step": 20210
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.061142718276459e-06,
+      "loss": 0.9647,
+      "step": 20211
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0605275294578196e-06,
+      "loss": 0.9919,
+      "step": 20212
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.059912509030635e-06,
+      "loss": 1.0128,
+      "step": 20213
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.059297657006489e-06,
+      "loss": 1.0049,
+      "step": 20214
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0586829733969696e-06,
+      "loss": 0.9571,
+      "step": 20215
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.058068458213649e-06,
+      "loss": 0.9169,
+      "step": 20216
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0574541114681043e-06,
+      "loss": 0.9943,
+      "step": 20217
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.056839933171908e-06,
+      "loss": 0.9901,
+      "step": 20218
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0562259233366335e-06,
+      "loss": 0.9759,
+      "step": 20219
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0556120819738402e-06,
+      "loss": 1.0268,
+      "step": 20220
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0549984090950926e-06,
+      "loss": 0.9918,
+      "step": 20221
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0543849047119547e-06,
+      "loss": 0.979,
+      "step": 20222
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0537715688359797e-06,
+      "loss": 0.9799,
+      "step": 20223
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0531584014787198e-06,
+      "loss": 0.9426,
+      "step": 20224
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0525454026517242e-06,
+      "loss": 0.9906,
+      "step": 20225
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0519325723665463e-06,
+      "loss": 0.9878,
+      "step": 20226
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.051319910634725e-06,
+      "loss": 0.9493,
+      "step": 20227
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0507074174677998e-06,
+      "loss": 1.0527,
+      "step": 20228
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0500950928773087e-06,
+      "loss": 0.9835,
+      "step": 20229
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0494829368747916e-06,
+      "loss": 0.9635,
+      "step": 20230
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.048870949471773e-06,
+      "loss": 0.9467,
+      "step": 20231
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0482591306797828e-06,
+      "loss": 1.0243,
+      "step": 20232
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0476474805103464e-06,
+      "loss": 0.9646,
+      "step": 20233
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.047035998974984e-06,
+      "loss": 0.9599,
+      "step": 20234
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0464246860852167e-06,
+      "loss": 0.9991,
+      "step": 20235
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.045813541852557e-06,
+      "loss": 1.0241,
+      "step": 20236
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.045202566288518e-06,
+      "loss": 0.9603,
+      "step": 20237
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.044591759404607e-06,
+      "loss": 0.9574,
+      "step": 20238
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.043981121212333e-06,
+      "loss": 1.0142,
+      "step": 20239
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0433706517231958e-06,
+      "loss": 1.0074,
+      "step": 20240
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.042760350948695e-06,
+      "loss": 0.9977,
+      "step": 20241
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0421502189003273e-06,
+      "loss": 1.03,
+      "step": 20242
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0415402555895858e-06,
+      "loss": 0.9975,
+      "step": 20243
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.04093046102796e-06,
+      "loss": 1.0542,
+      "step": 20244
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.040320835226937e-06,
+      "loss": 1.0878,
+      "step": 20245
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0397113781979993e-06,
+      "loss": 0.9393,
+      "step": 20246
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.039102089952628e-06,
+      "loss": 0.9923,
+      "step": 20247
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0384929705022993e-06,
+      "loss": 0.8911,
+      "step": 20248
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.037884019858488e-06,
+      "loss": 0.9595,
+      "step": 20249
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0372752380326644e-06,
+      "loss": 1.0368,
+      "step": 20250
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0366666250362967e-06,
+      "loss": 1.0234,
+      "step": 20251
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0360581808808477e-06,
+      "loss": 0.9837,
+      "step": 20252
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0354499055777797e-06,
+      "loss": 0.9205,
+      "step": 20253
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0348417991385507e-06,
+      "loss": 1.0468,
+      "step": 20254
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.034233861574615e-06,
+      "loss": 1.0241,
+      "step": 20255
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.033626092897425e-06,
+      "loss": 1.0126,
+      "step": 20256
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.033018493118427e-06,
+      "loss": 1.0634,
+      "step": 20257
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.03241106224907e-06,
+      "loss": 1.016,
+      "step": 20258
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.031803800300792e-06,
+      "loss": 1.0525,
+      "step": 20259
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.031196707285036e-06,
+      "loss": 0.9049,
+      "step": 20260
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.030589783213234e-06,
+      "loss": 1.0036,
+      "step": 20261
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0299830280968206e-06,
+      "loss": 0.9727,
+      "step": 20262
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.029376441947225e-06,
+      "loss": 0.9603,
+      "step": 20263
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0287700247758722e-06,
+      "loss": 0.9569,
+      "step": 20264
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0281637765941864e-06,
+      "loss": 0.9864,
+      "step": 20265
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.027557697413589e-06,
+      "loss": 1.0428,
+      "step": 20266
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0269517872454914e-06,
+      "loss": 0.9918,
+      "step": 20267
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0263460461013116e-06,
+      "loss": 1.0173,
+      "step": 20268
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0257404739924578e-06,
+      "loss": 1.0239,
+      "step": 20269
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0251350709303408e-06,
+      "loss": 0.9201,
+      "step": 20270
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.024529836926358e-06,
+      "loss": 0.9922,
+      "step": 20271
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0239247719919143e-06,
+      "loss": 0.991,
+      "step": 20272
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.02331987613841e-06,
+      "loss": 1.0308,
+      "step": 20273
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0227151493772326e-06,
+      "loss": 0.9341,
+      "step": 20274
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.022110591719775e-06,
+      "loss": 0.9607,
+      "step": 20275
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.021506203177429e-06,
+      "loss": 0.9445,
+      "step": 20276
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0209019837615786e-06,
+      "loss": 1.0179,
+      "step": 20277
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0202979334836005e-06,
+      "loss": 1.0073,
+      "step": 20278
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.019694052354878e-06,
+      "loss": 0.9659,
+      "step": 20279
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0190903403867846e-06,
+      "loss": 0.9618,
+      "step": 20280
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.018486797590695e-06,
+      "loss": 1.0677,
+      "step": 20281
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0178834239779707e-06,
+      "loss": 0.9585,
+      "step": 20282
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0172802195599864e-06,
+      "loss": 0.9855,
+      "step": 20283
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0166771843481006e-06,
+      "loss": 1.0251,
+      "step": 20284
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.01607431835367e-06,
+      "loss": 0.9735,
+      "step": 20285
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0154716215880522e-06,
+      "loss": 1.0656,
+      "step": 20286
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0148690940626026e-06,
+      "loss": 0.9547,
+      "step": 20287
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0142667357886713e-06,
+      "loss": 1.027,
+      "step": 20288
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0136645467776004e-06,
+      "loss": 0.925,
+      "step": 20289
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0130625270407348e-06,
+      "loss": 0.9533,
+      "step": 20290
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0124606765894172e-06,
+      "loss": 0.9501,
+      "step": 20291
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0118589954349844e-06,
+      "loss": 1.0272,
+      "step": 20292
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0112574835887666e-06,
+      "loss": 0.9809,
+      "step": 20293
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.010656141062095e-06,
+      "loss": 1.0532,
+      "step": 20294
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0100549678663017e-06,
+      "loss": 0.9516,
+      "step": 20295
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0094539640127064e-06,
+      "loss": 1.0097,
+      "step": 20296
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0088531295126313e-06,
+      "loss": 0.9068,
+      "step": 20297
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0082524643773913e-06,
+      "loss": 1.0241,
+      "step": 20298
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0076519686183093e-06,
+      "loss": 1.0289,
+      "step": 20299
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.007051642246689e-06,
+      "loss": 0.9784,
+      "step": 20300
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0064514852738407e-06,
+      "loss": 1.0337,
+      "step": 20301
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.005851497711069e-06,
+      "loss": 1.0478,
+      "step": 20302
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.00525167956968e-06,
+      "loss": 0.9742,
+      "step": 20303
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0046520308609663e-06,
+      "loss": 0.9442,
+      "step": 20304
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0040525515962274e-06,
+      "loss": 1.0611,
+      "step": 20305
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0034532417867537e-06,
+      "loss": 0.9524,
+      "step": 20306
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.002854101443834e-06,
+      "loss": 1.0533,
+      "step": 20307
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.002255130578756e-06,
+      "loss": 0.9633,
+      "step": 20308
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0016563292028003e-06,
+      "loss": 0.9871,
+      "step": 20309
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0010576973272513e-06,
+      "loss": 0.9061,
+      "step": 20310
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.000459234963379e-06,
+      "loss": 0.9821,
+      "step": 20311
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9998609421224605e-06,
+      "loss": 1.0291,
+      "step": 20312
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9992628188157616e-06,
+      "loss": 1.0391,
+      "step": 20313
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9986648650545562e-06,
+      "loss": 0.9123,
+      "step": 20314
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.998067080850101e-06,
+      "loss": 0.9898,
+      "step": 20315
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9974694662136608e-06,
+      "loss": 1.0726,
+      "step": 20316
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9968720211564894e-06,
+      "loss": 0.9732,
+      "step": 20317
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9962747456898425e-06,
+      "loss": 1.006,
+      "step": 20318
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.995677639824971e-06,
+      "loss": 1.0632,
+      "step": 20319
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9950807035731223e-06,
+      "loss": 0.9928,
+      "step": 20320
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9944839369455417e-06,
+      "loss": 0.9699,
+      "step": 20321
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9938873399534687e-06,
+      "loss": 0.8954,
+      "step": 20322
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9932909126081416e-06,
+      "loss": 1.0115,
+      "step": 20323
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.992694654920796e-06,
+      "loss": 0.9593,
+      "step": 20324
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9920985669026625e-06,
+      "loss": 1.0011,
+      "step": 20325
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9915026485649713e-06,
+      "loss": 0.9887,
+      "step": 20326
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9909068999189462e-06,
+      "loss": 1.0547,
+      "step": 20327
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9903113209758094e-06,
+      "loss": 0.9665,
+      "step": 20328
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.989715911746779e-06,
+      "loss": 1.0076,
+      "step": 20329
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9891206722430734e-06,
+      "loss": 0.9356,
+      "step": 20330
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9885256024759014e-06,
+      "loss": 0.9303,
+      "step": 20331
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.987930702456474e-06,
+      "loss": 0.9196,
+      "step": 20332
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.987335972195998e-06,
+      "loss": 1.0273,
+      "step": 20333
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.986741411705676e-06,
+      "loss": 0.9891,
+      "step": 20334
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.986147020996706e-06,
+      "loss": 1.0338,
+      "step": 20335
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9855528000802857e-06,
+      "loss": 0.9953,
+      "step": 20336
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.984958748967608e-06,
+      "loss": 0.9468,
+      "step": 20337
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9843648676698645e-06,
+      "loss": 1.0504,
+      "step": 20338
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9837711561982397e-06,
+      "loss": 1.0321,
+      "step": 20339
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.983177614563918e-06,
+      "loss": 0.9558,
+      "step": 20340
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9825842427780805e-06,
+      "loss": 0.9487,
+      "step": 20341
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.981991040851904e-06,
+      "loss": 0.9495,
+      "step": 20342
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9813980087965624e-06,
+      "loss": 1.0301,
+      "step": 20343
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.980805146623226e-06,
+      "loss": 1.0062,
+      "step": 20344
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.980212454343065e-06,
+      "loss": 0.931,
+      "step": 20345
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9796199319672415e-06,
+      "loss": 0.9873,
+      "step": 20346
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.979027579506916e-06,
+      "loss": 0.8711,
+      "step": 20347
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9784353969732487e-06,
+      "loss": 0.9696,
+      "step": 20348
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9778433843773925e-06,
+      "loss": 1.0006,
+      "step": 20349
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.977251541730504e-06,
+      "loss": 0.99,
+      "step": 20350
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.976659869043722e-06,
+      "loss": 1.0003,
+      "step": 20351
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.976068366328201e-06,
+      "loss": 0.9317,
+      "step": 20352
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9754770335950793e-06,
+      "loss": 0.9713,
+      "step": 20353
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9748858708554972e-06,
+      "loss": 0.947,
+      "step": 20354
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.974294878120586e-06,
+      "loss": 1.0462,
+      "step": 20355
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.973704055401483e-06,
+      "loss": 1.01,
+      "step": 20356
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9731134027093178e-06,
+      "loss": 1.0654,
+      "step": 20357
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.972522920055212e-06,
+      "loss": 1.0105,
+      "step": 20358
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.97193260745029e-06,
+      "loss": 0.9786,
+      "step": 20359
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9713424649056736e-06,
+      "loss": 0.9452,
+      "step": 20360
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9707524924324794e-06,
+      "loss": 1.0164,
+      "step": 20361
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.970162690041817e-06,
+      "loss": 1.0322,
+      "step": 20362
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9695730577447972e-06,
+      "loss": 0.9433,
+      "step": 20363
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9689835955525304e-06,
+      "loss": 0.979,
+      "step": 20364
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.968394303476119e-06,
+      "loss": 1.006,
+      "step": 20365
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9678051815266587e-06,
+      "loss": 1.0167,
+      "step": 20366
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.967216229715252e-06,
+      "loss": 1.0331,
+      "step": 20367
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9666274480529935e-06,
+      "loss": 0.8798,
+      "step": 20368
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.96603883655097e-06,
+      "loss": 0.9968,
+      "step": 20369
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9654503952202688e-06,
+      "loss": 1.0304,
+      "step": 20370
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.964862124071978e-06,
+      "loss": 0.9196,
+      "step": 20371
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9642740231171786e-06,
+      "loss": 0.9639,
+      "step": 20372
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9636860923669457e-06,
+      "loss": 0.9769,
+      "step": 20373
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9630983318323526e-06,
+      "loss": 1.0075,
+      "step": 20374
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.962510741524476e-06,
+      "loss": 0.9908,
+      "step": 20375
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.961923321454383e-06,
+      "loss": 0.9054,
+      "step": 20376
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9613360716331377e-06,
+      "loss": 1.0733,
+      "step": 20377
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.960748992071798e-06,
+      "loss": 0.9925,
+      "step": 20378
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9601620827814315e-06,
+      "loss": 0.9743,
+      "step": 20379
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.959575343773087e-06,
+      "loss": 0.9679,
+      "step": 20380
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.958988775057817e-06,
+      "loss": 1.0261,
+      "step": 20381
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.958402376646672e-06,
+      "loss": 0.9671,
+      "step": 20382
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.957816148550702e-06,
+      "loss": 1.0259,
+      "step": 20383
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9572300907809427e-06,
+      "loss": 0.9484,
+      "step": 20384
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.956644203348438e-06,
+      "loss": 0.9738,
+      "step": 20385
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9560584862642196e-06,
+      "loss": 0.9638,
+      "step": 20386
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.955472939539328e-06,
+      "loss": 0.9523,
+      "step": 20387
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.954887563184787e-06,
+      "loss": 0.9726,
+      "step": 20388
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.954302357211625e-06,
+      "loss": 0.9937,
+      "step": 20389
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9537173216308657e-06,
+      "loss": 1.0414,
+      "step": 20390
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.953132456453529e-06,
+      "loss": 0.9116,
+      "step": 20391
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.952547761690633e-06,
+      "loss": 0.9024,
+      "step": 20392
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9519632373531892e-06,
+      "loss": 1.0335,
+      "step": 20393
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9513788834522105e-06,
+      "loss": 0.9136,
+      "step": 20394
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.950794699998703e-06,
+      "loss": 0.9734,
+      "step": 20395
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9502106870036707e-06,
+      "loss": 0.9947,
+      "step": 20396
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.949626844478114e-06,
+      "loss": 1.0034,
+      "step": 20397
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.949043172433035e-06,
+      "loss": 0.9987,
+      "step": 20398
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9484596708794237e-06,
+      "loss": 0.9809,
+      "step": 20399
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.947876339828272e-06,
+      "loss": 1.1099,
+      "step": 20400
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9472931792905692e-06,
+      "loss": 0.9615,
+      "step": 20401
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9467101892772988e-06,
+      "loss": 0.9835,
+      "step": 20402
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9461273697994442e-06,
+      "loss": 0.9942,
+      "step": 20403
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9455447208679835e-06,
+      "loss": 1.0037,
+      "step": 20404
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9449622424938907e-06,
+      "loss": 0.8893,
+      "step": 20405
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.944379934688139e-06,
+      "loss": 0.939,
+      "step": 20406
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9437977974616974e-06,
+      "loss": 0.9615,
+      "step": 20407
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9432158308255302e-06,
+      "loss": 0.9902,
+      "step": 20408
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9426340347906015e-06,
+      "loss": 0.914,
+      "step": 20409
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9420524093678696e-06,
+      "loss": 1.0042,
+      "step": 20410
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.941470954568291e-06,
+      "loss": 0.9551,
+      "step": 20411
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9408896704028168e-06,
+      "loss": 0.982,
+      "step": 20412
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.940308556882399e-06,
+      "loss": 0.9795,
+      "step": 20413
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.939727614017982e-06,
+      "loss": 1.0249,
+      "step": 20414
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9391468418205093e-06,
+      "loss": 1.0045,
+      "step": 20415
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.938566240300922e-06,
+      "loss": 0.9199,
+      "step": 20416
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9379858094701564e-06,
+      "loss": 0.8793,
+      "step": 20417
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9374055493391454e-06,
+      "loss": 0.9318,
+      "step": 20418
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9368254599188187e-06,
+      "loss": 1.0111,
+      "step": 20419
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9362455412201043e-06,
+      "loss": 0.9475,
+      "step": 20420
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.935665793253927e-06,
+      "loss": 1.0245,
+      "step": 20421
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9350862160312047e-06,
+      "loss": 0.9948,
+      "step": 20422
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9345068095628573e-06,
+      "loss": 0.9228,
+      "step": 20423
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9339275738597973e-06,
+      "loss": 0.9653,
+      "step": 20424
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9333485089329375e-06,
+      "loss": 0.9755,
+      "step": 20425
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.932769614793184e-06,
+      "loss": 0.9263,
+      "step": 20426
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.932190891451442e-06,
+      "loss": 1.0769,
+      "step": 20427
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9316123389186137e-06,
+      "loss": 0.9933,
+      "step": 20428
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.931033957205595e-06,
+      "loss": 0.9858,
+      "step": 20429
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9304557463232844e-06,
+      "loss": 0.9777,
+      "step": 20430
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.929877706282568e-06,
+      "loss": 1.0357,
+      "step": 20431
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9292998370943385e-06,
+      "loss": 0.9538,
+      "step": 20432
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.92872213876948e-06,
+      "loss": 1.0533,
+      "step": 20433
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9281446113188775e-06,
+      "loss": 0.9992,
+      "step": 20434
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.927567254753403e-06,
+      "loss": 1.0133,
+      "step": 20435
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.926990069083937e-06,
+      "loss": 0.9468,
+      "step": 20436
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9264130543213514e-06,
+      "loss": 1.0396,
+      "step": 20437
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9258362104765173e-06,
+      "loss": 0.9883,
+      "step": 20438
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9252595375602935e-06,
+      "loss": 0.9952,
+      "step": 20439
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.924683035583549e-06,
+      "loss": 0.9325,
+      "step": 20440
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.924106704557144e-06,
+      "loss": 1.0065,
+      "step": 20441
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9235305444919304e-06,
+      "loss": 1.0202,
+      "step": 20442
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9229545553987605e-06,
+      "loss": 0.9954,
+      "step": 20443
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9223787372884897e-06,
+      "loss": 0.9944,
+      "step": 20444
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.921803090171963e-06,
+      "loss": 0.9856,
+      "step": 20445
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9212276140600217e-06,
+      "loss": 0.958,
+      "step": 20446
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9206523089635034e-06,
+      "loss": 0.9401,
+      "step": 20447
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.920077174893251e-06,
+      "loss": 0.961,
+      "step": 20448
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.919502211860097e-06,
+      "loss": 0.928,
+      "step": 20449
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9189274198748687e-06,
+      "loss": 1.0655,
+      "step": 20450
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.918352798948394e-06,
+      "loss": 0.9372,
+      "step": 20451
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9177783490915015e-06,
+      "loss": 0.9442,
+      "step": 20452
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9172040703150064e-06,
+      "loss": 0.9718,
+      "step": 20453
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.916629962629727e-06,
+      "loss": 1.026,
+      "step": 20454
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.91605602604648e-06,
+      "loss": 0.904,
+      "step": 20455
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.915482260576078e-06,
+      "loss": 0.9351,
+      "step": 20456
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.914908666229326e-06,
+      "loss": 1.1034,
+      "step": 20457
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.914335243017026e-06,
+      "loss": 0.9791,
+      "step": 20458
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.913761990949986e-06,
+      "loss": 0.9412,
+      "step": 20459
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.913188910039002e-06,
+      "loss": 1.0731,
+      "step": 20460
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.912616000294867e-06,
+      "loss": 0.9539,
+      "step": 20461
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.912043261728371e-06,
+      "loss": 1.0675,
+      "step": 20462
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9114706943503106e-06,
+      "loss": 1.0075,
+      "step": 20463
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.910898298171463e-06,
+      "loss": 0.9935,
+      "step": 20464
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.910326073202614e-06,
+      "loss": 0.9983,
+      "step": 20465
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9097540194545406e-06,
+      "loss": 0.9462,
+      "step": 20466
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9091821369380225e-06,
+      "loss": 0.9228,
+      "step": 20467
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.908610425663828e-06,
+      "loss": 0.8656,
+      "step": 20468
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.908038885642729e-06,
+      "loss": 0.9579,
+      "step": 20469
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9074675168854874e-06,
+      "loss": 0.9546,
+      "step": 20470
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9068963194028734e-06,
+      "loss": 0.9991,
+      "step": 20471
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.90632529320564e-06,
+      "loss": 0.9266,
+      "step": 20472
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9057544383045446e-06,
+      "loss": 0.9509,
+      "step": 20473
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9051837547103425e-06,
+      "loss": 0.9604,
+      "step": 20474
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9046132424337817e-06,
+      "loss": 1.0157,
+      "step": 20475
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.90404290148561e-06,
+      "loss": 0.9931,
+      "step": 20476
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.90347273187657e-06,
+      "loss": 1.0475,
+      "step": 20477
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.902902733617402e-06,
+      "loss": 0.9741,
+      "step": 20478
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9023329067188445e-06,
+      "loss": 1.0476,
+      "step": 20479
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9017632511916295e-06,
+      "loss": 1.0129,
+      "step": 20480
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9011937670464875e-06,
+      "loss": 1.0295,
+      "step": 20481
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9006244542941464e-06,
+      "loss": 1.0415,
+      "step": 20482
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9000553129453304e-06,
+      "loss": 0.9821,
+      "step": 20483
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8994863430107603e-06,
+      "loss": 1.0339,
+      "step": 20484
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8989175445011534e-06,
+      "loss": 0.907,
+      "step": 20485
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.898348917427224e-06,
+      "loss": 1.0023,
+      "step": 20486
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.897780461799683e-06,
+      "loss": 0.981,
+      "step": 20487
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8972121776292383e-06,
+      "loss": 0.995,
+      "step": 20488
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.896644064926596e-06,
+      "loss": 0.9941,
+      "step": 20489
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.896076123702457e-06,
+      "loss": 0.9325,
+      "step": 20490
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.895508353967518e-06,
+      "loss": 0.9197,
+      "step": 20491
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.894940755732475e-06,
+      "loss": 0.9493,
+      "step": 20492
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8943733290080205e-06,
+      "loss": 1.027,
+      "step": 20493
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8938060738048414e-06,
+      "loss": 0.9801,
+      "step": 20494
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8932389901336244e-06,
+      "loss": 0.9258,
+      "step": 20495
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8926720780050513e-06,
+      "loss": 0.9196,
+      "step": 20496
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8921053374298006e-06,
+      "loss": 0.9744,
+      "step": 20497
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8915387684185484e-06,
+      "loss": 0.9497,
+      "step": 20498
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.890972370981966e-06,
+      "loss": 1.0528,
+      "step": 20499
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8904061451307235e-06,
+      "loss": 0.9097,
+      "step": 20500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8898400908754867e-06,
+      "loss": 1.0566,
+      "step": 20501
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8892742082269176e-06,
+      "loss": 0.9781,
+      "step": 20502
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.888708497195677e-06,
+      "loss": 1.0115,
+      "step": 20503
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8881429577924207e-06,
+      "loss": 1.0236,
+      "step": 20504
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.887577590027799e-06,
+      "loss": 0.9829,
+      "step": 20505
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8870123939124663e-06,
+      "loss": 1.0171,
+      "step": 20506
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8864473694570672e-06,
+      "loss": 0.937,
+      "step": 20507
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8858825166722416e-06,
+      "loss": 0.9754,
+      "step": 20508
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.885317835568634e-06,
+      "loss": 0.9803,
+      "step": 20509
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8847533261568796e-06,
+      "loss": 0.9952,
+      "step": 20510
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8841889884476117e-06,
+      "loss": 1.0603,
+      "step": 20511
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8836248224514608e-06,
+      "loss": 1.0635,
+      "step": 20512
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8830608281790553e-06,
+      "loss": 0.9644,
+      "step": 20513
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8824970056410184e-06,
+      "loss": 0.9858,
+      "step": 20514
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.881933354847968e-06,
+      "loss": 1.0301,
+      "step": 20515
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8813698758105247e-06,
+      "loss": 1.0728,
+      "step": 20516
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8808065685393018e-06,
+      "loss": 0.9928,
+      "step": 20517
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8802434330449125e-06,
+      "loss": 0.9166,
+      "step": 20518
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8796804693379576e-06,
+      "loss": 0.983,
+      "step": 20519
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.87911767742905e-06,
+      "loss": 0.9667,
+      "step": 20520
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.878555057328785e-06,
+      "loss": 1.0257,
+      "step": 20521
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.877992609047765e-06,
+      "loss": 0.973,
+      "step": 20522
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8774303325965794e-06,
+      "loss": 0.9901,
+      "step": 20523
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.876868227985824e-06,
+      "loss": 0.9803,
+      "step": 20524
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.876306295226088e-06,
+      "loss": 0.9131,
+      "step": 20525
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.875744534327951e-06,
+      "loss": 0.9746,
+      "step": 20526
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8751829453019976e-06,
+      "loss": 0.9648,
+      "step": 20527
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8746215281588087e-06,
+      "loss": 1.0281,
+      "step": 20528
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8740602829089594e-06,
+      "loss": 1.0256,
+      "step": 20529
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.873499209563018e-06,
+      "loss": 1.0791,
+      "step": 20530
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8729383081315532e-06,
+      "loss": 1.0517,
+      "step": 20531
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8723775786251353e-06,
+      "loss": 0.9823,
+      "step": 20532
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.871817021054326e-06,
+      "loss": 0.9845,
+      "step": 20533
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8712566354296815e-06,
+      "loss": 0.9914,
+      "step": 20534
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.870696421761756e-06,
+      "loss": 0.896,
+      "step": 20535
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8701363800611093e-06,
+      "loss": 1.0075,
+      "step": 20536
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8695765103382848e-06,
+      "loss": 0.9769,
+      "step": 20537
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.86901681260383e-06,
+      "loss": 0.9614,
+      "step": 20538
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.868457286868287e-06,
+      "loss": 0.9482,
+      "step": 20539
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8678979331422e-06,
+      "loss": 1.0253,
+      "step": 20540
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.867338751436101e-06,
+      "loss": 0.9632,
+      "step": 20541
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.866779741760523e-06,
+      "loss": 1.0468,
+      "step": 20542
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.866220904125998e-06,
+      "loss": 0.974,
+      "step": 20543
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.865662238543055e-06,
+      "loss": 1.0242,
+      "step": 20544
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.865103745022212e-06,
+      "loss": 0.9989,
+      "step": 20545
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.86454542357399e-06,
+      "loss": 0.9778,
+      "step": 20546
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.863987274208913e-06,
+      "loss": 0.9376,
+      "step": 20547
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8634292969374867e-06,
+      "loss": 0.9455,
+      "step": 20548
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.862871491770225e-06,
+      "loss": 1.0246,
+      "step": 20549
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8623138587176328e-06,
+      "loss": 0.9311,
+      "step": 20550
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.86175639779022e-06,
+      "loss": 0.9127,
+      "step": 20551
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8611991089984816e-06,
+      "loss": 0.9702,
+      "step": 20552
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.860641992352917e-06,
+      "loss": 0.9905,
+      "step": 20553
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8600850478640198e-06,
+      "loss": 1.0532,
+      "step": 20554
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.859528275542285e-06,
+      "loss": 0.9545,
+      "step": 20555
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8589716753981956e-06,
+      "loss": 1.0229,
+      "step": 20556
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.858415247442238e-06,
+      "loss": 0.9635,
+      "step": 20557
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8578589916848933e-06,
+      "loss": 0.9771,
+      "step": 20558
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.85730290813664e-06,
+      "loss": 0.983,
+      "step": 20559
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8567469968079533e-06,
+      "loss": 0.9855,
+      "step": 20560
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8561912577093035e-06,
+      "loss": 0.9952,
+      "step": 20561
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8556356908511596e-06,
+      "loss": 0.9592,
+      "step": 20562
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8550802962439875e-06,
+      "loss": 0.9403,
+      "step": 20563
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.854525073898249e-06,
+      "loss": 0.9942,
+      "step": 20564
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8539700238244006e-06,
+      "loss": 0.981,
+      "step": 20565
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.853415146032901e-06,
+      "loss": 1.006,
+      "step": 20566
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8528604405342e-06,
+      "loss": 0.9018,
+      "step": 20567
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8523059073387472e-06,
+      "loss": 1.0664,
+      "step": 20568
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.851751546456988e-06,
+      "loss": 1.0528,
+      "step": 20569
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8511973578993647e-06,
+      "loss": 0.9816,
+      "step": 20570
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8506433416763184e-06,
+      "loss": 1.0369,
+      "step": 20571
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.850089497798282e-06,
+      "loss": 0.964,
+      "step": 20572
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.849535826275689e-06,
+      "loss": 0.9912,
+      "step": 20573
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.848982327118971e-06,
+      "loss": 0.9809,
+      "step": 20574
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.848429000338552e-06,
+      "loss": 0.9861,
+      "step": 20575
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8478758459448553e-06,
+      "loss": 0.9331,
+      "step": 20576
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.847322863948301e-06,
+      "loss": 0.9742,
+      "step": 20577
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.846770054359305e-06,
+      "loss": 1.0014,
+      "step": 20578
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.846217417188281e-06,
+      "loss": 0.9911,
+      "step": 20579
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.845664952445638e-06,
+      "loss": 0.9637,
+      "step": 20580
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8451126601417847e-06,
+      "loss": 1.0617,
+      "step": 20581
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8445605402871224e-06,
+      "loss": 0.9658,
+      "step": 20582
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8440085928920536e-06,
+      "loss": 0.9491,
+      "step": 20583
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8434568179669723e-06,
+      "loss": 0.9261,
+      "step": 20584
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.842905215522275e-06,
+      "loss": 0.9614,
+      "step": 20585
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8423537855683494e-06,
+      "loss": 0.9928,
+      "step": 20586
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8418025281155847e-06,
+      "loss": 1.0677,
+      "step": 20587
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8412514431743655e-06,
+      "loss": 0.9933,
+      "step": 20588
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8407005307550708e-06,
+      "loss": 0.9641,
+      "step": 20589
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8401497908680776e-06,
+      "loss": 0.9995,
+      "step": 20590
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.839599223523764e-06,
+      "loss": 0.9188,
+      "step": 20591
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8390488287324937e-06,
+      "loss": 0.9974,
+      "step": 20592
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8384986065046405e-06,
+      "loss": 1.0093,
+      "step": 20593
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8379485568505684e-06,
+      "loss": 0.9917,
+      "step": 20594
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8373986797806375e-06,
+      "loss": 1.0018,
+      "step": 20595
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8368489753052033e-06,
+      "loss": 0.9545,
+      "step": 20596
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8362994434346237e-06,
+      "loss": 1.0229,
+      "step": 20597
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.835750084179251e-06,
+      "loss": 1.0141,
+      "step": 20598
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8352008975494285e-06,
+      "loss": 0.9417,
+      "step": 20599
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.834651883555506e-06,
+      "loss": 1.0124,
+      "step": 20600
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8341030422078233e-06,
+      "loss": 1.0185,
+      "step": 20601
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8335543735167204e-06,
+      "loss": 1.0378,
+      "step": 20602
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8330058774925275e-06,
+      "loss": 1.0416,
+      "step": 20603
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8324575541455816e-06,
+      "loss": 0.9633,
+      "step": 20604
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8319094034862107e-06,
+      "loss": 0.9334,
+      "step": 20605
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.83136142552474e-06,
+      "loss": 0.9868,
+      "step": 20606
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.830813620271489e-06,
+      "loss": 0.9786,
+      "step": 20607
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.83026598773678e-06,
+      "loss": 0.931,
+      "step": 20608
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8297185279309283e-06,
+      "loss": 0.9677,
+      "step": 20609
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.829171240864245e-06,
+      "loss": 1.0085,
+      "step": 20610
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8286241265470363e-06,
+      "loss": 0.9786,
+      "step": 20611
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.828077184989614e-06,
+      "loss": 0.9322,
+      "step": 20612
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.82753041620228e-06,
+      "loss": 0.9798,
+      "step": 20613
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8269838201953304e-06,
+      "loss": 1.037,
+      "step": 20614
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8264373969790607e-06,
+      "loss": 0.9363,
+      "step": 20615
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8258911465637675e-06,
+      "loss": 0.9883,
+      "step": 20616
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8253450689597407e-06,
+      "loss": 0.9441,
+      "step": 20617
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8247991641772632e-06,
+      "loss": 1.0355,
+      "step": 20618
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.824253432226618e-06,
+      "loss": 1.0016,
+      "step": 20619
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8237078731180915e-06,
+      "loss": 0.9958,
+      "step": 20620
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8231624868619526e-06,
+      "loss": 1.0038,
+      "step": 20621
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.822617273468478e-06,
+      "loss": 0.9673,
+      "step": 20622
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8220722329479346e-06,
+      "loss": 0.924,
+      "step": 20623
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.821527365310596e-06,
+      "loss": 0.9937,
+      "step": 20624
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.820982670566721e-06,
+      "loss": 0.9898,
+      "step": 20625
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.82043814872657e-06,
+      "loss": 0.9487,
+      "step": 20626
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8198937998003986e-06,
+      "loss": 0.9513,
+      "step": 20627
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.819349623798468e-06,
+      "loss": 0.9496,
+      "step": 20628
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.81880562073102e-06,
+      "loss": 0.9577,
+      "step": 20629
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8182617906083046e-06,
+      "loss": 0.9245,
+      "step": 20630
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8177181334405704e-06,
+      "loss": 1.0383,
+      "step": 20631
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.817174649238053e-06,
+      "loss": 0.931,
+      "step": 20632
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8166313380109913e-06,
+      "loss": 0.9924,
+      "step": 20633
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8160881997696168e-06,
+      "loss": 0.9878,
+      "step": 20634
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8155452345241685e-06,
+      "loss": 0.9454,
+      "step": 20635
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.815002442284867e-06,
+      "loss": 0.9958,
+      "step": 20636
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.814459823061938e-06,
+      "loss": 0.9869,
+      "step": 20637
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8139173768656027e-06,
+      "loss": 0.9452,
+      "step": 20638
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.813375103706084e-06,
+      "loss": 0.9918,
+      "step": 20639
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8128330035935905e-06,
+      "loss": 0.925,
+      "step": 20640
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.812291076538335e-06,
+      "loss": 1.0227,
+      "step": 20641
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.811749322550528e-06,
+      "loss": 0.9849,
+      "step": 20642
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8112077416403716e-06,
+      "loss": 1.0404,
+      "step": 20643
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8106663338180684e-06,
+      "loss": 0.9513,
+      "step": 20644
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.810125099093818e-06,
+      "loss": 0.9864,
+      "step": 20645
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8095840374778153e-06,
+      "loss": 0.8981,
+      "step": 20646
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.809043148980251e-06,
+      "loss": 1.0254,
+      "step": 20647
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8085024336113144e-06,
+      "loss": 0.9647,
+      "step": 20648
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8079618913811915e-06,
+      "loss": 0.9129,
+      "step": 20649
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8074215223000626e-06,
+      "loss": 0.9881,
+      "step": 20650
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8068813263781083e-06,
+      "loss": 0.9518,
+      "step": 20651
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.806341303625504e-06,
+      "loss": 0.9536,
+      "step": 20652
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8058014540524206e-06,
+      "loss": 0.958,
+      "step": 20653
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8052617776690283e-06,
+      "loss": 1.0051,
+      "step": 20654
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8047222744854944e-06,
+      "loss": 0.9376,
+      "step": 20655
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8041829445119786e-06,
+      "loss": 0.9333,
+      "step": 20656
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.803643787758641e-06,
+      "loss": 1.07,
+      "step": 20657
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.803104804235639e-06,
+      "loss": 0.9712,
+      "step": 20658
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8025659939531237e-06,
+      "loss": 0.9546,
+      "step": 20659
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8020273569212455e-06,
+      "loss": 0.9152,
+      "step": 20660
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8014888931501506e-06,
+      "loss": 1.0096,
+      "step": 20661
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.800950602649981e-06,
+      "loss": 0.9521,
+      "step": 20662
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.800412485430878e-06,
+      "loss": 0.9947,
+      "step": 20663
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.799874541502976e-06,
+      "loss": 1.018,
+      "step": 20664
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.79933677087641e-06,
+      "loss": 0.9718,
+      "step": 20665
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7987991735613097e-06,
+      "loss": 1.0129,
+      "step": 20666
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7982617495678e-06,
+      "loss": 0.9552,
+      "step": 20667
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7977244989060072e-06,
+      "loss": 0.9907,
+      "step": 20668
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7971874215860486e-06,
+      "loss": 0.9665,
+      "step": 20669
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7966505176180425e-06,
+      "loss": 0.9675,
+      "step": 20670
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7961137870121028e-06,
+      "loss": 0.9928,
+      "step": 20671
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.795577229778338e-06,
+      "loss": 1.0076,
+      "step": 20672
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7950408459268567e-06,
+      "loss": 0.972,
+      "step": 20673
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.794504635467762e-06,
+      "loss": 0.9515,
+      "step": 20674
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7939685984111566e-06,
+      "loss": 0.9446,
+      "step": 20675
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7934327347671333e-06,
+      "loss": 1.0578,
+      "step": 20676
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7928970445457893e-06,
+      "loss": 1.0279,
+      "step": 20677
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.792361527757215e-06,
+      "loss": 1.0255,
+      "step": 20678
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7918261844114994e-06,
+      "loss": 0.9248,
+      "step": 20679
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7912910145187214e-06,
+      "loss": 0.9901,
+      "step": 20680
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7907560180889664e-06,
+      "loss": 1.0384,
+      "step": 20681
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7902211951323132e-06,
+      "loss": 1.0544,
+      "step": 20682
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.789686545658832e-06,
+      "loss": 0.9235,
+      "step": 20683
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7891520696785937e-06,
+      "loss": 0.975,
+      "step": 20684
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7886177672016705e-06,
+      "loss": 1.0008,
+      "step": 20685
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.788083638238126e-06,
+      "loss": 0.9894,
+      "step": 20686
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7875496827980177e-06,
+      "loss": 1.0106,
+      "step": 20687
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7870159008914067e-06,
+      "loss": 1.0109,
+      "step": 20688
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7864822925283487e-06,
+      "loss": 0.9481,
+      "step": 20689
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.785948857718894e-06,
+      "loss": 0.9189,
+      "step": 20690
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.785415596473088e-06,
+      "loss": 1.0624,
+      "step": 20691
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7848825088009792e-06,
+      "loss": 0.9488,
+      "step": 20692
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7843495947126112e-06,
+      "loss": 1.0622,
+      "step": 20693
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.783816854218017e-06,
+      "loss": 0.9929,
+      "step": 20694
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.783284287327233e-06,
+      "loss": 0.9413,
+      "step": 20695
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.782751894050293e-06,
+      "loss": 0.9689,
+      "step": 20696
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7822196743972273e-06,
+      "loss": 1.0331,
+      "step": 20697
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7816876283780583e-06,
+      "loss": 0.9684,
+      "step": 20698
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.781155756002806e-06,
+      "loss": 0.993,
+      "step": 20699
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7806240572814924e-06,
+      "loss": 1.0702,
+      "step": 20700
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.780092532224136e-06,
+      "loss": 0.9473,
+      "step": 20701
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.779561180840743e-06,
+      "loss": 0.9486,
+      "step": 20702
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7790300031413246e-06,
+      "loss": 0.9972,
+      "step": 20703
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7784989991358883e-06,
+      "loss": 1.0323,
+      "step": 20704
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7779681688344342e-06,
+      "loss": 0.9652,
+      "step": 20705
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.777437512246962e-06,
+      "loss": 1.0216,
+      "step": 20706
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7769070293834667e-06,
+      "loss": 1.039,
+      "step": 20707
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.776376720253946e-06,
+      "loss": 1.0051,
+      "step": 20708
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.775846584868384e-06,
+      "loss": 0.9767,
+      "step": 20709
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.775316623236767e-06,
+      "loss": 0.9488,
+      "step": 20710
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.774786835369078e-06,
+      "loss": 1.059,
+      "step": 20711
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7742572212753005e-06,
+      "loss": 1.0316,
+      "step": 20712
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7737277809654064e-06,
+      "loss": 0.949,
+      "step": 20713
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7731985144493703e-06,
+      "loss": 0.9985,
+      "step": 20714
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.772669421737161e-06,
+      "loss": 0.9728,
+      "step": 20715
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7721405028387463e-06,
+      "loss": 0.9491,
+      "step": 20716
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.771611757764088e-06,
+      "loss": 0.9666,
+      "step": 20717
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.771083186523146e-06,
+      "loss": 1.0047,
+      "step": 20718
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.77055478912588e-06,
+      "loss": 0.9704,
+      "step": 20719
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.77002656558224e-06,
+      "loss": 0.9894,
+      "step": 20720
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.769498515902176e-06,
+      "loss": 0.968,
+      "step": 20721
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7689706400956354e-06,
+      "loss": 1.002,
+      "step": 20722
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7684429381725645e-06,
+      "loss": 1.0076,
+      "step": 20723
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7679154101428995e-06,
+      "loss": 1.0043,
+      "step": 20724
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.76738805601658e-06,
+      "loss": 0.9727,
+      "step": 20725
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.766860875803538e-06,
+      "loss": 1.0586,
+      "step": 20726
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7663338695137043e-06,
+      "loss": 0.9977,
+      "step": 20727
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7658070371570067e-06,
+      "loss": 0.9966,
+      "step": 20728
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.765280378743369e-06,
+      "loss": 1.0379,
+      "step": 20729
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7647538942827117e-06,
+      "loss": 0.9238,
+      "step": 20730
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.764227583784951e-06,
+      "loss": 1.0146,
+      "step": 20731
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7637014472600024e-06,
+      "loss": 1.0358,
+      "step": 20732
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7631754847177762e-06,
+      "loss": 0.9465,
+      "step": 20733
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7626496961681795e-06,
+      "loss": 0.9915,
+      "step": 20734
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.762124081621117e-06,
+      "loss": 1.0376,
+      "step": 20735
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7615986410864893e-06,
+      "loss": 0.9614,
+      "step": 20736
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7610733745741938e-06,
+      "loss": 0.9597,
+      "step": 20737
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7605482820941254e-06,
+      "loss": 1.0462,
+      "step": 20738
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.760023363656175e-06,
+      "loss": 0.9477,
+      "step": 20739
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7594986192702304e-06,
+      "loss": 0.9987,
+      "step": 20740
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.758974048946176e-06,
+      "loss": 1.0086,
+      "step": 20741
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7584496526938935e-06,
+      "loss": 0.9808,
+      "step": 20742
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7579254305232597e-06,
+      "loss": 1.0303,
+      "step": 20743
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7574013824441504e-06,
+      "loss": 1.014,
+      "step": 20744
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.756877508466437e-06,
+      "loss": 0.9847,
+      "step": 20745
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.756353808599988e-06,
+      "loss": 0.9437,
+      "step": 20746
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.755830282854666e-06,
+      "loss": 0.9978,
+      "step": 20747
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.755306931240335e-06,
+      "loss": 0.9685,
+      "step": 20748
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.754783753766852e-06,
+      "loss": 1.2267,
+      "step": 20749
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7542607504440733e-06,
+      "loss": 0.9956,
+      "step": 20750
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7537379212818496e-06,
+      "loss": 0.997,
+      "step": 20751
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7532152662900294e-06,
+      "loss": 0.9131,
+      "step": 20752
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7526927854784583e-06,
+      "loss": 0.9448,
+      "step": 20753
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7521704788569783e-06,
+      "loss": 0.9588,
+      "step": 20754
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7516483464354273e-06,
+      "loss": 0.9252,
+      "step": 20755
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7511263882236406e-06,
+      "loss": 0.9843,
+      "step": 20756
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7506046042314506e-06,
+      "loss": 0.9911,
+      "step": 20757
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.750082994468686e-06,
+      "loss": 1.0377,
+      "step": 20758
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.749561558945174e-06,
+      "loss": 0.9722,
+      "step": 20759
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.749040297670732e-06,
+      "loss": 1.0221,
+      "step": 20760
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7485192106551843e-06,
+      "loss": 0.9646,
+      "step": 20761
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7479982979083444e-06,
+      "loss": 1.0631,
+      "step": 20762
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.747477559440025e-06,
+      "loss": 0.9745,
+      "step": 20763
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.746956995260033e-06,
+      "loss": 0.953,
+      "step": 20764
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7464366053781767e-06,
+      "loss": 0.922,
+      "step": 20765
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7459163898042598e-06,
+      "loss": 1.1067,
+      "step": 20766
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7453963485480776e-06,
+      "loss": 0.9973,
+      "step": 20767
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7448764816194253e-06,
+      "loss": 0.9518,
+      "step": 20768
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.744356789028101e-06,
+      "loss": 1.0459,
+      "step": 20769
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7438372707838936e-06,
+      "loss": 0.9816,
+      "step": 20770
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.743317926896584e-06,
+      "loss": 0.9266,
+      "step": 20771
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7427987573759575e-06,
+      "loss": 0.9504,
+      "step": 20772
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7422797622317944e-06,
+      "loss": 1.0794,
+      "step": 20773
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7417609414738724e-06,
+      "loss": 0.9585,
+      "step": 20774
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.74124229511196e-06,
+      "loss": 0.9399,
+      "step": 20775
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.740723823155831e-06,
+      "loss": 0.9664,
+      "step": 20776
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7402055256152518e-06,
+      "loss": 1.0107,
+      "step": 20777
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.739687402499981e-06,
+      "loss": 0.9439,
+      "step": 20778
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7391694538197814e-06,
+      "loss": 0.9111,
+      "step": 20779
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.73865167958441e-06,
+      "loss": 1.0,
+      "step": 20780
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.738134079803622e-06,
+      "loss": 0.9952,
+      "step": 20781
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.737616654487162e-06,
+      "loss": 0.8806,
+      "step": 20782
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7370994036447785e-06,
+      "loss": 0.9849,
+      "step": 20783
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.736582327286218e-06,
+      "loss": 0.9585,
+      "step": 20784
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7360654254212195e-06,
+      "loss": 1.016,
+      "step": 20785
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7355486980595173e-06,
+      "loss": 0.9346,
+      "step": 20786
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7350321452108444e-06,
+      "loss": 0.9813,
+      "step": 20787
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7345157668849365e-06,
+      "loss": 0.9815,
+      "step": 20788
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7339995630915153e-06,
+      "loss": 1.0098,
+      "step": 20789
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7334835338403054e-06,
+      "loss": 1.0501,
+      "step": 20790
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7329676791410277e-06,
+      "loss": 1.0609,
+      "step": 20791
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.732451999003402e-06,
+      "loss": 0.9131,
+      "step": 20792
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7319364934371374e-06,
+      "loss": 1.0127,
+      "step": 20793
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7314211624519473e-06,
+      "loss": 0.9818,
+      "step": 20794
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7309060060575368e-06,
+      "loss": 1.0181,
+      "step": 20795
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7303910242636146e-06,
+      "loss": 0.9554,
+      "step": 20796
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7298762170798755e-06,
+      "loss": 1.0018,
+      "step": 20797
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7293615845160193e-06,
+      "loss": 1.0153,
+      "step": 20798
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7288471265817406e-06,
+      "loss": 0.9819,
+      "step": 20799
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7283328432867296e-06,
+      "loss": 0.9747,
+      "step": 20800
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7278187346406736e-06,
+      "loss": 0.9807,
+      "step": 20801
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.727304800653257e-06,
+      "loss": 1.0099,
+      "step": 20802
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7267910413341603e-06,
+      "loss": 1.0111,
+      "step": 20803
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7262774566930627e-06,
+      "loss": 0.9744,
+      "step": 20804
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7257640467396364e-06,
+      "loss": 0.8499,
+      "step": 20805
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.725250811483551e-06,
+      "loss": 0.943,
+      "step": 20806
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.724737750934481e-06,
+      "loss": 1.0367,
+      "step": 20807
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7242248651020844e-06,
+      "loss": 0.9729,
+      "step": 20808
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7237121539960246e-06,
+      "loss": 0.9928,
+      "step": 20809
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.723199617625958e-06,
+      "loss": 0.9899,
+      "step": 20810
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7226872560015417e-06,
+      "loss": 0.9144,
+      "step": 20811
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7221750691324248e-06,
+      "loss": 0.9481,
+      "step": 20812
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.721663057028257e-06,
+      "loss": 1.0029,
+      "step": 20813
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7211512196986803e-06,
+      "loss": 1.1106,
+      "step": 20814
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7206395571533397e-06,
+      "loss": 0.9603,
+      "step": 20815
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7201280694018706e-06,
+      "loss": 0.9934,
+      "step": 20816
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7196167564539093e-06,
+      "loss": 1.0226,
+      "step": 20817
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.719105618319087e-06,
+      "loss": 0.9528,
+      "step": 20818
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.718594655007031e-06,
+      "loss": 0.9203,
+      "step": 20819
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.718083866527367e-06,
+      "loss": 0.8649,
+      "step": 20820
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7175732528897158e-06,
+      "loss": 0.9834,
+      "step": 20821
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7170628141036982e-06,
+      "loss": 1.0437,
+      "step": 20822
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7165525501789272e-06,
+      "loss": 0.8732,
+      "step": 20823
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7160424611250137e-06,
+      "loss": 0.9343,
+      "step": 20824
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7155325469515684e-06,
+      "loss": 1.069,
+      "step": 20825
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.715022807668195e-06,
+      "loss": 0.9686,
+      "step": 20826
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.714513243284497e-06,
+      "loss": 0.9838,
+      "step": 20827
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.714003853810071e-06,
+      "loss": 1.0123,
+      "step": 20828
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7134946392545127e-06,
+      "loss": 0.9733,
+      "step": 20829
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7129855996274147e-06,
+      "loss": 0.9354,
+      "step": 20830
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7124767349383664e-06,
+      "loss": 0.9536,
+      "step": 20831
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.711968045196952e-06,
+      "loss": 0.9915,
+      "step": 20832
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7114595304127534e-06,
+      "loss": 0.9604,
+      "step": 20833
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7109511905953504e-06,
+      "loss": 0.9368,
+      "step": 20834
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.710443025754318e-06,
+      "loss": 1.02,
+      "step": 20835
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.709935035899228e-06,
+      "loss": 0.9576,
+      "step": 20836
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7094272210396503e-06,
+      "loss": 0.969,
+      "step": 20837
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.70891958118515e-06,
+      "loss": 1.0073,
+      "step": 20838
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7084121163452893e-06,
+      "loss": 0.9367,
+      "step": 20839
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7079048265296284e-06,
+      "loss": 1.1019,
+      "step": 20840
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7073977117477213e-06,
+      "loss": 1.0708,
+      "step": 20841
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.706890772009121e-06,
+      "loss": 0.9774,
+      "step": 20842
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7063840073233787e-06,
+      "loss": 1.0253,
+      "step": 20843
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.705877417700035e-06,
+      "loss": 0.8826,
+      "step": 20844
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.705371003148638e-06,
+      "loss": 0.9965,
+      "step": 20845
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.704864763678724e-06,
+      "loss": 0.973,
+      "step": 20846
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.704358699299832e-06,
+      "loss": 0.9888,
+      "step": 20847
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.703852810021489e-06,
+      "loss": 1.0615,
+      "step": 20848
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7033470958532293e-06,
+      "loss": 1.032,
+      "step": 20849
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.70284155680458e-06,
+      "loss": 0.9216,
+      "step": 20850
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.702336192885058e-06,
+      "loss": 0.9667,
+      "step": 20851
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7018310041041854e-06,
+      "loss": 0.9832,
+      "step": 20852
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.701325990471481e-06,
+      "loss": 1.0022,
+      "step": 20853
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7008211519964557e-06,
+      "loss": 0.9463,
+      "step": 20854
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.700316488688619e-06,
+      "loss": 0.9083,
+      "step": 20855
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.699812000557475e-06,
+      "loss": 0.9754,
+      "step": 20856
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6993076876125304e-06,
+      "loss": 0.989,
+      "step": 20857
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.698803549863285e-06,
+      "loss": 0.9598,
+      "step": 20858
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6982995873192323e-06,
+      "loss": 1.0802,
+      "step": 20859
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6977957999898642e-06,
+      "loss": 1.0304,
+      "step": 20860
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.697292187884677e-06,
+      "loss": 0.9137,
+      "step": 20861
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6967887510131507e-06,
+      "loss": 0.9712,
+      "step": 20862
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.696285489384769e-06,
+      "loss": 0.9988,
+      "step": 20863
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6957824030090166e-06,
+      "loss": 0.9953,
+      "step": 20864
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6952794918953664e-06,
+      "loss": 0.9423,
+      "step": 20865
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6947767560532923e-06,
+      "loss": 0.9573,
+      "step": 20866
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6942741954922615e-06,
+      "loss": 0.956,
+      "step": 20867
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.693771810221746e-06,
+      "loss": 0.9868,
+      "step": 20868
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.693269600251207e-06,
+      "loss": 0.9704,
+      "step": 20869
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.692767565590103e-06,
+      "loss": 1.0066,
+      "step": 20870
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6922657062478896e-06,
+      "loss": 0.99,
+      "step": 20871
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.691764022234027e-06,
+      "loss": 0.9253,
+      "step": 20872
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6912625135579586e-06,
+      "loss": 0.9906,
+      "step": 20873
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6907611802291325e-06,
+      "loss": 1.0114,
+      "step": 20874
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.690260022256993e-06,
+      "loss": 1.0126,
+      "step": 20875
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6897590396509823e-06,
+      "loss": 1.0094,
+      "step": 20876
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6892582324205353e-06,
+      "loss": 1.0033,
+      "step": 20877
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.688757600575085e-06,
+      "loss": 0.9941,
+      "step": 20878
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6882571441240612e-06,
+      "loss": 0.9975,
+      "step": 20879
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.687756863076896e-06,
+      "loss": 0.908,
+      "step": 20880
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6872567574430065e-06,
+      "loss": 0.9527,
+      "step": 20881
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.686756827231817e-06,
+      "loss": 1.0375,
+      "step": 20882
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.686257072452744e-06,
+      "loss": 0.9546,
+      "step": 20883
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6857574931152e-06,
+      "loss": 0.9285,
+      "step": 20884
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6852580892285975e-06,
+      "loss": 1.0118,
+      "step": 20885
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6847588608023413e-06,
+      "loss": 0.945,
+      "step": 20886
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6842598078458372e-06,
+      "loss": 0.9696,
+      "step": 20887
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6837609303684856e-06,
+      "loss": 0.9437,
+      "step": 20888
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.683262228379684e-06,
+      "loss": 0.9939,
+      "step": 20889
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.682763701888824e-06,
+      "loss": 0.9834,
+      "step": 20890
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6822653509053e-06,
+      "loss": 0.9887,
+      "step": 20891
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6817671754384958e-06,
+      "loss": 0.9708,
+      "step": 20892
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.681269175497797e-06,
+      "loss": 0.9669,
+      "step": 20893
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.680771351092586e-06,
+      "loss": 0.9969,
+      "step": 20894
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.680273702232237e-06,
+      "loss": 1.0411,
+      "step": 20895
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6797762289261267e-06,
+      "loss": 0.9995,
+      "step": 20896
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6792789311836247e-06,
+      "loss": 0.9371,
+      "step": 20897
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6787818090140985e-06,
+      "loss": 0.9067,
+      "step": 20898
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6782848624269133e-06,
+      "loss": 0.9933,
+      "step": 20899
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6777880914314296e-06,
+      "loss": 0.9717,
+      "step": 20900
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6772914960370037e-06,
+      "loss": 1.0098,
+      "step": 20901
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.676795076252991e-06,
+      "loss": 0.9465,
+      "step": 20902
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.676298832088743e-06,
+      "loss": 1.0175,
+      "step": 20903
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6758027635536057e-06,
+      "loss": 0.9738,
+      "step": 20904
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6753068706569255e-06,
+      "loss": 1.0107,
+      "step": 20905
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6748111534080424e-06,
+      "loss": 0.9564,
+      "step": 20906
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.674315611816294e-06,
+      "loss": 1.0281,
+      "step": 20907
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6738202458910158e-06,
+      "loss": 1.0312,
+      "step": 20908
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.673325055641537e-06,
+      "loss": 0.92,
+      "step": 20909
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.672830041077187e-06,
+      "loss": 1.018,
+      "step": 20910
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6723352022072904e-06,
+      "loss": 0.968,
+      "step": 20911
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6718405390411677e-06,
+      "loss": 1.0088,
+      "step": 20912
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6713460515881365e-06,
+      "loss": 0.9974,
+      "step": 20913
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.670851739857512e-06,
+      "loss": 1.0026,
+      "step": 20914
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6703576038586046e-06,
+      "loss": 0.9904,
+      "step": 20915
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6698636436007255e-06,
+      "loss": 0.9625,
+      "step": 20916
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.669369859093173e-06,
+      "loss": 0.9768,
+      "step": 20917
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.668876250345255e-06,
+      "loss": 0.9781,
+      "step": 20918
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6683828173662655e-06,
+      "loss": 1.0098,
+      "step": 20919
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.667889560165501e-06,
+      "loss": 0.9821,
+      "step": 20920
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6673964787522523e-06,
+      "loss": 0.953,
+      "step": 20921
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6669035731358073e-06,
+      "loss": 0.9871,
+      "step": 20922
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6664108433254534e-06,
+      "loss": 0.9754,
+      "step": 20923
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6659182893304673e-06,
+      "loss": 0.9547,
+      "step": 20924
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6654259111601303e-06,
+      "loss": 0.9799,
+      "step": 20925
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6649337088237162e-06,
+      "loss": 0.9392,
+      "step": 20926
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6644416823304988e-06,
+      "loss": 1.0236,
+      "step": 20927
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.663949831689742e-06,
+      "loss": 0.9412,
+      "step": 20928
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.663458156910713e-06,
+      "loss": 0.8949,
+      "step": 20929
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.662966658002675e-06,
+      "loss": 1.0115,
+      "step": 20930
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6624753349748867e-06,
+      "loss": 0.9137,
+      "step": 20931
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.661984187836597e-06,
+      "loss": 0.9889,
+      "step": 20932
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6614932165970644e-06,
+      "loss": 0.971,
+      "step": 20933
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.661002421265536e-06,
+      "loss": 1.0389,
+      "step": 20934
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6605118018512534e-06,
+      "loss": 0.9675,
+      "step": 20935
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6600213583634594e-06,
+      "loss": 0.9704,
+      "step": 20936
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.659531090811395e-06,
+      "loss": 0.9868,
+      "step": 20937
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6590409992042955e-06,
+      "loss": 0.9928,
+      "step": 20938
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6585510835513888e-06,
+      "loss": 0.9219,
+      "step": 20939
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6580613438619044e-06,
+      "loss": 1.0445,
+      "step": 20940
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6575717801450687e-06,
+      "loss": 1.0401,
+      "step": 20941
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6570823924101067e-06,
+      "loss": 0.934,
+      "step": 20942
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6565931806662307e-06,
+      "loss": 1.0327,
+      "step": 20943
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.656104144922658e-06,
+      "loss": 1.0574,
+      "step": 20944
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6556152851886045e-06,
+      "loss": 1.0087,
+      "step": 20945
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6551266014732738e-06,
+      "loss": 1.0148,
+      "step": 20946
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.654638093785872e-06,
+      "loss": 1.0433,
+      "step": 20947
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6541497621356007e-06,
+      "loss": 1.0206,
+      "step": 20948
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6536616065316635e-06,
+      "loss": 1.058,
+      "step": 20949
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.65317362698325e-06,
+      "loss": 0.929,
+      "step": 20950
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6526858234995537e-06,
+      "loss": 0.9887,
+      "step": 20951
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.652198196089764e-06,
+      "loss": 0.9826,
+      "step": 20952
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.651710744763068e-06,
+      "loss": 1.0067,
+      "step": 20953
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6512234695286447e-06,
+      "loss": 1.0134,
+      "step": 20954
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6507363703956714e-06,
+      "loss": 0.9639,
+      "step": 20955
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.650249447373329e-06,
+      "loss": 1.0272,
+      "step": 20956
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6497627004707862e-06,
+      "loss": 0.9738,
+      "step": 20957
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6492761296972118e-06,
+      "loss": 0.9938,
+      "step": 20958
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6487897350617687e-06,
+      "loss": 0.9588,
+      "step": 20959
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6483035165736274e-06,
+      "loss": 1.0238,
+      "step": 20960
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6478174742419384e-06,
+      "loss": 0.9899,
+      "step": 20961
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6473316080758594e-06,
+      "loss": 1.0081,
+      "step": 20962
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.646845918084543e-06,
+      "loss": 0.9998,
+      "step": 20963
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.646360404277141e-06,
+      "loss": 1.0531,
+      "step": 20964
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6458750666627943e-06,
+      "loss": 0.9725,
+      "step": 20965
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6453899052506483e-06,
+      "loss": 0.9861,
+      "step": 20966
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.64490492004984e-06,
+      "loss": 0.964,
+      "step": 20967
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6444201110695057e-06,
+      "loss": 1.0086,
+      "step": 20968
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.643935478318778e-06,
+      "loss": 1.0414,
+      "step": 20969
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6434510218067848e-06,
+      "loss": 0.9639,
+      "step": 20970
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6429667415426532e-06,
+      "loss": 0.9834,
+      "step": 20971
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.642482637535504e-06,
+      "loss": 0.9647,
+      "step": 20972
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6419987097944578e-06,
+      "loss": 0.9299,
+      "step": 20973
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6415149583286296e-06,
+      "loss": 0.9851,
+      "step": 20974
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6410313831471315e-06,
+      "loss": 0.9249,
+      "step": 20975
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.640547984259072e-06,
+      "loss": 0.9942,
+      "step": 20976
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.640064761673558e-06,
+      "loss": 0.9866,
+      "step": 20977
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6395817153996907e-06,
+      "loss": 0.9699,
+      "step": 20978
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.639098845446571e-06,
+      "loss": 1.0635,
+      "step": 20979
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6386161518232926e-06,
+      "loss": 0.9968,
+      "step": 20980
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.638133634538949e-06,
+      "loss": 1.01,
+      "step": 20981
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.637651293602628e-06,
+      "loss": 0.9871,
+      "step": 20982
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.637169129023417e-06,
+      "loss": 0.9984,
+      "step": 20983
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6366871408103992e-06,
+      "loss": 1.0587,
+      "step": 20984
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.636205328972651e-06,
+      "loss": 0.9348,
+      "step": 20985
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.635723693519251e-06,
+      "loss": 1.013,
+      "step": 20986
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.635242234459269e-06,
+      "loss": 0.9459,
+      "step": 20987
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.634760951801776e-06,
+      "loss": 1.0172,
+      "step": 20988
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6342798455558382e-06,
+      "loss": 0.957,
+      "step": 20989
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.633798915730517e-06,
+      "loss": 0.9865,
+      "step": 20990
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6333181623348714e-06,
+      "loss": 1.0341,
+      "step": 20991
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.632837585377958e-06,
+      "loss": 0.9771,
+      "step": 20992
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.632357184868829e-06,
+      "loss": 1.0188,
+      "step": 20993
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.631876960816533e-06,
+      "loss": 0.9914,
+      "step": 20994
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6313969132301177e-06,
+      "loss": 1.0295,
+      "step": 20995
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6309170421186247e-06,
+      "loss": 1.0355,
+      "step": 20996
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.630437347491093e-06,
+      "loss": 0.9936,
+      "step": 20997
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.629957829356559e-06,
+      "loss": 0.9718,
+      "step": 20998
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6294784877240543e-06,
+      "loss": 0.9945,
+      "step": 20999
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6289993226026115e-06,
+      "loss": 0.9981,
+      "step": 21000
+    },
+    {
+      "epoch": 0.89,
+      "eval_text_loss": 0.7550748586654663,
+      "eval_text_runtime": 1.4155,
+      "eval_text_samples_per_second": 52.279,
+      "eval_text_steps_per_second": 1.413,
+      "step": 21000
+    },
+    {
+      "epoch": 0.89,
+      "eval_svit_loss": 1.074927568435669,
+      "eval_svit_runtime": 4.9925,
+      "eval_svit_samples_per_second": 136.404,
+      "eval_svit_steps_per_second": 2.203,
+      "step": 21000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.62852033400125e-06,
+      "loss": 1.0137,
+      "step": 21001
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.628041521928998e-06,
+      "loss": 0.9976,
+      "step": 21002
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6275628863948744e-06,
+      "loss": 0.9718,
+      "step": 21003
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.627084427407895e-06,
+      "loss": 1.026,
+      "step": 21004
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6266061449770676e-06,
+      "loss": 0.9426,
+      "step": 21005
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6261280391114076e-06,
+      "loss": 0.9516,
+      "step": 21006
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6256501098199197e-06,
+      "loss": 0.959,
+      "step": 21007
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.625172357111603e-06,
+      "loss": 0.9678,
+      "step": 21008
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6246947809954617e-06,
+      "loss": 0.9691,
+      "step": 21009
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6242173814804897e-06,
+      "loss": 0.9957,
+      "step": 21010
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6237401585756806e-06,
+      "loss": 0.9918,
+      "step": 21011
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.62326311229002e-06,
+      "loss": 0.9171,
+      "step": 21012
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.622786242632498e-06,
+      "loss": 1.045,
+      "step": 21013
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6223095496120966e-06,
+      "loss": 0.942,
+      "step": 21014
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.621833033237795e-06,
+      "loss": 0.9773,
+      "step": 21015
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6213566935185675e-06,
+      "loss": 0.9157,
+      "step": 21016
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.620880530463389e-06,
+      "loss": 0.9567,
+      "step": 21017
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.62040454408123e-06,
+      "loss": 0.9275,
+      "step": 21018
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6199287343810534e-06,
+      "loss": 0.9826,
+      "step": 21019
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6194531013718217e-06,
+      "loss": 0.9209,
+      "step": 21020
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6189776450624983e-06,
+      "loss": 1.0242,
+      "step": 21021
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6185023654620383e-06,
+      "loss": 0.9946,
+      "step": 21022
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.61802726257939e-06,
+      "loss": 0.9465,
+      "step": 21023
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.617552336423506e-06,
+      "loss": 1.0374,
+      "step": 21024
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.617077587003333e-06,
+      "loss": 0.9107,
+      "step": 21025
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6166030143278155e-06,
+      "loss": 0.9789,
+      "step": 21026
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.616128618405889e-06,
+      "loss": 0.9473,
+      "step": 21027
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.615654399246489e-06,
+      "loss": 1.0179,
+      "step": 21028
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6151803568585546e-06,
+      "loss": 0.918,
+      "step": 21029
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6147064912510094e-06,
+      "loss": 0.9776,
+      "step": 21030
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.61423280243278e-06,
+      "loss": 0.9286,
+      "step": 21031
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6137592904127907e-06,
+      "loss": 1.058,
+      "step": 21032
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6132859551999637e-06,
+      "loss": 0.9395,
+      "step": 21033
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6128127968032105e-06,
+      "loss": 0.9343,
+      "step": 21034
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.612339815231445e-06,
+      "loss": 1.0047,
+      "step": 21035
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.611867010493576e-06,
+      "loss": 1.0013,
+      "step": 21036
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6113943825985146e-06,
+      "loss": 0.9642,
+      "step": 21037
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6109219315551584e-06,
+      "loss": 0.9438,
+      "step": 21038
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.610449657372407e-06,
+      "loss": 0.9468,
+      "step": 21039
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6099775600591605e-06,
+      "loss": 0.9523,
+      "step": 21040
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.609505639624309e-06,
+      "loss": 0.9314,
+      "step": 21041
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6090338960767417e-06,
+      "loss": 0.8767,
+      "step": 21042
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6085623294253426e-06,
+      "loss": 1.0095,
+      "step": 21043
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6080909396790023e-06,
+      "loss": 1.0552,
+      "step": 21044
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.607619726846592e-06,
+      "loss": 1.0374,
+      "step": 21045
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6071486909369927e-06,
+      "loss": 1.0083,
+      "step": 21046
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6066778319590727e-06,
+      "loss": 0.99,
+      "step": 21047
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.606207149921708e-06,
+      "loss": 0.9711,
+      "step": 21048
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6057366448337606e-06,
+      "loss": 0.9559,
+      "step": 21049
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.605266316704093e-06,
+      "loss": 1.0104,
+      "step": 21050
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6047961655415658e-06,
+      "loss": 0.9998,
+      "step": 21051
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6043261913550347e-06,
+      "loss": 0.9779,
+      "step": 21052
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.603856394153354e-06,
+      "loss": 0.9801,
+      "step": 21053
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.60338677394537e-06,
+      "loss": 0.9976,
+      "step": 21054
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6029173307399325e-06,
+      "loss": 0.9451,
+      "step": 21055
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.602448064545883e-06,
+      "loss": 0.9585,
+      "step": 21056
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6019789753720595e-06,
+      "loss": 0.9865,
+      "step": 21057
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6015100632272998e-06,
+      "loss": 0.9724,
+      "step": 21058
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6010413281204364e-06,
+      "loss": 1.0747,
+      "step": 21059
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6005727700602993e-06,
+      "loss": 0.9139,
+      "step": 21060
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.600104389055713e-06,
+      "loss": 1.0214,
+      "step": 21061
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5996361851155018e-06,
+      "loss": 0.9844,
+      "step": 21062
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.599168158248485e-06,
+      "loss": 0.9446,
+      "step": 21063
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5987003084634787e-06,
+      "loss": 0.962,
+      "step": 21064
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5982326357692947e-06,
+      "loss": 0.9904,
+      "step": 21065
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.597765140174744e-06,
+      "loss": 0.923,
+      "step": 21066
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5972978216886316e-06,
+      "loss": 0.9814,
+      "step": 21067
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.596830680319761e-06,
+      "loss": 1.0934,
+      "step": 21068
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.596363716076932e-06,
+      "loss": 1.0268,
+      "step": 21069
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5958969289689386e-06,
+      "loss": 1.0284,
+      "step": 21070
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5954303190045762e-06,
+      "loss": 0.926,
+      "step": 21071
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5949638861926333e-06,
+      "loss": 0.9871,
+      "step": 21072
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5944976305418957e-06,
+      "loss": 1.007,
+      "step": 21073
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.594031552061146e-06,
+      "loss": 0.9158,
+      "step": 21074
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.593565650759165e-06,
+      "loss": 0.9414,
+      "step": 21075
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5930999266447278e-06,
+      "loss": 1.0311,
+      "step": 21076
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5926343797266063e-06,
+      "loss": 1.0382,
+      "step": 21077
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.592169010013571e-06,
+      "loss": 1.0358,
+      "step": 21078
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5917038175143886e-06,
+      "loss": 1.0092,
+      "step": 21079
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5912388022378212e-06,
+      "loss": 0.9348,
+      "step": 21080
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.590773964192628e-06,
+      "loss": 0.9568,
+      "step": 21081
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.590309303387565e-06,
+      "loss": 1.0513,
+      "step": 21082
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5898448198313842e-06,
+      "loss": 0.9938,
+      "step": 21083
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.589380513532839e-06,
+      "loss": 0.9803,
+      "step": 21084
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5889163845006694e-06,
+      "loss": 0.9865,
+      "step": 21085
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5884524327436213e-06,
+      "loss": 0.9633,
+      "step": 21086
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5879886582704354e-06,
+      "loss": 0.9193,
+      "step": 21087
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5875250610898473e-06,
+      "loss": 1.0,
+      "step": 21088
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5870616412105854e-06,
+      "loss": 0.9198,
+      "step": 21089
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5865983986413846e-06,
+      "loss": 1.0308,
+      "step": 21090
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.586135333390969e-06,
+      "loss": 1.0383,
+      "step": 21091
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5856724454680606e-06,
+      "loss": 0.9484,
+      "step": 21092
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.585209734881378e-06,
+      "loss": 0.9453,
+      "step": 21093
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.584747201639639e-06,
+      "loss": 1.0039,
+      "step": 21094
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.584284845751557e-06,
+      "loss": 1.0679,
+      "step": 21095
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5838226672258375e-06,
+      "loss": 0.9379,
+      "step": 21096
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.583360666071191e-06,
+      "loss": 0.9713,
+      "step": 21097
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.582898842296317e-06,
+      "loss": 1.0029,
+      "step": 21098
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.582437195909917e-06,
+      "loss": 0.9818,
+      "step": 21099
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5819757269206836e-06,
+      "loss": 0.9559,
+      "step": 21100
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5815144353373136e-06,
+      "loss": 1.0496,
+      "step": 21101
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5810533211684953e-06,
+      "loss": 1.002,
+      "step": 21102
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5805923844229123e-06,
+      "loss": 0.9813,
+      "step": 21103
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.580131625109247e-06,
+      "loss": 0.924,
+      "step": 21104
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5796710432361817e-06,
+      "loss": 0.9035,
+      "step": 21105
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.579210638812393e-06,
+      "loss": 0.9778,
+      "step": 21106
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5787504118465493e-06,
+      "loss": 1.0392,
+      "step": 21107
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.57829036234732e-06,
+      "loss": 0.9667,
+      "step": 21108
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5778304903233743e-06,
+      "loss": 1.0898,
+      "step": 21109
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.577370795783375e-06,
+      "loss": 0.9495,
+      "step": 21110
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5769112787359788e-06,
+      "loss": 1.0437,
+      "step": 21111
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.576451939189839e-06,
+      "loss": 1.0005,
+      "step": 21112
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5759927771536166e-06,
+      "loss": 0.923,
+      "step": 21113
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5755337926359527e-06,
+      "loss": 1.0003,
+      "step": 21114
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.575074985645496e-06,
+      "loss": 0.9922,
+      "step": 21115
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5746163561908877e-06,
+      "loss": 0.9686,
+      "step": 21116
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5741579042807707e-06,
+      "loss": 0.95,
+      "step": 21117
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5736996299237774e-06,
+      "loss": 1.0284,
+      "step": 21118
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.573241533128541e-06,
+      "loss": 1.0081,
+      "step": 21119
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.57278361390369e-06,
+      "loss": 1.0266,
+      "step": 21120
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5723258722578527e-06,
+      "loss": 1.0081,
+      "step": 21121
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5718683081996488e-06,
+      "loss": 1.0053,
+      "step": 21122
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5714109217376985e-06,
+      "loss": 0.9601,
+      "step": 21123
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5709537128806164e-06,
+      "loss": 0.9121,
+      "step": 21124
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.570496681637016e-06,
+      "loss": 0.9695,
+      "step": 21125
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5700398280155062e-06,
+      "loss": 1.0117,
+      "step": 21126
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.569583152024691e-06,
+      "loss": 0.9344,
+      "step": 21127
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5691266536731774e-06,
+      "loss": 0.9967,
+      "step": 21128
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5686703329695597e-06,
+      "loss": 1.0111,
+      "step": 21129
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5682141899224345e-06,
+      "loss": 0.9567,
+      "step": 21130
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5677582245403946e-06,
+      "loss": 1.0329,
+      "step": 21131
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.567302436832031e-06,
+      "loss": 1.0008,
+      "step": 21132
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5668468268059256e-06,
+      "loss": 0.9705,
+      "step": 21133
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.566391394470663e-06,
+      "loss": 1.0414,
+      "step": 21134
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5659361398348215e-06,
+      "loss": 0.9435,
+      "step": 21135
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.565481062906976e-06,
+      "loss": 0.9895,
+      "step": 21136
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.565026163695701e-06,
+      "loss": 0.9719,
+      "step": 21137
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5645714422095627e-06,
+      "loss": 0.9097,
+      "step": 21138
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5641168984571274e-06,
+      "loss": 0.9498,
+      "step": 21139
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.563662532446958e-06,
+      "loss": 1.0272,
+      "step": 21140
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5632083441876124e-06,
+      "loss": 0.9932,
+      "step": 21141
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5627543336876473e-06,
+      "loss": 1.0594,
+      "step": 21142
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.562300500955614e-06,
+      "loss": 0.9574,
+      "step": 21143
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5618468460000602e-06,
+      "loss": 1.0374,
+      "step": 21144
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5613933688295335e-06,
+      "loss": 1.0209,
+      "step": 21145
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5609400694525755e-06,
+      "loss": 1.0077,
+      "step": 21146
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.560486947877723e-06,
+      "loss": 0.9597,
+      "step": 21147
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.560034004113513e-06,
+      "loss": 1.015,
+      "step": 21148
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.559581238168478e-06,
+      "loss": 0.9837,
+      "step": 21149
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.559128650051146e-06,
+      "loss": 0.946,
+      "step": 21150
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5586762397700415e-06,
+      "loss": 0.9858,
+      "step": 21151
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5582240073336885e-06,
+      "loss": 1.0814,
+      "step": 21152
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.557771952750603e-06,
+      "loss": 1.0545,
+      "step": 21153
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5573200760293027e-06,
+      "loss": 1.0094,
+      "step": 21154
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5568683771782987e-06,
+      "loss": 1.0131,
+      "step": 21155
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5564168562060987e-06,
+      "loss": 0.95,
+      "step": 21156
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.555965513121209e-06,
+      "loss": 0.9916,
+      "step": 21157
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.555514347932131e-06,
+      "loss": 0.9758,
+      "step": 21158
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.555063360647363e-06,
+      "loss": 1.0941,
+      "step": 21159
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5546125512754017e-06,
+      "loss": 0.9604,
+      "step": 21160
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.554161919824737e-06,
+      "loss": 0.9273,
+      "step": 21161
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.553711466303858e-06,
+      "loss": 1.0061,
+      "step": 21162
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.55326119072125e-06,
+      "loss": 0.9805,
+      "step": 21163
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5528110930853947e-06,
+      "loss": 0.9685,
+      "step": 21164
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5523611734047708e-06,
+      "loss": 0.9723,
+      "step": 21165
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5519114316878536e-06,
+      "loss": 1.0011,
+      "step": 21166
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.551461867943114e-06,
+      "loss": 1.0528,
+      "step": 21167
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.551012482179021e-06,
+      "loss": 1.0207,
+      "step": 21168
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.550563274404038e-06,
+      "loss": 0.9584,
+      "step": 21169
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.550114244626628e-06,
+      "loss": 0.9944,
+      "step": 21170
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5496653928552497e-06,
+      "loss": 0.9741,
+      "step": 21171
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.54921671909836e-06,
+      "loss": 0.9787,
+      "step": 21172
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5487682233644043e-06,
+      "loss": 0.9686,
+      "step": 21173
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.548319905661836e-06,
+      "loss": 0.9511,
+      "step": 21174
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5478717659991005e-06,
+      "loss": 0.9442,
+      "step": 21175
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5474238043846367e-06,
+      "loss": 0.9592,
+      "step": 21176
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5469760208268814e-06,
+      "loss": 0.9939,
+      "step": 21177
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.546528415334273e-06,
+      "loss": 0.9411,
+      "step": 21178
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5460809879152426e-06,
+      "loss": 0.9853,
+      "step": 21179
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5456337385782173e-06,
+      "loss": 1.0627,
+      "step": 21180
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5451866673316183e-06,
+      "loss": 0.9654,
+      "step": 21181
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.544739774183873e-06,
+      "loss": 0.9794,
+      "step": 21182
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.544293059143399e-06,
+      "loss": 0.9676,
+      "step": 21183
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.543846522218606e-06,
+      "loss": 0.8972,
+      "step": 21184
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.54340016341791e-06,
+      "loss": 0.888,
+      "step": 21185
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.542953982749719e-06,
+      "loss": 1.0276,
+      "step": 21186
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.542507980222433e-06,
+      "loss": 0.9262,
+      "step": 21187
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.542062155844456e-06,
+      "loss": 1.0215,
+      "step": 21188
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5416165096241875e-06,
+      "loss": 0.9893,
+      "step": 21189
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.541171041570022e-06,
+      "loss": 0.9593,
+      "step": 21190
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5407257516903484e-06,
+      "loss": 0.952,
+      "step": 21191
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.540280639993554e-06,
+      "loss": 0.9715,
+      "step": 21192
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5398357064880264e-06,
+      "loss": 0.9765,
+      "step": 21193
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.539390951182147e-06,
+      "loss": 1.0228,
+      "step": 21194
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.53894637408429e-06,
+      "loss": 1.0059,
+      "step": 21195
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.538501975202831e-06,
+      "loss": 0.975,
+      "step": 21196
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.538057754546144e-06,
+      "loss": 0.9788,
+      "step": 21197
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5376137121225934e-06,
+      "loss": 0.9864,
+      "step": 21198
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.537169847940545e-06,
+      "loss": 1.007,
+      "step": 21199
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5367261620083573e-06,
+      "loss": 0.9655,
+      "step": 21200
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5362826543343935e-06,
+      "loss": 0.9704,
+      "step": 21201
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.535839324927002e-06,
+      "loss": 0.9853,
+      "step": 21202
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.535396173794538e-06,
+      "loss": 0.9335,
+      "step": 21203
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5349532009453438e-06,
+      "loss": 1.0202,
+      "step": 21204
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5345104063877716e-06,
+      "loss": 0.9665,
+      "step": 21205
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.534067790130156e-06,
+      "loss": 1.0371,
+      "step": 21206
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5336253521808354e-06,
+      "loss": 1.0204,
+      "step": 21207
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5331830925481453e-06,
+      "loss": 0.9665,
+      "step": 21208
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.532741011240416e-06,
+      "loss": 0.9683,
+      "step": 21209
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5322991082659743e-06,
+      "loss": 1.0358,
+      "step": 21210
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5318573836331465e-06,
+      "loss": 0.9587,
+      "step": 21211
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5314158373502496e-06,
+      "loss": 0.92,
+      "step": 21212
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.530974469425605e-06,
+      "loss": 0.9326,
+      "step": 21213
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.530533279867522e-06,
+      "loss": 0.9857,
+      "step": 21214
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5300922686843146e-06,
+      "loss": 0.9979,
+      "step": 21215
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.529651435884292e-06,
+      "loss": 1.013,
+      "step": 21216
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.529210781475753e-06,
+      "loss": 0.9663,
+      "step": 21217
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.528770305467001e-06,
+      "loss": 0.9421,
+      "step": 21218
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.528330007866333e-06,
+      "loss": 0.9873,
+      "step": 21219
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5278898886820425e-06,
+      "loss": 1.0798,
+      "step": 21220
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5274499479224195e-06,
+      "loss": 1.0037,
+      "step": 21221
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.527010185595752e-06,
+      "loss": 0.9457,
+      "step": 21222
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.526570601710323e-06,
+      "loss": 1.026,
+      "step": 21223
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.526131196274414e-06,
+      "loss": 0.937,
+      "step": 21224
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5256919692963e-06,
+      "loss": 0.9565,
+      "step": 21225
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5252529207842573e-06,
+      "loss": 0.9874,
+      "step": 21226
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5248140507465535e-06,
+      "loss": 0.9608,
+      "step": 21227
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.524375359191458e-06,
+      "loss": 0.938,
+      "step": 21228
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.523936846127232e-06,
+      "loss": 1.045,
+      "step": 21229
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.523498511562137e-06,
+      "loss": 0.948,
+      "step": 21230
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.52306035550443e-06,
+      "loss": 0.9203,
+      "step": 21231
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.522622377962365e-06,
+      "loss": 0.9563,
+      "step": 21232
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.52218457894419e-06,
+      "loss": 0.9914,
+      "step": 21233
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5217469584581536e-06,
+      "loss": 1.027,
+      "step": 21234
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.521309516512499e-06,
+      "loss": 1.0269,
+      "step": 21235
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5208722531154653e-06,
+      "loss": 0.9206,
+      "step": 21236
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5204351682752892e-06,
+      "loss": 0.9438,
+      "step": 21237
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.519998262000206e-06,
+      "loss": 1.0149,
+      "step": 21238
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5195615342984427e-06,
+      "loss": 0.9549,
+      "step": 21239
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5191249851782285e-06,
+      "loss": 0.9711,
+      "step": 21240
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.518688614647785e-06,
+      "loss": 0.9698,
+      "step": 21241
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.518252422715333e-06,
+      "loss": 0.984,
+      "step": 21242
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5178164093890867e-06,
+      "loss": 0.9896,
+      "step": 21243
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.517380574677263e-06,
+      "loss": 0.972,
+      "step": 21244
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.516944918588068e-06,
+      "loss": 1.0017,
+      "step": 21245
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.516509441129711e-06,
+      "loss": 0.9945,
+      "step": 21246
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.516074142310393e-06,
+      "loss": 0.9152,
+      "step": 21247
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5156390221383145e-06,
+      "loss": 0.9143,
+      "step": 21248
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.515204080621672e-06,
+      "loss": 0.9668,
+      "step": 21249
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5147693177686583e-06,
+      "loss": 0.9499,
+      "step": 21250
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5143347335874626e-06,
+      "loss": 0.968,
+      "step": 21251
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.513900328086272e-06,
+      "loss": 0.9267,
+      "step": 21252
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5134661012732667e-06,
+      "loss": 0.9373,
+      "step": 21253
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.513032053156628e-06,
+      "loss": 1.0089,
+      "step": 21254
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5125981837445337e-06,
+      "loss": 0.9158,
+      "step": 21255
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.512164493045155e-06,
+      "loss": 0.9461,
+      "step": 21256
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5117309810666594e-06,
+      "loss": 0.9259,
+      "step": 21257
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.511297647817216e-06,
+      "loss": 1.0,
+      "step": 21258
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5108644933049867e-06,
+      "loss": 1.052,
+      "step": 21259
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.510431517538129e-06,
+      "loss": 0.958,
+      "step": 21260
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.509998720524799e-06,
+      "loss": 1.0077,
+      "step": 21261
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5095661022731513e-06,
+      "loss": 0.9445,
+      "step": 21262
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5091336627913344e-06,
+      "loss": 1.0373,
+      "step": 21263
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5087014020874934e-06,
+      "loss": 1.0057,
+      "step": 21264
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.50826932016977e-06,
+      "loss": 0.977,
+      "step": 21265
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.507837417046304e-06,
+      "loss": 0.9958,
+      "step": 21266
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5074056927252337e-06,
+      "loss": 0.9654,
+      "step": 21267
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.506974147214687e-06,
+      "loss": 0.9496,
+      "step": 21268
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.506542780522794e-06,
+      "loss": 1.0095,
+      "step": 21269
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.506111592657684e-06,
+      "loss": 0.9194,
+      "step": 21270
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5056805836274756e-06,
+      "loss": 0.916,
+      "step": 21271
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5052497534402866e-06,
+      "loss": 0.9649,
+      "step": 21272
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.504819102104235e-06,
+      "loss": 1.0025,
+      "step": 21273
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.504388629627435e-06,
+      "loss": 0.9467,
+      "step": 21274
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.50395833601799e-06,
+      "loss": 0.9832,
+      "step": 21275
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5035282212840064e-06,
+      "loss": 1.0464,
+      "step": 21276
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5030982854335896e-06,
+      "loss": 1.0509,
+      "step": 21277
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5026685284748377e-06,
+      "loss": 1.0018,
+      "step": 21278
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.502238950415843e-06,
+      "loss": 1.0086,
+      "step": 21279
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.501809551264698e-06,
+      "loss": 1.0234,
+      "step": 21280
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.501380331029495e-06,
+      "loss": 1.043,
+      "step": 21281
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5009512897183154e-06,
+      "loss": 0.9688,
+      "step": 21282
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5005224273392415e-06,
+      "loss": 0.9788,
+      "step": 21283
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.500093743900352e-06,
+      "loss": 0.9211,
+      "step": 21284
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.499665239409723e-06,
+      "loss": 1.0386,
+      "step": 21285
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.499236913875426e-06,
+      "loss": 1.0179,
+      "step": 21286
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4988087673055276e-06,
+      "loss": 1.0209,
+      "step": 21287
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4983807997080924e-06,
+      "loss": 0.946,
+      "step": 21288
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4979530110911856e-06,
+      "loss": 0.9699,
+      "step": 21289
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4975254014628628e-06,
+      "loss": 0.9305,
+      "step": 21290
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4970979708311784e-06,
+      "loss": 0.9558,
+      "step": 21291
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4966707192041853e-06,
+      "loss": 0.986,
+      "step": 21292
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.49624364658993e-06,
+      "loss": 0.9467,
+      "step": 21293
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4958167529964585e-06,
+      "loss": 0.9607,
+      "step": 21294
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.495390038431811e-06,
+      "loss": 1.0501,
+      "step": 21295
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.494963502904028e-06,
+      "loss": 0.99,
+      "step": 21296
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.49453714642114e-06,
+      "loss": 0.99,
+      "step": 21297
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4941109689911814e-06,
+      "loss": 1.0218,
+      "step": 21298
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.493684970622179e-06,
+      "loss": 0.9367,
+      "step": 21299
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4932591513221585e-06,
+      "loss": 0.905,
+      "step": 21300
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4928335110991386e-06,
+      "loss": 0.9785,
+      "step": 21301
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4924080499611384e-06,
+      "loss": 0.9834,
+      "step": 21302
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4919827679161715e-06,
+      "loss": 1.0003,
+      "step": 21303
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.491557664972251e-06,
+      "loss": 0.9308,
+      "step": 21304
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.491132741137381e-06,
+      "loss": 1.0428,
+      "step": 21305
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4907079964195692e-06,
+      "loss": 0.9357,
+      "step": 21306
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.490283430826814e-06,
+      "loss": 0.9904,
+      "step": 21307
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.489859044367115e-06,
+      "loss": 1.0072,
+      "step": 21308
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4894348370484646e-06,
+      "loss": 0.9675,
+      "step": 21309
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4890108088788535e-06,
+      "loss": 1.0009,
+      "step": 21310
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4885869598662708e-06,
+      "loss": 1.037,
+      "step": 21311
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.488163290018698e-06,
+      "loss": 1.0155,
+      "step": 21312
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.487739799344118e-06,
+      "loss": 1.0092,
+      "step": 21313
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4873164878505067e-06,
+      "loss": 0.9569,
+      "step": 21314
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4868933555458378e-06,
+      "loss": 0.927,
+      "step": 21315
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.486470402438082e-06,
+      "loss": 0.9588,
+      "step": 21316
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.486047628535208e-06,
+      "loss": 1.041,
+      "step": 21317
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4856250338451763e-06,
+      "loss": 0.9487,
+      "step": 21318
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4852026183759504e-06,
+      "loss": 1.0418,
+      "step": 21319
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4847803821354856e-06,
+      "loss": 1.0457,
+      "step": 21320
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4843583251317363e-06,
+      "loss": 1.0352,
+      "step": 21321
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4839364473726515e-06,
+      "loss": 1.0009,
+      "step": 21322
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4835147488661794e-06,
+      "loss": 0.9561,
+      "step": 21323
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.483093229620264e-06,
+      "loss": 1.0187,
+      "step": 21324
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4826718896428437e-06,
+      "loss": 0.9492,
+      "step": 21325
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.482250728941854e-06,
+      "loss": 0.9871,
+      "step": 21326
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4818297475252317e-06,
+      "loss": 1.0432,
+      "step": 21327
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.481408945400905e-06,
+      "loss": 1.0192,
+      "step": 21328
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.480988322576801e-06,
+      "loss": 1.0426,
+      "step": 21329
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.480567879060841e-06,
+      "loss": 0.9642,
+      "step": 21330
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4801476148609483e-06,
+      "loss": 0.9699,
+      "step": 21331
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4797275299850366e-06,
+      "loss": 1.0249,
+      "step": 21332
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.47930762444102e-06,
+      "loss": 0.9669,
+      "step": 21333
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.478887898236807e-06,
+      "loss": 1.0195,
+      "step": 21334
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4784683513803057e-06,
+      "loss": 0.9685,
+      "step": 21335
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.47804898387942e-06,
+      "loss": 0.9195,
+      "step": 21336
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4776297957420436e-06,
+      "loss": 0.9917,
+      "step": 21337
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.47721078697608e-06,
+      "loss": 1.0683,
+      "step": 21338
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4767919575894173e-06,
+      "loss": 0.9677,
+      "step": 21339
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.476373307589948e-06,
+      "loss": 1.01,
+      "step": 21340
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4759548369855548e-06,
+      "loss": 0.9259,
+      "step": 21341
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.475536545784122e-06,
+      "loss": 0.9858,
+      "step": 21342
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.475118433993531e-06,
+      "loss": 0.9051,
+      "step": 21343
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4747005016216555e-06,
+      "loss": 1.0031,
+      "step": 21344
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.474282748676366e-06,
+      "loss": 1.0256,
+      "step": 21345
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.473865175165534e-06,
+      "loss": 0.9076,
+      "step": 21346
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4734477810970284e-06,
+      "loss": 0.911,
+      "step": 21347
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4730305664787058e-06,
+      "loss": 0.9671,
+      "step": 21348
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4726135313184267e-06,
+      "loss": 1.0045,
+      "step": 21349
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.472196675624049e-06,
+      "loss": 0.9548,
+      "step": 21350
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4717799994034255e-06,
+      "loss": 1.0151,
+      "step": 21351
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4713635026644013e-06,
+      "loss": 1.0001,
+      "step": 21352
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.470947185414824e-06,
+      "loss": 0.929,
+      "step": 21353
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.470531047662537e-06,
+      "loss": 1.0326,
+      "step": 21354
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.470115089415376e-06,
+      "loss": 1.0314,
+      "step": 21355
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4696993106811785e-06,
+      "loss": 0.9455,
+      "step": 21356
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.469283711467775e-06,
+      "loss": 0.9932,
+      "step": 21357
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.468868291782997e-06,
+      "loss": 0.9451,
+      "step": 21358
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4684530516346663e-06,
+      "loss": 1.0522,
+      "step": 21359
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.468037991030605e-06,
+      "loss": 0.9382,
+      "step": 21360
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4676231099786332e-06,
+      "loss": 0.9603,
+      "step": 21361
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4672084084865678e-06,
+      "loss": 1.0243,
+      "step": 21362
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4667938865622167e-06,
+      "loss": 1.0139,
+      "step": 21363
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4663795442133876e-06,
+      "loss": 1.0048,
+      "step": 21364
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.46596538144789e-06,
+      "loss": 1.0066,
+      "step": 21365
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4655513982735216e-06,
+      "loss": 0.981,
+      "step": 21366
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.465137594698082e-06,
+      "loss": 0.9552,
+      "step": 21367
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4647239707293638e-06,
+      "loss": 0.9307,
+      "step": 21368
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4643105263751625e-06,
+      "loss": 0.9798,
+      "step": 21369
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4638972616432634e-06,
+      "loss": 0.9563,
+      "step": 21370
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.463484176541451e-06,
+      "loss": 0.9234,
+      "step": 21371
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4630712710775055e-06,
+      "loss": 0.9652,
+      "step": 21372
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.462658545259209e-06,
+      "loss": 0.9613,
+      "step": 21373
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4622459990943315e-06,
+      "loss": 0.9573,
+      "step": 21374
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.461833632590647e-06,
+      "loss": 1.0359,
+      "step": 21375
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4614214457559206e-06,
+      "loss": 1.0372,
+      "step": 21376
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4610094385979192e-06,
+      "loss": 1.0507,
+      "step": 21377
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4605976111244013e-06,
+      "loss": 0.9344,
+      "step": 21378
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.460185963343126e-06,
+      "loss": 1.0554,
+      "step": 21379
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4597744952618476e-06,
+      "loss": 1.0001,
+      "step": 21380
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.459363206888316e-06,
+      "loss": 0.9837,
+      "step": 21381
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4589520982302784e-06,
+      "loss": 1.055,
+      "step": 21382
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.45854116929548e-06,
+      "loss": 1.0071,
+      "step": 21383
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.45813042009166e-06,
+      "loss": 0.9634,
+      "step": 21384
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4577198506265566e-06,
+      "loss": 1.0375,
+      "step": 21385
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.457309460907903e-06,
+      "loss": 0.9864,
+      "step": 21386
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.456899250943431e-06,
+      "loss": 1.0394,
+      "step": 21387
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4564892207408654e-06,
+      "loss": 0.9651,
+      "step": 21388
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.456079370307931e-06,
+      "loss": 1.0357,
+      "step": 21389
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.45566969965235e-06,
+      "loss": 1.0191,
+      "step": 21390
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4552602087818356e-06,
+      "loss": 0.9941,
+      "step": 21391
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4548508977041043e-06,
+      "loss": 0.9818,
+      "step": 21392
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.454441766426865e-06,
+      "loss": 0.9742,
+      "step": 21393
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4540328149578244e-06,
+      "loss": 0.9719,
+      "step": 21394
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4536240433046856e-06,
+      "loss": 0.9368,
+      "step": 21395
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.45321545147515e-06,
+      "loss": 0.9103,
+      "step": 21396
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4528070394769124e-06,
+      "loss": 0.9532,
+      "step": 21397
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4523988073176666e-06,
+      "loss": 1.0287,
+      "step": 21398
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.451990755005104e-06,
+      "loss": 0.9011,
+      "step": 21399
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.451582882546908e-06,
+      "loss": 0.9562,
+      "step": 21400
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4511751899507644e-06,
+      "loss": 0.9832,
+      "step": 21401
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4507676772243505e-06,
+      "loss": 0.991,
+      "step": 21402
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4503603443753444e-06,
+      "loss": 1.0394,
+      "step": 21403
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4499531914114187e-06,
+      "loss": 0.9166,
+      "step": 21404
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.449546218340242e-06,
+      "loss": 0.9232,
+      "step": 21405
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.44913942516948e-06,
+      "loss": 1.0535,
+      "step": 21406
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.448732811906798e-06,
+      "loss": 0.998,
+      "step": 21407
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4483263785598525e-06,
+      "loss": 0.9413,
+      "step": 21408
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.447920125136302e-06,
+      "loss": 0.9703,
+      "step": 21409
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.447514051643796e-06,
+      "loss": 0.9299,
+      "step": 21410
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.447108158089985e-06,
+      "loss": 0.9599,
+      "step": 21411
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.446702444482515e-06,
+      "loss": 0.9603,
+      "step": 21412
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4462969108290303e-06,
+      "loss": 0.9495,
+      "step": 21413
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4458915571371646e-06,
+      "loss": 0.9441,
+      "step": 21414
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4454863834145582e-06,
+      "loss": 0.9843,
+      "step": 21415
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4450813896688433e-06,
+      "loss": 0.9382,
+      "step": 21416
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4446765759076453e-06,
+      "loss": 0.985,
+      "step": 21417
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.444271942138592e-06,
+      "loss": 0.9263,
+      "step": 21418
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.443867488369306e-06,
+      "loss": 0.971,
+      "step": 21419
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.443463214607406e-06,
+      "loss": 0.9399,
+      "step": 21420
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4430591208605028e-06,
+      "loss": 0.9793,
+      "step": 21421
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4426552071362132e-06,
+      "loss": 0.9976,
+      "step": 21422
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4422514734421435e-06,
+      "loss": 0.9816,
+      "step": 21423
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4418479197859017e-06,
+      "loss": 1.0033,
+      "step": 21424
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.441444546175084e-06,
+      "loss": 1.0321,
+      "step": 21425
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4410413526172927e-06,
+      "loss": 0.9524,
+      "step": 21426
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.440638339120124e-06,
+      "loss": 1.0478,
+      "step": 21427
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4402355056911656e-06,
+      "loss": 0.9837,
+      "step": 21428
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4398328523380056e-06,
+      "loss": 1.036,
+      "step": 21429
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.439430379068232e-06,
+      "loss": 0.9934,
+      "step": 21430
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4390280858894257e-06,
+      "loss": 0.9826,
+      "step": 21431
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4386259728091613e-06,
+      "loss": 0.9394,
+      "step": 21432
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4382240398350155e-06,
+      "loss": 0.9598,
+      "step": 21433
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4378222869745596e-06,
+      "loss": 0.9929,
+      "step": 21434
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4374207142353626e-06,
+      "loss": 0.969,
+      "step": 21435
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4370193216249873e-06,
+      "loss": 0.9586,
+      "step": 21436
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.436618109150992e-06,
+      "loss": 0.9879,
+      "step": 21437
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4362170768209407e-06,
+      "loss": 0.9819,
+      "step": 21438
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4358162246423813e-06,
+      "loss": 0.9644,
+      "step": 21439
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4354155526228687e-06,
+      "loss": 0.9572,
+      "step": 21440
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4350150607699466e-06,
+      "loss": 0.9762,
+      "step": 21441
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4346147490911644e-06,
+      "loss": 1.0024,
+      "step": 21442
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4342146175940576e-06,
+      "loss": 1.011,
+      "step": 21443
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.433814666286166e-06,
+      "loss": 0.9097,
+      "step": 21444
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4334148951750205e-06,
+      "loss": 1.0016,
+      "step": 21445
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4330153042681574e-06,
+      "loss": 0.97,
+      "step": 21446
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4326158935730984e-06,
+      "loss": 1.0182,
+      "step": 21447
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.432216663097368e-06,
+      "loss": 0.9582,
+      "step": 21448
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4318176128484895e-06,
+      "loss": 0.9576,
+      "step": 21449
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4314187428339765e-06,
+      "loss": 0.9949,
+      "step": 21450
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.431020053061343e-06,
+      "loss": 1.0451,
+      "step": 21451
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.430621543538099e-06,
+      "loss": 1.0064,
+      "step": 21452
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.430223214271754e-06,
+      "loss": 0.9544,
+      "step": 21453
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4298250652698077e-06,
+      "loss": 0.9433,
+      "step": 21454
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4294270965397608e-06,
+      "loss": 0.8932,
+      "step": 21455
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4290293080891095e-06,
+      "loss": 1.0377,
+      "step": 21456
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4286316999253503e-06,
+      "loss": 0.9684,
+      "step": 21457
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4282342720559682e-06,
+      "loss": 0.9831,
+      "step": 21458
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.427837024488452e-06,
+      "loss": 0.9742,
+      "step": 21459
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4274399572302844e-06,
+      "loss": 0.9255,
+      "step": 21460
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.427043070288945e-06,
+      "loss": 0.9851,
+      "step": 21461
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4266463636719084e-06,
+      "loss": 0.9597,
+      "step": 21462
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4262498373866485e-06,
+      "loss": 0.9821,
+      "step": 21463
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.425853491440634e-06,
+      "loss": 0.933,
+      "step": 21464
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.425457325841333e-06,
+      "loss": 1.0105,
+      "step": 21465
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4250613405962043e-06,
+      "loss": 0.9878,
+      "step": 21466
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4246655357127103e-06,
+      "loss": 1.0663,
+      "step": 21467
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.424269911198305e-06,
+      "loss": 1.1139,
+      "step": 21468
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.423874467060441e-06,
+      "loss": 0.9676,
+      "step": 21469
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.423479203306568e-06,
+      "loss": 1.0271,
+      "step": 21470
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4230841199441303e-06,
+      "loss": 0.9638,
+      "step": 21471
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4226892169805706e-06,
+      "loss": 0.9752,
+      "step": 21472
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4222944944233287e-06,
+      "loss": 0.886,
+      "step": 21473
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.421899952279837e-06,
+      "loss": 0.9457,
+      "step": 21474
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4215055905575317e-06,
+      "loss": 0.9027,
+      "step": 21475
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.421111409263837e-06,
+      "loss": 1.0675,
+      "step": 21476
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4207174084061803e-06,
+      "loss": 0.9665,
+      "step": 21477
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.420323587991983e-06,
+      "loss": 0.972,
+      "step": 21478
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4199299480286633e-06,
+      "loss": 1.0275,
+      "step": 21479
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4195364885236376e-06,
+      "loss": 0.9664,
+      "step": 21480
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4191432094843137e-06,
+      "loss": 0.9403,
+      "step": 21481
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4187501109181046e-06,
+      "loss": 1.0168,
+      "step": 21482
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4183571928324116e-06,
+      "loss": 0.9953,
+      "step": 21483
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.417964455234636e-06,
+      "loss": 0.9922,
+      "step": 21484
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4175718981321772e-06,
+      "loss": 0.9345,
+      "step": 21485
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.41717952153243e-06,
+      "loss": 0.9501,
+      "step": 21486
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4167873254427843e-06,
+      "loss": 1.0161,
+      "step": 21487
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4163953098706284e-06,
+      "loss": 0.9494,
+      "step": 21488
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4160034748233464e-06,
+      "loss": 0.9853,
+      "step": 21489
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4156118203083203e-06,
+      "loss": 0.9861,
+      "step": 21490
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4152203463329264e-06,
+      "loss": 0.9021,
+      "step": 21491
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4148290529045385e-06,
+      "loss": 0.9363,
+      "step": 21492
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4144379400305297e-06,
+      "loss": 1.0164,
+      "step": 21493
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4140470077182632e-06,
+      "loss": 1.0179,
+      "step": 21494
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4136562559751065e-06,
+      "loss": 0.9457,
+      "step": 21495
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.413265684808419e-06,
+      "loss": 0.9195,
+      "step": 21496
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4128752942255595e-06,
+      "loss": 1.0125,
+      "step": 21497
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.412485084233878e-06,
+      "loss": 1.0007,
+      "step": 21498
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4120950548407266e-06,
+      "loss": 0.9693,
+      "step": 21499
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.411705206053455e-06,
+      "loss": 1.0363,
+      "step": 21500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.411315537879403e-06,
+      "loss": 1.0125,
+      "step": 21501
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4109260503259102e-06,
+      "loss": 1.027,
+      "step": 21502
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.410536743400316e-06,
+      "loss": 0.9466,
+      "step": 21503
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.410147617109955e-06,
+      "loss": 1.1026,
+      "step": 21504
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4097586714621512e-06,
+      "loss": 0.9373,
+      "step": 21505
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4093699064642376e-06,
+      "loss": 1.0065,
+      "step": 21506
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.408981322123533e-06,
+      "loss": 0.9574,
+      "step": 21507
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4085929184473613e-06,
+      "loss": 0.9339,
+      "step": 21508
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.408204695443033e-06,
+      "loss": 0.9622,
+      "step": 21509
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.407816653117865e-06,
+      "loss": 1.0164,
+      "step": 21510
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4074287914791673e-06,
+      "loss": 0.9463,
+      "step": 21511
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.407041110534243e-06,
+      "loss": 1.0103,
+      "step": 21512
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.406653610290396e-06,
+      "loss": 0.8932,
+      "step": 21513
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4062662907549262e-06,
+      "loss": 0.9398,
+      "step": 21514
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.405879151935131e-06,
+      "loss": 1.0156,
+      "step": 21515
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4054921938383005e-06,
+      "loss": 0.9831,
+      "step": 21516
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4051054164717226e-06,
+      "loss": 0.9907,
+      "step": 21517
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4047188198426866e-06,
+      "loss": 0.951,
+      "step": 21518
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4043324039584727e-06,
+      "loss": 1.0115,
+      "step": 21519
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.40394616882636e-06,
+      "loss": 1.0335,
+      "step": 21520
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4035601144536227e-06,
+      "loss": 1.0293,
+      "step": 21521
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.403174240847536e-06,
+      "loss": 0.9678,
+      "step": 21522
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4027885480153657e-06,
+      "loss": 0.9908,
+      "step": 21523
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.402403035964377e-06,
+      "loss": 0.8984,
+      "step": 21524
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4020177047018323e-06,
+      "loss": 0.975,
+      "step": 21525
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.401632554234993e-06,
+      "loss": 0.9656,
+      "step": 21526
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4012475845711105e-06,
+      "loss": 1.0116,
+      "step": 21527
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.400862795717437e-06,
+      "loss": 1.0364,
+      "step": 21528
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4004781876812202e-06,
+      "loss": 0.9527,
+      "step": 21529
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4000937604697072e-06,
+      "loss": 0.9793,
+      "step": 21530
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3997095140901384e-06,
+      "loss": 1.0177,
+      "step": 21531
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.39932544854975e-06,
+      "loss": 1.0213,
+      "step": 21532
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.398941563855779e-06,
+      "loss": 1.0719,
+      "step": 21533
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3985578600154548e-06,
+      "loss": 0.9577,
+      "step": 21534
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3981743370360067e-06,
+      "loss": 1.0058,
+      "step": 21535
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.397790994924657e-06,
+      "loss": 0.9637,
+      "step": 21536
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.39740783368863e-06,
+      "loss": 1.0516,
+      "step": 21537
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3970248533351404e-06,
+      "loss": 0.948,
+      "step": 21538
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3966420538714034e-06,
+      "loss": 0.9907,
+      "step": 21539
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.396259435304628e-06,
+      "loss": 0.9443,
+      "step": 21540
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3958769976420252e-06,
+      "loss": 1.0392,
+      "step": 21541
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.395494740890796e-06,
+      "loss": 0.9393,
+      "step": 21542
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3951126650581414e-06,
+      "loss": 0.9239,
+      "step": 21543
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.394730770151258e-06,
+      "loss": 1.0743,
+      "step": 21544
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.394349056177341e-06,
+      "loss": 1.016,
+      "step": 21545
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.39396752314358e-06,
+      "loss": 1.0073,
+      "step": 21546
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3935861710571618e-06,
+      "loss": 0.9335,
+      "step": 21547
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3932049999252707e-06,
+      "loss": 0.9935,
+      "step": 21548
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.392824009755085e-06,
+      "loss": 1.006,
+      "step": 21549
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3924432005537827e-06,
+      "loss": 0.9565,
+      "step": 21550
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3920625723285354e-06,
+      "loss": 0.8913,
+      "step": 21551
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3916821250865153e-06,
+      "loss": 1.0424,
+      "step": 21552
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3913018588348882e-06,
+      "loss": 1.0471,
+      "step": 21553
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3909217735808155e-06,
+      "loss": 0.984,
+      "step": 21554
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3905418693314588e-06,
+      "loss": 0.9938,
+      "step": 21555
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.390162146093973e-06,
+      "loss": 0.9917,
+      "step": 21556
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3897826038755115e-06,
+      "loss": 0.986,
+      "step": 21557
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3894032426832228e-06,
+      "loss": 0.9367,
+      "step": 21558
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3890240625242547e-06,
+      "loss": 0.992,
+      "step": 21559
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3886450634057482e-06,
+      "loss": 1.0545,
+      "step": 21560
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.388266245334843e-06,
+      "loss": 1.0444,
+      "step": 21561
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3878876083186745e-06,
+      "loss": 1.1075,
+      "step": 21562
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3875091523643747e-06,
+      "loss": 0.985,
+      "step": 21563
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.387130877479074e-06,
+      "loss": 0.9255,
+      "step": 21564
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3867527836698956e-06,
+      "loss": 1.0159,
+      "step": 21565
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.386374870943964e-06,
+      "loss": 0.9659,
+      "step": 21566
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.385997139308397e-06,
+      "loss": 0.9793,
+      "step": 21567
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3856195887703095e-06,
+      "loss": 1.0155,
+      "step": 21568
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3852422193368133e-06,
+      "loss": 0.9415,
+      "step": 21569
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3848650310150168e-06,
+      "loss": 0.9606,
+      "step": 21570
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3844880238120264e-06,
+      "loss": 0.9145,
+      "step": 21571
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.384111197734942e-06,
+      "loss": 1.0004,
+      "step": 21572
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.383734552790863e-06,
+      "loss": 0.9555,
+      "step": 21573
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3833580889868835e-06,
+      "loss": 0.9704,
+      "step": 21574
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.382981806330095e-06,
+      "loss": 0.9671,
+      "step": 21575
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3826057048275853e-06,
+      "loss": 0.9604,
+      "step": 21576
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3822297844864407e-06,
+      "loss": 0.9907,
+      "step": 21577
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3818540453137377e-06,
+      "loss": 0.9427,
+      "step": 21578
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.38147848731656e-06,
+      "loss": 0.9591,
+      "step": 21579
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.381103110501978e-06,
+      "loss": 0.9059,
+      "step": 21580
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3807279148770647e-06,
+      "loss": 0.9182,
+      "step": 21581
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3803529004488846e-06,
+      "loss": 1.0218,
+      "step": 21582
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3799780672245055e-06,
+      "loss": 1.0142,
+      "step": 21583
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3796034152109873e-06,
+      "loss": 0.9434,
+      "step": 21584
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.379228944415385e-06,
+      "loss": 0.9828,
+      "step": 21585
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3788546548447534e-06,
+      "loss": 1.0132,
+      "step": 21586
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3784805465061447e-06,
+      "loss": 0.9553,
+      "step": 21587
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.378106619406605e-06,
+      "loss": 1.044,
+      "step": 21588
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.377732873553176e-06,
+      "loss": 0.9233,
+      "step": 21589
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.377359308952899e-06,
+      "loss": 0.9597,
+      "step": 21590
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3769859256128113e-06,
+      "loss": 0.9471,
+      "step": 21591
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3766127235399487e-06,
+      "loss": 1.0084,
+      "step": 21592
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3762397027413355e-06,
+      "loss": 0.9728,
+      "step": 21593
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3758668632240023e-06,
+      "loss": 1.0458,
+      "step": 21594
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3754942049949737e-06,
+      "loss": 0.9599,
+      "step": 21595
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.375121728061265e-06,
+      "loss": 0.9871,
+      "step": 21596
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3747494324298926e-06,
+      "loss": 1.0342,
+      "step": 21597
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3743773181078724e-06,
+      "loss": 1.0543,
+      "step": 21598
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.374005385102215e-06,
+      "loss": 0.9323,
+      "step": 21599
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.373633633419922e-06,
+      "loss": 0.9901,
+      "step": 21600
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.373262063067997e-06,
+      "loss": 0.9401,
+      "step": 21601
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.372890674053442e-06,
+      "loss": 0.9801,
+      "step": 21602
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.372519466383252e-06,
+      "loss": 1.0502,
+      "step": 21603
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.372148440064418e-06,
+      "loss": 1.0091,
+      "step": 21604
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.371777595103928e-06,
+      "loss": 1.0711,
+      "step": 21605
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.371406931508771e-06,
+      "loss": 0.9768,
+      "step": 21606
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3710364492859264e-06,
+      "loss": 0.9817,
+      "step": 21607
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3706661484423736e-06,
+      "loss": 0.9872,
+      "step": 21608
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3702960289850866e-06,
+      "loss": 0.9369,
+      "step": 21609
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.369926090921041e-06,
+      "loss": 1.0036,
+      "step": 21610
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3695563342572015e-06,
+      "loss": 0.895,
+      "step": 21611
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.369186759000536e-06,
+      "loss": 1.0331,
+      "step": 21612
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.368817365158003e-06,
+      "loss": 0.9541,
+      "step": 21613
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3684481527365636e-06,
+      "loss": 0.911,
+      "step": 21614
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3680791217431713e-06,
+      "loss": 0.9316,
+      "step": 21615
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.367710272184777e-06,
+      "loss": 0.9504,
+      "step": 21616
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3673416040683296e-06,
+      "loss": 1.0341,
+      "step": 21617
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.366973117400774e-06,
+      "loss": 0.975,
+      "step": 21618
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3666048121890497e-06,
+      "loss": 0.9851,
+      "step": 21619
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.366236688440096e-06,
+      "loss": 0.9739,
+      "step": 21620
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.365868746160847e-06,
+      "loss": 0.9381,
+      "step": 21621
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3655009853582315e-06,
+      "loss": 1.0218,
+      "step": 21622
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.36513340603918e-06,
+      "loss": 0.9725,
+      "step": 21623
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3647660082106128e-06,
+      "loss": 0.9737,
+      "step": 21624
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3643987918794567e-06,
+      "loss": 1.0134,
+      "step": 21625
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.364031757052623e-06,
+      "loss": 0.9431,
+      "step": 21626
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3636649037370277e-06,
+      "loss": 1.008,
+      "step": 21627
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3632982319395805e-06,
+      "loss": 0.9961,
+      "step": 21628
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3629317416671885e-06,
+      "loss": 0.9778,
+      "step": 21629
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3625654329267557e-06,
+      "loss": 0.9108,
+      "step": 21630
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3621993057251825e-06,
+      "loss": 1.0147,
+      "step": 21631
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3618333600693655e-06,
+      "loss": 0.9794,
+      "step": 21632
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3614675959661974e-06,
+      "loss": 0.9582,
+      "step": 21633
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.361102013422567e-06,
+      "loss": 1.0005,
+      "step": 21634
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3607366124453635e-06,
+      "loss": 0.9227,
+      "step": 21635
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3603713930414675e-06,
+      "loss": 0.9521,
+      "step": 21636
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.360006355217759e-06,
+      "loss": 0.9963,
+      "step": 21637
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.359641498981116e-06,
+      "loss": 1.0256,
+      "step": 21638
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.359276824338408e-06,
+      "loss": 0.9638,
+      "step": 21639
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3589123312965068e-06,
+      "loss": 0.9663,
+      "step": 21640
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.358548019862278e-06,
+      "loss": 1.019,
+      "step": 21641
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3581838900425837e-06,
+      "loss": 0.9944,
+      "step": 21642
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.357819941844283e-06,
+      "loss": 1.0097,
+      "step": 21643
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3574561752742322e-06,
+      "loss": 1.0256,
+      "step": 21644
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.357092590339282e-06,
+      "loss": 0.9955,
+      "step": 21645
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.356729187046282e-06,
+      "loss": 1.0393,
+      "step": 21646
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3563659654020765e-06,
+      "loss": 0.9855,
+      "step": 21647
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.35600292541351e-06,
+      "loss": 1.0022,
+      "step": 21648
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3556400670874194e-06,
+      "loss": 1.0524,
+      "step": 21649
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3552773904306405e-06,
+      "loss": 0.9506,
+      "step": 21650
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3549148954500024e-06,
+      "loss": 0.9218,
+      "step": 21651
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3545525821523364e-06,
+      "loss": 0.8949,
+      "step": 21652
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.354190450544467e-06,
+      "loss": 0.9261,
+      "step": 21653
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3538285006332154e-06,
+      "loss": 1.0194,
+      "step": 21654
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.353466732425398e-06,
+      "loss": 1.0358,
+      "step": 21655
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3531051459278313e-06,
+      "loss": 1.0343,
+      "step": 21656
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3527437411473245e-06,
+      "loss": 0.948,
+      "step": 21657
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3523825180906868e-06,
+      "loss": 1.0134,
+      "step": 21658
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3520214767647234e-06,
+      "loss": 0.9963,
+      "step": 21659
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3516606171762337e-06,
+      "loss": 1.04,
+      "step": 21660
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.351299939332016e-06,
+      "loss": 0.9892,
+      "step": 21661
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3509394432388614e-06,
+      "loss": 0.8944,
+      "step": 21662
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.350579128903564e-06,
+      "loss": 1.0224,
+      "step": 21663
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3502189963329105e-06,
+      "loss": 1.0153,
+      "step": 21664
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3498590455336847e-06,
+      "loss": 0.9667,
+      "step": 21665
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3494992765126638e-06,
+      "loss": 1.0091,
+      "step": 21666
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3491396892766286e-06,
+      "loss": 1.0358,
+      "step": 21667
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3487802838323526e-06,
+      "loss": 0.9423,
+      "step": 21668
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3484210601866035e-06,
+      "loss": 1.0329,
+      "step": 21669
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.348062018346147e-06,
+      "loss": 1.027,
+      "step": 21670
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3477031583177494e-06,
+      "loss": 0.996,
+      "step": 21671
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3473444801081703e-06,
+      "loss": 1.009,
+      "step": 21672
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.346985983724165e-06,
+      "loss": 1.0261,
+      "step": 21673
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3466276691724847e-06,
+      "loss": 1.0619,
+      "step": 21674
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.346269536459881e-06,
+      "loss": 1.0825,
+      "step": 21675
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3459115855931016e-06,
+      "loss": 0.9271,
+      "step": 21676
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3455538165788863e-06,
+      "loss": 0.974,
+      "step": 21677
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.345196229423974e-06,
+      "loss": 0.9471,
+      "step": 21678
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.344838824135104e-06,
+      "loss": 0.9852,
+      "step": 21679
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3444816007190055e-06,
+      "loss": 1.0283,
+      "step": 21680
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3441245591824072e-06,
+      "loss": 0.8964,
+      "step": 21681
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3437676995320372e-06,
+      "loss": 0.975,
+      "step": 21682
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.343411021774618e-06,
+      "loss": 0.9166,
+      "step": 21683
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3430545259168638e-06,
+      "loss": 0.9679,
+      "step": 21684
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.342698211965493e-06,
+      "loss": 0.9166,
+      "step": 21685
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.342342079927218e-06,
+      "loss": 0.9746,
+      "step": 21686
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.341986129808747e-06,
+      "loss": 0.9817,
+      "step": 21687
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3416303616167824e-06,
+      "loss": 0.9765,
+      "step": 21688
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.341274775358028e-06,
+      "loss": 0.9702,
+      "step": 21689
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3409193710391833e-06,
+      "loss": 0.9372,
+      "step": 21690
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.34056414866694e-06,
+      "loss": 1.0483,
+      "step": 21691
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3402091082479906e-06,
+      "loss": 0.945,
+      "step": 21692
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.339854249789021e-06,
+      "loss": 0.944,
+      "step": 21693
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.339499573296721e-06,
+      "loss": 1.0803,
+      "step": 21694
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.339145078777766e-06,
+      "loss": 1.0294,
+      "step": 21695
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3387907662388367e-06,
+      "loss": 0.9085,
+      "step": 21696
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3384366356866053e-06,
+      "loss": 1.0079,
+      "step": 21697
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.338082687127744e-06,
+      "loss": 0.95,
+      "step": 21698
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3377289205689202e-06,
+      "loss": 0.9793,
+      "step": 21699
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3373753360167964e-06,
+      "loss": 0.9872,
+      "step": 21700
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3370219334780333e-06,
+      "loss": 0.9611,
+      "step": 21701
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.336668712959288e-06,
+      "loss": 1.0631,
+      "step": 21702
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3363156744672156e-06,
+      "loss": 1.0002,
+      "step": 21703
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3359628180084637e-06,
+      "loss": 1.0232,
+      "step": 21704
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.335610143589681e-06,
+      "loss": 0.9859,
+      "step": 21705
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3352576512175097e-06,
+      "loss": 0.9175,
+      "step": 21706
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3349053408985903e-06,
+      "loss": 0.9563,
+      "step": 21707
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.334553212639558e-06,
+      "loss": 1.0521,
+      "step": 21708
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3342012664470472e-06,
+      "loss": 0.9926,
+      "step": 21709
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3338495023276876e-06,
+      "loss": 0.9491,
+      "step": 21710
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3334979202881037e-06,
+      "loss": 0.9822,
+      "step": 21711
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3331465203349193e-06,
+      "loss": 0.9612,
+      "step": 21712
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3327953024747544e-06,
+      "loss": 0.9721,
+      "step": 21713
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3324442667142226e-06,
+      "loss": 0.9242,
+      "step": 21714
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3320934130599384e-06,
+      "loss": 1.0089,
+      "step": 21715
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3317427415185104e-06,
+      "loss": 0.9764,
+      "step": 21716
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3313922520965428e-06,
+      "loss": 1.0284,
+      "step": 21717
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3310419448006396e-06,
+      "loss": 1.0699,
+      "step": 21718
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3306918196373987e-06,
+      "loss": 1.0144,
+      "step": 21719
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3303418766134147e-06,
+      "loss": 0.9804,
+      "step": 21720
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3299921157352792e-06,
+      "loss": 0.9957,
+      "step": 21721
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.329642537009582e-06,
+      "loss": 0.8924,
+      "step": 21722
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3292931404429076e-06,
+      "loss": 1.0719,
+      "step": 21723
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.328943926041838e-06,
+      "loss": 0.973,
+      "step": 21724
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.328594893812949e-06,
+      "loss": 1.0344,
+      "step": 21725
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3282460437628187e-06,
+      "loss": 0.9534,
+      "step": 21726
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.327897375898015e-06,
+      "loss": 1.0273,
+      "step": 21727
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3275488902251077e-06,
+      "loss": 1.0123,
+      "step": 21728
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.327200586750661e-06,
+      "loss": 0.9181,
+      "step": 21729
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3268524654812354e-06,
+      "loss": 1.0084,
+      "step": 21730
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3265045264233888e-06,
+      "loss": 0.9734,
+      "step": 21731
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3261567695836744e-06,
+      "loss": 0.9444,
+      "step": 21732
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3258091949686435e-06,
+      "loss": 0.9603,
+      "step": 21733
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3254618025848455e-06,
+      "loss": 0.9169,
+      "step": 21734
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3251145924388185e-06,
+      "loss": 1.005,
+      "step": 21735
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3247675645371085e-06,
+      "loss": 0.9804,
+      "step": 21736
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.324420718886249e-06,
+      "loss": 0.9575,
+      "step": 21737
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3240740554927766e-06,
+      "loss": 0.9752,
+      "step": 21738
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.323727574363217e-06,
+      "loss": 0.9944,
+      "step": 21739
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3233812755041e-06,
+      "loss": 0.9761,
+      "step": 21740
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3230351589219482e-06,
+      "loss": 1.0045,
+      "step": 21741
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3226892246232812e-06,
+      "loss": 0.9768,
+      "step": 21742
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3223434726146155e-06,
+      "loss": 0.9386,
+      "step": 21743
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3219979029024634e-06,
+      "loss": 0.9126,
+      "step": 21744
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3216525154933357e-06,
+      "loss": 0.984,
+      "step": 21745
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.321307310393736e-06,
+      "loss": 1.0259,
+      "step": 21746
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.320962287610169e-06,
+      "loss": 0.9297,
+      "step": 21747
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.320617447149133e-06,
+      "loss": 1.0317,
+      "step": 21748
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.320272789017124e-06,
+      "loss": 0.8891,
+      "step": 21749
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.319928313220633e-06,
+      "loss": 1.0065,
+      "step": 21750
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3195840197661514e-06,
+      "loss": 0.91,
+      "step": 21751
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3192399086601636e-06,
+      "loss": 0.9246,
+      "step": 21752
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3188959799091495e-06,
+      "loss": 1.0297,
+      "step": 21753
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.318552233519589e-06,
+      "loss": 1.0151,
+      "step": 21754
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.318208669497958e-06,
+      "loss": 1.0099,
+      "step": 21755
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3178652878507292e-06,
+      "loss": 0.9261,
+      "step": 21756
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3175220885843675e-06,
+      "loss": 1.049,
+      "step": 21757
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.317179071705339e-06,
+      "loss": 0.9763,
+      "step": 21758
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3168362372201062e-06,
+      "loss": 0.9884,
+      "step": 21759
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.316493585135127e-06,
+      "loss": 1.0182,
+      "step": 21760
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.316151115456854e-06,
+      "loss": 0.9225,
+      "step": 21761
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3158088281917392e-06,
+      "loss": 0.9561,
+      "step": 21762
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3154667233462317e-06,
+      "loss": 0.9509,
+      "step": 21763
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3151248009267737e-06,
+      "loss": 0.9459,
+      "step": 21764
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.314783060939807e-06,
+      "loss": 0.9309,
+      "step": 21765
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.314441503391767e-06,
+      "loss": 1.0112,
+      "step": 21766
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.314100128289092e-06,
+      "loss": 0.9403,
+      "step": 21767
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3137589356382076e-06,
+      "loss": 0.9465,
+      "step": 21768
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3134179254455426e-06,
+      "loss": 0.9898,
+      "step": 21769
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.313077097717521e-06,
+      "loss": 0.9163,
+      "step": 21770
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3127364524605636e-06,
+      "loss": 0.9863,
+      "step": 21771
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3123959896810855e-06,
+      "loss": 0.9658,
+      "step": 21772
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.312055709385499e-06,
+      "loss": 0.9518,
+      "step": 21773
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.311715611580218e-06,
+      "loss": 1.0145,
+      "step": 21774
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.311375696271645e-06,
+      "loss": 0.9999,
+      "step": 21775
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.311035963466184e-06,
+      "loss": 1.0177,
+      "step": 21776
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.310696413170234e-06,
+      "loss": 0.965,
+      "step": 21777
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3103570453901936e-06,
+      "loss": 0.9505,
+      "step": 21778
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.310017860132453e-06,
+      "loss": 0.9053,
+      "step": 21779
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3096788574034028e-06,
+      "loss": 0.9503,
+      "step": 21780
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.309340037209426e-06,
+      "loss": 1.0283,
+      "step": 21781
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3090013995569084e-06,
+      "loss": 0.9574,
+      "step": 21782
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.308662944452227e-06,
+      "loss": 1.0214,
+      "step": 21783
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.308324671901756e-06,
+      "loss": 0.9829,
+      "step": 21784
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.307986581911871e-06,
+      "loss": 0.962,
+      "step": 21785
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3076486744889374e-06,
+      "loss": 0.9603,
+      "step": 21786
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3073109496393224e-06,
+      "loss": 1.0256,
+      "step": 21787
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.306973407369385e-06,
+      "loss": 1.0162,
+      "step": 21788
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.306636047685487e-06,
+      "loss": 0.9383,
+      "step": 21789
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3062988705939795e-06,
+      "loss": 0.991,
+      "step": 21790
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.305961876101216e-06,
+      "loss": 0.9656,
+      "step": 21791
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.305625064213545e-06,
+      "loss": 0.9825,
+      "step": 21792
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3052884349373097e-06,
+      "loss": 1.0398,
+      "step": 21793
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3049519882788507e-06,
+      "loss": 0.9992,
+      "step": 21794
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.304615724244507e-06,
+      "loss": 0.9728,
+      "step": 21795
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3042796428406113e-06,
+      "loss": 0.9825,
+      "step": 21796
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3039437440734957e-06,
+      "loss": 0.9861,
+      "step": 21797
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.303608027949487e-06,
+      "loss": 0.9712,
+      "step": 21798
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3032724944749085e-06,
+      "loss": 0.9298,
+      "step": 21799
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3029371436560822e-06,
+      "loss": 0.9623,
+      "step": 21800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3026019754993224e-06,
+      "loss": 0.9553,
+      "step": 21801
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3022669900109444e-06,
+      "loss": 1.027,
+      "step": 21802
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3019321871972578e-06,
+      "loss": 0.9562,
+      "step": 21803
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3015975670645696e-06,
+      "loss": 0.9819,
+      "step": 21804
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3012631296191835e-06,
+      "loss": 0.9521,
+      "step": 21805
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3009288748673975e-06,
+      "loss": 1.0246,
+      "step": 21806
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.300594802815509e-06,
+      "loss": 0.9263,
+      "step": 21807
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3002609134698117e-06,
+      "loss": 1.0309,
+      "step": 21808
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2999272068365933e-06,
+      "loss": 0.9665,
+      "step": 21809
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.299593682922141e-06,
+      "loss": 0.9598,
+      "step": 21810
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2992603417327366e-06,
+      "loss": 0.9794,
+      "step": 21811
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2989271832746586e-06,
+      "loss": 1.0297,
+      "step": 21812
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2985942075541846e-06,
+      "loss": 0.9811,
+      "step": 21813
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2982614145775863e-06,
+      "loss": 0.9954,
+      "step": 21814
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2979288043511307e-06,
+      "loss": 0.96,
+      "step": 21815
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.297596376881085e-06,
+      "loss": 0.9496,
+      "step": 21816
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2972641321737103e-06,
+      "loss": 0.9751,
+      "step": 21817
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.296932070235266e-06,
+      "loss": 0.9674,
+      "step": 21818
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.296600191072005e-06,
+      "loss": 1.0027,
+      "step": 21819
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2962684946901804e-06,
+      "loss": 0.9912,
+      "step": 21820
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2959369810960397e-06,
+      "loss": 0.9527,
+      "step": 21821
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.29560565029583e-06,
+      "loss": 0.9634,
+      "step": 21822
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.295274502295788e-06,
+      "loss": 0.9919,
+      "step": 21823
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.294943537102155e-06,
+      "loss": 0.9541,
+      "step": 21824
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.294612754721164e-06,
+      "loss": 0.9275,
+      "step": 21825
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2942821551590478e-06,
+      "loss": 1.0541,
+      "step": 21826
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2939517384220304e-06,
+      "loss": 0.9756,
+      "step": 21827
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.293621504516337e-06,
+      "loss": 0.9823,
+      "step": 21828
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2932914534481913e-06,
+      "loss": 0.95,
+      "step": 21829
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2929615852238054e-06,
+      "loss": 1.0255,
+      "step": 21830
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.292631899849397e-06,
+      "loss": 1.0093,
+      "step": 21831
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2923023973311744e-06,
+      "loss": 1.0479,
+      "step": 21832
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.291973077675346e-06,
+      "loss": 0.9557,
+      "step": 21833
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.291643940888111e-06,
+      "loss": 0.9254,
+      "step": 21834
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2913149869756747e-06,
+      "loss": 0.9626,
+      "step": 21835
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.290986215944231e-06,
+      "loss": 0.9365,
+      "step": 21836
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2906576277999727e-06,
+      "loss": 0.9805,
+      "step": 21837
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.290329222549088e-06,
+      "loss": 0.9585,
+      "step": 21838
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.290001000197766e-06,
+      "loss": 0.9311,
+      "step": 21839
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.28967296075219e-06,
+      "loss": 0.9593,
+      "step": 21840
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2893451042185357e-06,
+      "loss": 1.0001,
+      "step": 21841
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.289017430602979e-06,
+      "loss": 1.0458,
+      "step": 21842
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.288689939911695e-06,
+      "loss": 0.9977,
+      "step": 21843
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2883626321508523e-06,
+      "loss": 0.9477,
+      "step": 21844
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2880355073266147e-06,
+      "loss": 1.0064,
+      "step": 21845
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.287708565445144e-06,
+      "loss": 1.008,
+      "step": 21846
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.287381806512601e-06,
+      "loss": 0.9757,
+      "step": 21847
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.287055230535138e-06,
+      "loss": 0.9506,
+      "step": 21848
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2867288375189093e-06,
+      "loss": 0.8939,
+      "step": 21849
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2864026274700602e-06,
+      "loss": 1.0399,
+      "step": 21850
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.286076600394739e-06,
+      "loss": 0.9491,
+      "step": 21851
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2857507562990847e-06,
+      "loss": 0.9766,
+      "step": 21852
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2854250951892355e-06,
+      "loss": 1.0269,
+      "step": 21853
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.285099617071325e-06,
+      "loss": 1.0162,
+      "step": 21854
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2847743219514874e-06,
+      "loss": 1.1284,
+      "step": 21855
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2844492098358464e-06,
+      "loss": 1.059,
+      "step": 21856
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2841242807305277e-06,
+      "loss": 1.0074,
+      "step": 21857
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.283799534641653e-06,
+      "loss": 0.9113,
+      "step": 21858
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.283474971575338e-06,
+      "loss": 1.0098,
+      "step": 21859
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.283150591537696e-06,
+      "loss": 1.0109,
+      "step": 21860
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2828263945348376e-06,
+      "loss": 0.9375,
+      "step": 21861
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2825023805728733e-06,
+      "loss": 1.0687,
+      "step": 21862
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.282178549657901e-06,
+      "loss": 0.9001,
+      "step": 21863
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2818549017960237e-06,
+      "loss": 1.0418,
+      "step": 21864
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2815314369933352e-06,
+      "loss": 0.9753,
+      "step": 21865
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2812081552559343e-06,
+      "loss": 1.0353,
+      "step": 21866
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2808850565899047e-06,
+      "loss": 1.0159,
+      "step": 21867
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.280562141001336e-06,
+      "loss": 0.9147,
+      "step": 21868
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.280239408496309e-06,
+      "loss": 1.0122,
+      "step": 21869
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.279916859080903e-06,
+      "loss": 1.064,
+      "step": 21870
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2795944927611947e-06,
+      "loss": 0.8932,
+      "step": 21871
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.279272309543257e-06,
+      "loss": 0.9906,
+      "step": 21872
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.278950309433158e-06,
+      "loss": 0.8944,
+      "step": 21873
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2786284924369632e-06,
+      "loss": 0.96,
+      "step": 21874
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.278306858560734e-06,
+      "loss": 0.972,
+      "step": 21875
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2779854078105302e-06,
+      "loss": 0.9216,
+      "step": 21876
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2776641401924062e-06,
+      "loss": 1.0539,
+      "step": 21877
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2773430557124134e-06,
+      "loss": 0.9456,
+      "step": 21878
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2770221543766002e-06,
+      "loss": 1.0467,
+      "step": 21879
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2767014361910126e-06,
+      "loss": 1.0041,
+      "step": 21880
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.27638090116169e-06,
+      "loss": 1.0228,
+      "step": 21881
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2760605492946722e-06,
+      "loss": 1.0446,
+      "step": 21882
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2757403805959923e-06,
+      "loss": 0.921,
+      "step": 21883
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2754203950716804e-06,
+      "loss": 0.9942,
+      "step": 21884
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.275100592727766e-06,
+      "loss": 1.034,
+      "step": 21885
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.274780973570273e-06,
+      "loss": 0.9853,
+      "step": 21886
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2744615376052204e-06,
+      "loss": 1.0615,
+      "step": 21887
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2741422848386266e-06,
+      "loss": 1.0042,
+      "step": 21888
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.273823215276505e-06,
+      "loss": 0.9921,
+      "step": 21889
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.273504328924866e-06,
+      "loss": 0.8612,
+      "step": 21890
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2731856257897167e-06,
+      "loss": 0.983,
+      "step": 21891
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2728671058770593e-06,
+      "loss": 1.0283,
+      "step": 21892
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2725487691928953e-06,
+      "loss": 1.0156,
+      "step": 21893
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2722306157432195e-06,
+      "loss": 0.9821,
+      "step": 21894
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2719126455340257e-06,
+      "loss": 1.0248,
+      "step": 21895
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2715948585713044e-06,
+      "loss": 1.0256,
+      "step": 21896
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.271277254861041e-06,
+      "loss": 1.0208,
+      "step": 21897
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2709598344092166e-06,
+      "loss": 0.9192,
+      "step": 21898
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2706425972218127e-06,
+      "loss": 1.0168,
+      "step": 21899
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2703255433048045e-06,
+      "loss": 0.9528,
+      "step": 21900
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2700086726641623e-06,
+      "loss": 0.9792,
+      "step": 21901
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2696919853058586e-06,
+      "loss": 1.0048,
+      "step": 21902
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.269375481235855e-06,
+      "loss": 0.9541,
+      "step": 21903
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2690591604601164e-06,
+      "loss": 0.9521,
+      "step": 21904
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2687430229846e-06,
+      "loss": 1.0415,
+      "step": 21905
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2684270688152613e-06,
+      "loss": 0.9449,
+      "step": 21906
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.268111297958049e-06,
+      "loss": 0.9461,
+      "step": 21907
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2677957104189154e-06,
+      "loss": 1.0375,
+      "step": 21908
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.267480306203803e-06,
+      "loss": 0.9964,
+      "step": 21909
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.267165085318656e-06,
+      "loss": 0.93,
+      "step": 21910
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2668500477694058e-06,
+      "loss": 0.9422,
+      "step": 21911
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.266535193561993e-06,
+      "loss": 1.0112,
+      "step": 21912
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2662205227023463e-06,
+      "loss": 0.8967,
+      "step": 21913
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2659060351963915e-06,
+      "loss": 0.9777,
+      "step": 21914
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2655917310500534e-06,
+      "loss": 1.0253,
+      "step": 21915
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2652776102692544e-06,
+      "loss": 1.0061,
+      "step": 21916
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2649636728599102e-06,
+      "loss": 0.97,
+      "step": 21917
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2646499188279327e-06,
+      "loss": 1.0126,
+      "step": 21918
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.264336348179235e-06,
+      "loss": 1.0081,
+      "step": 21919
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.264022960919723e-06,
+      "loss": 1.0557,
+      "step": 21920
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.263709757055299e-06,
+      "loss": 0.9873,
+      "step": 21921
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.263396736591862e-06,
+      "loss": 0.9385,
+      "step": 21922
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.26308389953531e-06,
+      "loss": 0.954,
+      "step": 21923
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2627712458915364e-06,
+      "loss": 1.0334,
+      "step": 21924
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2624587756664286e-06,
+      "loss": 0.9367,
+      "step": 21925
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.262146488865873e-06,
+      "loss": 0.9704,
+      "step": 21926
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.261834385495753e-06,
+      "loss": 1.0102,
+      "step": 21927
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2615224655619487e-06,
+      "loss": 0.9563,
+      "step": 21928
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2612107290703324e-06,
+      "loss": 0.8819,
+      "step": 21929
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2608991760267775e-06,
+      "loss": 1.0586,
+      "step": 21930
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2605878064371546e-06,
+      "loss": 0.9988,
+      "step": 21931
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.260276620307328e-06,
+      "loss": 0.9318,
+      "step": 21932
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2599656176431573e-06,
+      "loss": 0.9466,
+      "step": 21933
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2596547984505024e-06,
+      "loss": 0.9902,
+      "step": 21934
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.259344162735221e-06,
+      "loss": 1.0085,
+      "step": 21935
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2590337105031604e-06,
+      "loss": 1.0194,
+      "step": 21936
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.258723441760169e-06,
+      "loss": 0.8943,
+      "step": 21937
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2584133565120916e-06,
+      "loss": 0.999,
+      "step": 21938
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.258103454764772e-06,
+      "loss": 0.9214,
+      "step": 21939
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2577937365240444e-06,
+      "loss": 0.983,
+      "step": 21940
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2574842017957444e-06,
+      "loss": 1.0112,
+      "step": 21941
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.257174850585703e-06,
+      "loss": 1.0239,
+      "step": 21942
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.256865682899745e-06,
+      "loss": 0.9949,
+      "step": 21943
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2565566987436974e-06,
+      "loss": 0.9079,
+      "step": 21944
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.256247898123377e-06,
+      "loss": 1.0319,
+      "step": 21945
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.255939281044606e-06,
+      "loss": 0.9941,
+      "step": 21946
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2556308475131917e-06,
+      "loss": 1.0694,
+      "step": 21947
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2553225975349483e-06,
+      "loss": 0.9693,
+      "step": 21948
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.255014531115679e-06,
+      "loss": 0.9766,
+      "step": 21949
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2547066482611904e-06,
+      "loss": 0.9711,
+      "step": 21950
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.254398948977279e-06,
+      "loss": 0.9338,
+      "step": 21951
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2540914332697418e-06,
+      "loss": 0.9623,
+      "step": 21952
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.253784101144372e-06,
+      "loss": 0.9358,
+      "step": 21953
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.253476952606958e-06,
+      "loss": 0.974,
+      "step": 21954
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2531699876632865e-06,
+      "loss": 0.9606,
+      "step": 21955
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2528632063191394e-06,
+      "loss": 0.9858,
+      "step": 21956
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.252556608580294e-06,
+      "loss": 0.9559,
+      "step": 21957
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2522501944525278e-06,
+      "loss": 0.9495,
+      "step": 21958
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.251943963941611e-06,
+      "loss": 1.0029,
+      "step": 21959
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2516379170533135e-06,
+      "loss": 0.9302,
+      "step": 21960
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2513320537933993e-06,
+      "loss": 0.9828,
+      "step": 21961
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.25102637416763e-06,
+      "loss": 1.0334,
+      "step": 21962
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2507208781817636e-06,
+      "loss": 0.9323,
+      "step": 21963
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2504155658415553e-06,
+      "loss": 1.0056,
+      "step": 21964
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2501104371527564e-06,
+      "loss": 0.9195,
+      "step": 21965
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2498054921211122e-06,
+      "loss": 1.0387,
+      "step": 21966
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.24950073075237e-06,
+      "loss": 0.9666,
+      "step": 21967
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.249196153052269e-06,
+      "loss": 1.0511,
+      "step": 21968
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.248891759026546e-06,
+      "loss": 1.0219,
+      "step": 21969
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2485875486809362e-06,
+      "loss": 0.9317,
+      "step": 21970
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2482835220211685e-06,
+      "loss": 0.9853,
+      "step": 21971
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.247979679052971e-06,
+      "loss": 0.9347,
+      "step": 21972
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.247676019782066e-06,
+      "loss": 0.9459,
+      "step": 21973
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.247372544214175e-06,
+      "loss": 1.0362,
+      "step": 21974
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2470692523550134e-06,
+      "loss": 0.9769,
+      "step": 21975
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2467661442102952e-06,
+      "loss": 1.001,
+      "step": 21976
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2464632197857286e-06,
+      "loss": 1.005,
+      "step": 21977
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.246160479087021e-06,
+      "loss": 0.9473,
+      "step": 21978
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.245857922119873e-06,
+      "loss": 1.031,
+      "step": 21979
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2455555488899875e-06,
+      "loss": 1.0536,
+      "step": 21980
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.245253359403057e-06,
+      "loss": 1.0051,
+      "step": 21981
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.244951353664775e-06,
+      "loss": 0.9999,
+      "step": 21982
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2446495316808305e-06,
+      "loss": 1.0029,
+      "step": 21983
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.244347893456909e-06,
+      "loss": 0.9674,
+      "step": 21984
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.244046438998692e-06,
+      "loss": 0.8998,
+      "step": 21985
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2437451683118586e-06,
+      "loss": 0.9373,
+      "step": 21986
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.243444081402081e-06,
+      "loss": 0.9978,
+      "step": 21987
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2431431782750336e-06,
+      "loss": 0.9433,
+      "step": 21988
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.242842458936385e-06,
+      "loss": 0.9514,
+      "step": 21989
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.242541923391799e-06,
+      "loss": 0.9516,
+      "step": 21990
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2422415716469337e-06,
+      "loss": 0.9069,
+      "step": 21991
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2419414037074517e-06,
+      "loss": 0.9496,
+      "step": 21992
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.241641419579005e-06,
+      "loss": 0.9044,
+      "step": 21993
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2413416192672435e-06,
+      "loss": 1.0376,
+      "step": 21994
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2410420027778146e-06,
+      "loss": 1.0464,
+      "step": 21995
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2407425701163633e-06,
+      "loss": 1.0053,
+      "step": 21996
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2404433212885313e-06,
+      "loss": 1.1214,
+      "step": 21997
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.240144256299952e-06,
+      "loss": 1.0081,
+      "step": 21998
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.23984537515626e-06,
+      "loss": 0.9011,
+      "step": 21999
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.239546677863086e-06,
+      "loss": 0.9607,
+      "step": 22000
+    },
+    {
+      "epoch": 0.93,
+      "eval_text_loss": 0.7531290650367737,
+      "eval_text_runtime": 1.4008,
+      "eval_text_samples_per_second": 52.828,
+      "eval_text_steps_per_second": 1.428,
+      "step": 22000
+    },
+    {
+      "epoch": 0.93,
+      "eval_svit_loss": 1.070781946182251,
+      "eval_svit_runtime": 5.0024,
+      "eval_svit_samples_per_second": 136.135,
+      "eval_svit_steps_per_second": 2.199,
+      "step": 22000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.239248164426058e-06,
+      "loss": 0.9462,
+      "step": 22001
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2389498348507967e-06,
+      "loss": 0.9119,
+      "step": 22002
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.23865168914292e-06,
+      "loss": 0.9781,
+      "step": 22003
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2383537273080506e-06,
+      "loss": 0.9851,
+      "step": 22004
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.238055949351794e-06,
+      "loss": 1.0208,
+      "step": 22005
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2377583552797625e-06,
+      "loss": 0.9532,
+      "step": 22006
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2374609450975615e-06,
+      "loss": 1.0272,
+      "step": 22007
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2371637188107954e-06,
+      "loss": 1.0034,
+      "step": 22008
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2368666764250593e-06,
+      "loss": 0.9919,
+      "step": 22009
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2365698179459494e-06,
+      "loss": 1.0107,
+      "step": 22010
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2362731433790597e-06,
+      "loss": 0.8971,
+      "step": 22011
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2359766527299788e-06,
+      "loss": 0.9492,
+      "step": 22012
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2356803460042886e-06,
+      "loss": 0.9487,
+      "step": 22013
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.235384223207572e-06,
+      "loss": 0.9944,
+      "step": 22014
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.235088284345408e-06,
+      "loss": 0.9846,
+      "step": 22015
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.234792529423371e-06,
+      "loss": 1.0291,
+      "step": 22016
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2344969584470302e-06,
+      "loss": 1.0308,
+      "step": 22017
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2342015714219537e-06,
+      "loss": 1.0319,
+      "step": 22018
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.233906368353709e-06,
+      "loss": 0.9842,
+      "step": 22019
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.233611349247853e-06,
+      "loss": 0.9883,
+      "step": 22020
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2333165141099434e-06,
+      "loss": 1.0195,
+      "step": 22021
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2330218629455348e-06,
+      "loss": 1.0401,
+      "step": 22022
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2327273957601792e-06,
+      "loss": 0.9621,
+      "step": 22023
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.23243311255942e-06,
+      "loss": 0.9381,
+      "step": 22024
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2321390133488032e-06,
+      "loss": 0.9902,
+      "step": 22025
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.231845098133867e-06,
+      "loss": 0.9947,
+      "step": 22026
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2315513669201496e-06,
+      "loss": 0.96,
+      "step": 22027
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.231257819713182e-06,
+      "loss": 1.0112,
+      "step": 22028
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2309644565184944e-06,
+      "loss": 1.032,
+      "step": 22029
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2306712773416143e-06,
+      "loss": 1.0063,
+      "step": 22030
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2303782821880617e-06,
+      "loss": 0.9161,
+      "step": 22031
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.230085471063358e-06,
+      "loss": 1.0325,
+      "step": 22032
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2297928439730166e-06,
+      "loss": 0.943,
+      "step": 22033
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2295004009225533e-06,
+      "loss": 0.9176,
+      "step": 22034
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.229208141917473e-06,
+      "loss": 0.9222,
+      "step": 22035
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2289160669632827e-06,
+      "loss": 0.9675,
+      "step": 22036
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.228624176065484e-06,
+      "loss": 1.0924,
+      "step": 22037
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.228332469229574e-06,
+      "loss": 1.0019,
+      "step": 22038
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.22804094646105e-06,
+      "loss": 0.9843,
+      "step": 22039
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2277496077654018e-06,
+      "loss": 1.0584,
+      "step": 22040
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.227458453148117e-06,
+      "loss": 0.9728,
+      "step": 22041
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.227167482614681e-06,
+      "loss": 0.9963,
+      "step": 22042
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2268766961705738e-06,
+      "loss": 1.0279,
+      "step": 22043
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2265860938212737e-06,
+      "loss": 1.0098,
+      "step": 22044
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2262956755722545e-06,
+      "loss": 0.8902,
+      "step": 22045
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.226005441428986e-06,
+      "loss": 0.9761,
+      "step": 22046
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2257153913969367e-06,
+      "loss": 0.9385,
+      "step": 22047
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2254255254815696e-06,
+      "loss": 1.0003,
+      "step": 22048
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2251358436883445e-06,
+      "loss": 0.9751,
+      "step": 22049
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.224846346022718e-06,
+      "loss": 1.0211,
+      "step": 22050
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2245570324901435e-06,
+      "loss": 0.9356,
+      "step": 22051
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2242679030960723e-06,
+      "loss": 0.9594,
+      "step": 22052
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2239789578459483e-06,
+      "loss": 0.9819,
+      "step": 22053
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2236901967452154e-06,
+      "loss": 0.9475,
+      "step": 22054
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.223401619799314e-06,
+      "loss": 0.997,
+      "step": 22055
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.223113227013677e-06,
+      "loss": 0.9796,
+      "step": 22056
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2228250183937395e-06,
+      "loss": 0.9996,
+      "step": 22057
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2225369939449306e-06,
+      "loss": 0.9587,
+      "step": 22058
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2222491536726746e-06,
+      "loss": 1.0345,
+      "step": 22059
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.221961497582391e-06,
+      "loss": 0.9825,
+      "step": 22060
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2216740256795033e-06,
+      "loss": 0.8694,
+      "step": 22061
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2213867379694246e-06,
+      "loss": 0.9526,
+      "step": 22062
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2210996344575655e-06,
+      "loss": 1.0295,
+      "step": 22063
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2208127151493353e-06,
+      "loss": 0.972,
+      "step": 22064
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.220525980050138e-06,
+      "loss": 0.9097,
+      "step": 22065
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.220239429165374e-06,
+      "loss": 1.0331,
+      "step": 22066
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.219953062500443e-06,
+      "loss": 0.9944,
+      "step": 22067
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.219666880060738e-06,
+      "loss": 1.0473,
+      "step": 22068
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2193808818516496e-06,
+      "loss": 1.0634,
+      "step": 22069
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2190950678785676e-06,
+      "loss": 1.0062,
+      "step": 22070
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.218809438146871e-06,
+      "loss": 0.9968,
+      "step": 22071
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.218523992661943e-06,
+      "loss": 0.9674,
+      "step": 22072
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.218238731429162e-06,
+      "loss": 0.9223,
+      "step": 22073
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2179536544538995e-06,
+      "loss": 0.9926,
+      "step": 22074
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.217668761741524e-06,
+      "loss": 0.9718,
+      "step": 22075
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2173840532974058e-06,
+      "loss": 1.0578,
+      "step": 22076
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2170995291269052e-06,
+      "loss": 0.9972,
+      "step": 22077
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.216815189235383e-06,
+      "loss": 0.984,
+      "step": 22078
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2165310336281923e-06,
+      "loss": 1.0198,
+      "step": 22079
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2162470623106895e-06,
+      "loss": 1.0179,
+      "step": 22080
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.215963275288223e-06,
+      "loss": 0.9871,
+      "step": 22081
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.215679672566138e-06,
+      "loss": 1.0037,
+      "step": 22082
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2153962541497742e-06,
+      "loss": 0.9986,
+      "step": 22083
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.215113020044474e-06,
+      "loss": 0.9166,
+      "step": 22084
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.214829970255571e-06,
+      "loss": 0.8513,
+      "step": 22085
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2145471047883974e-06,
+      "loss": 0.9336,
+      "step": 22086
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2142644236482804e-06,
+      "loss": 0.9704,
+      "step": 22087
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.213981926840546e-06,
+      "loss": 0.9694,
+      "step": 22088
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.213699614370516e-06,
+      "loss": 0.9385,
+      "step": 22089
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2134174862435067e-06,
+      "loss": 0.946,
+      "step": 22090
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2131355424648318e-06,
+      "loss": 0.9441,
+      "step": 22091
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2128537830398057e-06,
+      "loss": 1.0309,
+      "step": 22092
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2125722079737333e-06,
+      "loss": 1.0121,
+      "step": 22093
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2122908172719183e-06,
+      "loss": 0.9415,
+      "step": 22094
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.212009610939663e-06,
+      "loss": 1.0357,
+      "step": 22095
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2117285889822637e-06,
+      "loss": 1.0095,
+      "step": 22096
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.211447751405013e-06,
+      "loss": 1.0451,
+      "step": 22097
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2111670982132007e-06,
+      "loss": 1.0168,
+      "step": 22098
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.210886629412117e-06,
+      "loss": 1.0753,
+      "step": 22099
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2106063450070415e-06,
+      "loss": 0.9895,
+      "step": 22100
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2103262450032548e-06,
+      "loss": 1.0027,
+      "step": 22101
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.210046329406032e-06,
+      "loss": 1.0116,
+      "step": 22102
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2097665982206484e-06,
+      "loss": 1.0059,
+      "step": 22103
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2094870514523714e-06,
+      "loss": 0.998,
+      "step": 22104
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2092076891064676e-06,
+      "loss": 0.9463,
+      "step": 22105
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2089285111881975e-06,
+      "loss": 0.9915,
+      "step": 22106
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2086495177028233e-06,
+      "loss": 0.952,
+      "step": 22107
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2083707086555975e-06,
+      "loss": 1.0341,
+      "step": 22108
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.208092084051773e-06,
+      "loss": 0.9539,
+      "step": 22109
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2078136438965982e-06,
+      "loss": 0.9132,
+      "step": 22110
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.207535388195318e-06,
+      "loss": 0.9916,
+      "step": 22111
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2072573169531744e-06,
+      "loss": 1.0139,
+      "step": 22112
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2069794301754035e-06,
+      "loss": 1.0258,
+      "step": 22113
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2067017278672423e-06,
+      "loss": 0.9947,
+      "step": 22114
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2064242100339195e-06,
+      "loss": 0.96,
+      "step": 22115
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.206146876680664e-06,
+      "loss": 0.9738,
+      "step": 22116
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.205869727812701e-06,
+      "loss": 0.9252,
+      "step": 22117
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.205592763435248e-06,
+      "loss": 0.9833,
+      "step": 22118
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2053159835535254e-06,
+      "loss": 0.996,
+      "step": 22119
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.205039388172744e-06,
+      "loss": 0.9392,
+      "step": 22120
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.204762977298116e-06,
+      "loss": 1.0522,
+      "step": 22121
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.204486750934848e-06,
+      "loss": 0.9429,
+      "step": 22122
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.204210709088143e-06,
+      "loss": 1.0335,
+      "step": 22123
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2039348517632e-06,
+      "loss": 0.9574,
+      "step": 22124
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2036591789652153e-06,
+      "loss": 1.0139,
+      "step": 22125
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2033836906993823e-06,
+      "loss": 0.955,
+      "step": 22126
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2031083869708906e-06,
+      "loss": 0.9865,
+      "step": 22127
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2028332677849255e-06,
+      "loss": 1.0608,
+      "step": 22128
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2025583331466695e-06,
+      "loss": 1.0016,
+      "step": 22129
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.202283583061301e-06,
+      "loss": 0.9901,
+      "step": 22130
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2020090175339974e-06,
+      "loss": 0.9455,
+      "step": 22131
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2017346365699277e-06,
+      "loss": 0.9482,
+      "step": 22132
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2014604401742623e-06,
+      "loss": 1.0008,
+      "step": 22133
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.201186428352166e-06,
+      "loss": 1.0347,
+      "step": 22134
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2009126011088004e-06,
+      "loss": 0.9613,
+      "step": 22135
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2006389584493216e-06,
+      "loss": 0.9721,
+      "step": 22136
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.200365500378887e-06,
+      "loss": 0.9466,
+      "step": 22137
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2000922269026473e-06,
+      "loss": 1.0031,
+      "step": 22138
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1998191380257484e-06,
+      "loss": 0.9928,
+      "step": 22139
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1995462337533348e-06,
+      "loss": 1.0414,
+      "step": 22140
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.199273514090549e-06,
+      "loss": 1.0907,
+      "step": 22141
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.199000979042525e-06,
+      "loss": 0.9724,
+      "step": 22142
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1987286286144007e-06,
+      "loss": 0.9782,
+      "step": 22143
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1984564628113017e-06,
+      "loss": 0.9098,
+      "step": 22144
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1981844816383582e-06,
+      "loss": 1.012,
+      "step": 22145
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.197912685100692e-06,
+      "loss": 0.9658,
+      "step": 22146
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.197641073203424e-06,
+      "loss": 1.0064,
+      "step": 22147
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.197369645951668e-06,
+      "loss": 1.0183,
+      "step": 22148
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1970984033505394e-06,
+      "loss": 0.9262,
+      "step": 22149
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.196827345405147e-06,
+      "loss": 0.9756,
+      "step": 22150
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.196556472120597e-06,
+      "loss": 0.9991,
+      "step": 22151
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1962857835019892e-06,
+      "loss": 0.9295,
+      "step": 22152
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1960152795544257e-06,
+      "loss": 0.9783,
+      "step": 22153
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.195744960283002e-06,
+      "loss": 0.9501,
+      "step": 22154
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.195474825692806e-06,
+      "loss": 0.9584,
+      "step": 22155
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.19520487578893e-06,
+      "loss": 0.9595,
+      "step": 22156
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1949351105764585e-06,
+      "loss": 0.9982,
+      "step": 22157
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1946655300604735e-06,
+      "loss": 1.0076,
+      "step": 22158
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.19439613424605e-06,
+      "loss": 0.9754,
+      "step": 22159
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1941269231382657e-06,
+      "loss": 0.9383,
+      "step": 22160
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.193857896742191e-06,
+      "loss": 1.096,
+      "step": 22161
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1935890550628925e-06,
+      "loss": 0.9609,
+      "step": 22162
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.193320398105433e-06,
+      "loss": 1.0028,
+      "step": 22163
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1930519258748773e-06,
+      "loss": 0.967,
+      "step": 22164
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1927836383762814e-06,
+      "loss": 0.9818,
+      "step": 22165
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1925155356146955e-06,
+      "loss": 0.8957,
+      "step": 22166
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1922476175951726e-06,
+      "loss": 0.9819,
+      "step": 22167
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1919798843227604e-06,
+      "loss": 0.9728,
+      "step": 22168
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1917123358025007e-06,
+      "loss": 0.939,
+      "step": 22169
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1914449720394327e-06,
+      "loss": 0.9879,
+      "step": 22170
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1911777930385926e-06,
+      "loss": 1.0046,
+      "step": 22171
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1909107988050164e-06,
+      "loss": 0.9347,
+      "step": 22172
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.190643989343728e-06,
+      "loss": 0.9997,
+      "step": 22173
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1903773646597585e-06,
+      "loss": 1.0047,
+      "step": 22174
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1901109247581265e-06,
+      "loss": 0.9841,
+      "step": 22175
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.189844669643854e-06,
+      "loss": 1.0045,
+      "step": 22176
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1895785993219544e-06,
+      "loss": 1.0386,
+      "step": 22177
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1893127137974394e-06,
+      "loss": 0.8659,
+      "step": 22178
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1890470130753172e-06,
+      "loss": 0.8436,
+      "step": 22179
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.188781497160596e-06,
+      "loss": 0.9458,
+      "step": 22180
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1885161660582745e-06,
+      "loss": 1.0632,
+      "step": 22181
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.188251019773349e-06,
+      "loss": 0.9569,
+      "step": 22182
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1879860583108184e-06,
+      "loss": 1.0813,
+      "step": 22183
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1877212816756712e-06,
+      "loss": 0.9693,
+      "step": 22184
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1874566898728947e-06,
+      "loss": 0.9965,
+      "step": 22185
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1871922829074736e-06,
+      "loss": 1.0282,
+      "step": 22186
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1869280607843893e-06,
+      "loss": 0.8923,
+      "step": 22187
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.186664023508617e-06,
+      "loss": 0.9165,
+      "step": 22188
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1864001710851313e-06,
+      "loss": 0.9838,
+      "step": 22189
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.186136503518902e-06,
+      "loss": 1.0047,
+      "step": 22190
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1858730208148974e-06,
+      "loss": 0.9612,
+      "step": 22191
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1856097229780793e-06,
+      "loss": 0.9401,
+      "step": 22192
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.185346610013407e-06,
+      "loss": 1.0328,
+      "step": 22193
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1850836819258384e-06,
+      "loss": 1.0026,
+      "step": 22194
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1848209387203238e-06,
+      "loss": 1.0042,
+      "step": 22195
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1845583804018134e-06,
+      "loss": 0.9526,
+      "step": 22196
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1842960069752552e-06,
+      "loss": 1.0143,
+      "step": 22197
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.184033818445588e-06,
+      "loss": 1.0335,
+      "step": 22198
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1837718148177533e-06,
+      "loss": 0.985,
+      "step": 22199
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1835099960966843e-06,
+      "loss": 0.9632,
+      "step": 22200
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.183248362287315e-06,
+      "loss": 0.9379,
+      "step": 22201
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1829869133945716e-06,
+      "loss": 0.9493,
+      "step": 22202
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1827256494233806e-06,
+      "loss": 0.9318,
+      "step": 22203
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.182464570378663e-06,
+      "loss": 0.9591,
+      "step": 22204
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1822036762653367e-06,
+      "loss": 1.0471,
+      "step": 22205
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1819429670883154e-06,
+      "loss": 0.9742,
+      "step": 22206
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.181682442852511e-06,
+      "loss": 1.044,
+      "step": 22207
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1814221035628295e-06,
+      "loss": 0.9729,
+      "step": 22208
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1811619492241765e-06,
+      "loss": 0.9772,
+      "step": 22209
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1809019798414525e-06,
+      "loss": 1.0113,
+      "step": 22210
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1806421954195534e-06,
+      "loss": 1.0158,
+      "step": 22211
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.180382595963373e-06,
+      "loss": 0.9952,
+      "step": 22212
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1801231814778026e-06,
+      "loss": 0.9967,
+      "step": 22213
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1798639519677257e-06,
+      "loss": 1.0079,
+      "step": 22214
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.179604907438029e-06,
+      "loss": 0.9574,
+      "step": 22215
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1793460478935904e-06,
+      "loss": 1.0169,
+      "step": 22216
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.179087373339285e-06,
+      "loss": 1.0113,
+      "step": 22217
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.178828883779988e-06,
+      "loss": 1.007,
+      "step": 22218
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.178570579220566e-06,
+      "loss": 0.9548,
+      "step": 22219
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1783124596658856e-06,
+      "loss": 0.9649,
+      "step": 22220
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1780545251208086e-06,
+      "loss": 0.9892,
+      "step": 22221
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1777967755901943e-06,
+      "loss": 0.9772,
+      "step": 22222
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1775392110788985e-06,
+      "loss": 1.0418,
+      "step": 22223
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1772818315917713e-06,
+      "loss": 1.0169,
+      "step": 22224
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1770246371336614e-06,
+      "loss": 1.0196,
+      "step": 22225
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1767676277094143e-06,
+      "loss": 1.0148,
+      "step": 22226
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1765108033238717e-06,
+      "loss": 1.0128,
+      "step": 22227
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.176254163981868e-06,
+      "loss": 0.9974,
+      "step": 22228
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1759977096882414e-06,
+      "loss": 0.9798,
+      "step": 22229
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1757414404478206e-06,
+      "loss": 0.9751,
+      "step": 22230
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1754853562654356e-06,
+      "loss": 0.9724,
+      "step": 22231
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1752294571459056e-06,
+      "loss": 0.9333,
+      "step": 22232
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1749737430940533e-06,
+      "loss": 0.9384,
+      "step": 22233
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1747182141146975e-06,
+      "loss": 0.9582,
+      "step": 22234
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.174462870212649e-06,
+      "loss": 0.9638,
+      "step": 22235
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1742077113927175e-06,
+      "loss": 0.9575,
+      "step": 22236
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1739527376597106e-06,
+      "loss": 0.9557,
+      "step": 22237
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.173697949018432e-06,
+      "loss": 0.9756,
+      "step": 22238
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1734433454736784e-06,
+      "loss": 0.9917,
+      "step": 22239
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1731889270302488e-06,
+      "loss": 0.8997,
+      "step": 22240
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.172934693692933e-06,
+      "loss": 1.0143,
+      "step": 22241
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.172680645466523e-06,
+      "loss": 1.0251,
+      "step": 22242
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1724267823558e-06,
+      "loss": 1.0103,
+      "step": 22243
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1721731043655493e-06,
+      "loss": 0.9726,
+      "step": 22244
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1719196115005495e-06,
+      "loss": 0.9293,
+      "step": 22245
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.171666303765574e-06,
+      "loss": 0.8734,
+      "step": 22246
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.171413181165393e-06,
+      "loss": 1.0135,
+      "step": 22247
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.171160243704779e-06,
+      "loss": 0.9205,
+      "step": 22248
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1709074913884934e-06,
+      "loss": 0.9345,
+      "step": 22249
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1706549242212973e-06,
+      "loss": 1.0129,
+      "step": 22250
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.170402542207948e-06,
+      "loss": 0.9741,
+      "step": 22251
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1701503453532015e-06,
+      "loss": 0.9859,
+      "step": 22252
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1698983336618076e-06,
+      "loss": 0.981,
+      "step": 22253
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.169646507138512e-06,
+      "loss": 0.9914,
+      "step": 22254
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1693948657880594e-06,
+      "loss": 1.0223,
+      "step": 22255
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.169143409615191e-06,
+      "loss": 0.9969,
+      "step": 22256
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.168892138624642e-06,
+      "loss": 0.9669,
+      "step": 22257
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1686410528211456e-06,
+      "loss": 0.8959,
+      "step": 22258
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.168390152209431e-06,
+      "loss": 0.9076,
+      "step": 22259
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1681394367942267e-06,
+      "loss": 0.9398,
+      "step": 22260
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1678889065802525e-06,
+      "loss": 0.9976,
+      "step": 22261
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.16763856157223e-06,
+      "loss": 0.8784,
+      "step": 22262
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.167388401774873e-06,
+      "loss": 0.9797,
+      "step": 22263
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1671384271928954e-06,
+      "loss": 0.9647,
+      "step": 22264
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1668886378310043e-06,
+      "loss": 0.9588,
+      "step": 22265
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.166639033693906e-06,
+      "loss": 1.037,
+      "step": 22266
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.166389614786302e-06,
+      "loss": 0.976,
+      "step": 22267
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.16614038111289e-06,
+      "loss": 1.0059,
+      "step": 22268
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.165891332678366e-06,
+      "loss": 1.0186,
+      "step": 22269
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1656424694874203e-06,
+      "loss": 0.9352,
+      "step": 22270
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1653937915447422e-06,
+      "loss": 0.9643,
+      "step": 22271
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1651452988550137e-06,
+      "loss": 1.035,
+      "step": 22272
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1648969914229168e-06,
+      "loss": 1.0601,
+      "step": 22273
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1646488692531287e-06,
+      "loss": 0.9619,
+      "step": 22274
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1644009323503248e-06,
+      "loss": 0.9478,
+      "step": 22275
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.164153180719172e-06,
+      "loss": 0.9527,
+      "step": 22276
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.16390561436434e-06,
+      "loss": 0.9773,
+      "step": 22277
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1636582332904914e-06,
+      "loss": 0.9101,
+      "step": 22278
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1634110375022855e-06,
+      "loss": 0.9885,
+      "step": 22279
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1631640270043787e-06,
+      "loss": 0.9954,
+      "step": 22280
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.162917201801426e-06,
+      "loss": 1.0181,
+      "step": 22281
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1626705618980738e-06,
+      "loss": 0.9511,
+      "step": 22282
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1624241072989697e-06,
+      "loss": 1.0434,
+      "step": 22283
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1621778380087555e-06,
+      "loss": 0.9642,
+      "step": 22284
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1619317540320706e-06,
+      "loss": 0.9796,
+      "step": 22285
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1616858553735513e-06,
+      "loss": 0.9776,
+      "step": 22286
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1614401420378275e-06,
+      "loss": 0.9048,
+      "step": 22287
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1611946140295296e-06,
+      "loss": 0.9858,
+      "step": 22288
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.160949271353281e-06,
+      "loss": 0.9496,
+      "step": 22289
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.160704114013703e-06,
+      "loss": 0.9339,
+      "step": 22290
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1604591420154153e-06,
+      "loss": 0.962,
+      "step": 22291
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1602143553630317e-06,
+      "loss": 0.9606,
+      "step": 22292
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1599697540611632e-06,
+      "loss": 0.9976,
+      "step": 22293
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.159725338114417e-06,
+      "loss": 0.9495,
+      "step": 22294
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1594811075273983e-06,
+      "loss": 1.0036,
+      "step": 22295
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1592370623047044e-06,
+      "loss": 1.0307,
+      "step": 22296
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1589932024509367e-06,
+      "loss": 0.9764,
+      "step": 22297
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1587495279706853e-06,
+      "loss": 0.9742,
+      "step": 22298
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1585060388685423e-06,
+      "loss": 1.007,
+      "step": 22299
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.158262735149093e-06,
+      "loss": 0.9587,
+      "step": 22300
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1580196168169213e-06,
+      "loss": 0.9701,
+      "step": 22301
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.157776683876607e-06,
+      "loss": 0.977,
+      "step": 22302
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1575339363327255e-06,
+      "loss": 0.9596,
+      "step": 22303
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1572913741898484e-06,
+      "loss": 0.9622,
+      "step": 22304
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1570489974525477e-06,
+      "loss": 1.0126,
+      "step": 22305
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.156806806125387e-06,
+      "loss": 0.9826,
+      "step": 22306
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.156564800212927e-06,
+      "loss": 0.9709,
+      "step": 22307
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1563229797197294e-06,
+      "loss": 1.026,
+      "step": 22308
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.156081344650348e-06,
+      "loss": 0.9841,
+      "step": 22309
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1558398950093353e-06,
+      "loss": 0.9601,
+      "step": 22310
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1555986308012378e-06,
+      "loss": 0.9327,
+      "step": 22311
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1553575520306007e-06,
+      "loss": 0.9932,
+      "step": 22312
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.155116658701966e-06,
+      "loss": 0.9865,
+      "step": 22313
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.154875950819869e-06,
+      "loss": 0.9443,
+      "step": 22314
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1546354283888485e-06,
+      "loss": 0.9993,
+      "step": 22315
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1543950914134306e-06,
+      "loss": 1.0147,
+      "step": 22316
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1541549398981444e-06,
+      "loss": 1.0406,
+      "step": 22317
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.153914973847513e-06,
+      "loss": 1.0333,
+      "step": 22318
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1536751932660585e-06,
+      "loss": 1.0078,
+      "step": 22319
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1534355981582945e-06,
+      "loss": 0.9653,
+      "step": 22320
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1531961885287367e-06,
+      "loss": 0.9498,
+      "step": 22321
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.152956964381895e-06,
+      "loss": 0.9184,
+      "step": 22322
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1527179257222733e-06,
+      "loss": 1.0336,
+      "step": 22323
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1524790725543746e-06,
+      "loss": 1.0527,
+      "step": 22324
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1522404048827006e-06,
+      "loss": 1.0495,
+      "step": 22325
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1520019227117464e-06,
+      "loss": 1.0324,
+      "step": 22326
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.151763626046002e-06,
+      "loss": 1.0295,
+      "step": 22327
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1515255148899585e-06,
+      "loss": 0.9988,
+      "step": 22328
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1512875892481e-06,
+      "loss": 0.9838,
+      "step": 22329
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.151049849124909e-06,
+      "loss": 0.9892,
+      "step": 22330
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1508122945248624e-06,
+      "loss": 0.9306,
+      "step": 22331
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1505749254524357e-06,
+      "loss": 1.0084,
+      "step": 22332
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.150337741912102e-06,
+      "loss": 0.9412,
+      "step": 22333
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1501007439083266e-06,
+      "loss": 1.0387,
+      "step": 22334
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.149863931445574e-06,
+      "loss": 0.923,
+      "step": 22335
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.149627304528307e-06,
+      "loss": 0.9469,
+      "step": 22336
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1493908631609817e-06,
+      "loss": 1.0351,
+      "step": 22337
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.149154607348051e-06,
+      "loss": 1.0233,
+      "step": 22338
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.148918537093965e-06,
+      "loss": 0.9935,
+      "step": 22339
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1486826524031737e-06,
+      "loss": 1.0069,
+      "step": 22340
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1484469532801168e-06,
+      "loss": 0.9004,
+      "step": 22341
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1482114397292366e-06,
+      "loss": 0.9285,
+      "step": 22342
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.147976111754967e-06,
+      "loss": 1.0382,
+      "step": 22343
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1477409693617437e-06,
+      "loss": 0.9691,
+      "step": 22344
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.147506012553994e-06,
+      "loss": 0.9553,
+      "step": 22345
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.147271241336144e-06,
+      "loss": 0.9627,
+      "step": 22346
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.147036655712617e-06,
+      "loss": 0.9673,
+      "step": 22347
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.146802255687831e-06,
+      "loss": 1.0176,
+      "step": 22348
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1465680412662016e-06,
+      "loss": 1.0656,
+      "step": 22349
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1463340124521405e-06,
+      "loss": 0.9347,
+      "step": 22350
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1461001692500574e-06,
+      "loss": 1.0324,
+      "step": 22351
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.145866511664354e-06,
+      "loss": 0.9721,
+      "step": 22352
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1456330396994353e-06,
+      "loss": 0.9915,
+      "step": 22353
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1453997533596973e-06,
+      "loss": 0.9798,
+      "step": 22354
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.145166652649534e-06,
+      "loss": 0.994,
+      "step": 22355
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1449337375733374e-06,
+      "loss": 1.0185,
+      "step": 22356
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1447010081354938e-06,
+      "loss": 1.0002,
+      "step": 22357
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.144468464340387e-06,
+      "loss": 0.9798,
+      "step": 22358
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1442361061924e-06,
+      "loss": 1.0063,
+      "step": 22359
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1440039336959065e-06,
+      "loss": 0.9333,
+      "step": 22360
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1437719468552824e-06,
+      "loss": 0.9587,
+      "step": 22361
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1435401456748952e-06,
+      "loss": 0.9703,
+      "step": 22362
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.143308530159113e-06,
+      "loss": 1.0149,
+      "step": 22363
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1430771003122975e-06,
+      "loss": 0.9674,
+      "step": 22364
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.142845856138809e-06,
+      "loss": 1.0541,
+      "step": 22365
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1426147976430037e-06,
+      "loss": 1.044,
+      "step": 22366
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1423839248292335e-06,
+      "loss": 1.0152,
+      "step": 22367
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1421532377018465e-06,
+      "loss": 1.1099,
+      "step": 22368
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1419227362651896e-06,
+      "loss": 0.9871,
+      "step": 22369
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.141692420523604e-06,
+      "loss": 1.0085,
+      "step": 22370
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1414622904814276e-06,
+      "loss": 1.0349,
+      "step": 22371
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1412323461429963e-06,
+      "loss": 0.9462,
+      "step": 22372
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.141002587512642e-06,
+      "loss": 0.9547,
+      "step": 22373
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1407730145946904e-06,
+      "loss": 0.9846,
+      "step": 22374
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.140543627393468e-06,
+      "loss": 0.9722,
+      "step": 22375
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.140314425913295e-06,
+      "loss": 1.0416,
+      "step": 22376
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.140085410158488e-06,
+      "loss": 0.9609,
+      "step": 22377
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.139856580133363e-06,
+      "loss": 0.9407,
+      "step": 22378
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.139627935842228e-06,
+      "loss": 1.0422,
+      "step": 22379
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1393994772893925e-06,
+      "loss": 0.9966,
+      "step": 22380
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.139171204479158e-06,
+      "loss": 1.008,
+      "step": 22381
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.138943117415825e-06,
+      "loss": 0.9544,
+      "step": 22382
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1387152161036896e-06,
+      "loss": 1.0191,
+      "step": 22383
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.138487500547045e-06,
+      "loss": 0.9239,
+      "step": 22384
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1382599707501817e-06,
+      "loss": 0.983,
+      "step": 22385
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.138032626717385e-06,
+      "loss": 0.9448,
+      "step": 22386
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.137805468452935e-06,
+      "loss": 0.9673,
+      "step": 22387
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.137578495961115e-06,
+      "loss": 0.9997,
+      "step": 22388
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.137351709246196e-06,
+      "loss": 0.9539,
+      "step": 22389
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1371251083124534e-06,
+      "loss": 1.0486,
+      "step": 22390
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1368986931641546e-06,
+      "loss": 0.9752,
+      "step": 22391
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.136672463805564e-06,
+      "loss": 0.9559,
+      "step": 22392
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.136446420240942e-06,
+      "loss": 0.9203,
+      "step": 22393
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.136220562474549e-06,
+      "loss": 0.9761,
+      "step": 22394
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1359948905106387e-06,
+      "loss": 0.9436,
+      "step": 22395
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.135769404353461e-06,
+      "loss": 1.0047,
+      "step": 22396
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1355441040072646e-06,
+      "loss": 0.9733,
+      "step": 22397
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.135318989476293e-06,
+      "loss": 1.0569,
+      "step": 22398
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1350940607647866e-06,
+      "loss": 0.9981,
+      "step": 22399
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.13486931787698e-06,
+      "loss": 0.9881,
+      "step": 22400
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1346447608171107e-06,
+      "loss": 0.9843,
+      "step": 22401
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.134420389589407e-06,
+      "loss": 1.0306,
+      "step": 22402
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.134196204198096e-06,
+      "loss": 0.9759,
+      "step": 22403
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1339722046473976e-06,
+      "loss": 0.9785,
+      "step": 22404
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1337483909415347e-06,
+      "loss": 0.9533,
+      "step": 22405
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1335247630847225e-06,
+      "loss": 0.9713,
+      "step": 22406
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.133301321081173e-06,
+      "loss": 0.9389,
+      "step": 22407
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.133078064935094e-06,
+      "loss": 0.9992,
+      "step": 22408
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.132854994650692e-06,
+      "loss": 0.9893,
+      "step": 22409
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1326321102321704e-06,
+      "loss": 0.9688,
+      "step": 22410
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1324094116837246e-06,
+      "loss": 0.9028,
+      "step": 22411
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.132186899009552e-06,
+      "loss": 0.9401,
+      "step": 22412
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.131964572213844e-06,
+      "loss": 0.9114,
+      "step": 22413
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.131742431300786e-06,
+      "loss": 0.9889,
+      "step": 22414
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1315204762745657e-06,
+      "loss": 0.9695,
+      "step": 22415
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.131298707139361e-06,
+      "loss": 0.9944,
+      "step": 22416
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1310771238993533e-06,
+      "loss": 1.0433,
+      "step": 22417
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1308557265587117e-06,
+      "loss": 0.976,
+      "step": 22418
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1306345151216093e-06,
+      "loss": 1.024,
+      "step": 22419
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1304134895922138e-06,
+      "loss": 0.9755,
+      "step": 22420
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1301926499746874e-06,
+      "loss": 1.0237,
+      "step": 22421
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1299719962731906e-06,
+      "loss": 1.1239,
+      "step": 22422
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.129751528491877e-06,
+      "loss": 0.9217,
+      "step": 22423
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.129531246634904e-06,
+      "loss": 0.9778,
+      "step": 22424
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1293111507064192e-06,
+      "loss": 0.9828,
+      "step": 22425
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.129091240710567e-06,
+      "loss": 0.952,
+      "step": 22426
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.128871516651491e-06,
+      "loss": 0.9273,
+      "step": 22427
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.128651978533332e-06,
+      "loss": 1.0215,
+      "step": 22428
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.128432626360222e-06,
+      "loss": 1.0269,
+      "step": 22429
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.128213460136294e-06,
+      "loss": 0.9894,
+      "step": 22430
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.127994479865677e-06,
+      "loss": 0.999,
+      "step": 22431
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1277756855524967e-06,
+      "loss": 0.8995,
+      "step": 22432
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1275570772008737e-06,
+      "loss": 0.9843,
+      "step": 22433
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1273386548149247e-06,
+      "loss": 0.969,
+      "step": 22434
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1271204183987663e-06,
+      "loss": 0.9768,
+      "step": 22435
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1269023679565074e-06,
+      "loss": 1.0111,
+      "step": 22436
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.126684503492256e-06,
+      "loss": 0.9347,
+      "step": 22437
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1264668250101163e-06,
+      "loss": 0.9931,
+      "step": 22438
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1262493325141887e-06,
+      "loss": 0.9811,
+      "step": 22439
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.12603202600857e-06,
+      "loss": 0.9841,
+      "step": 22440
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1258149054973534e-06,
+      "loss": 0.971,
+      "step": 22441
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.125597970984629e-06,
+      "loss": 0.9807,
+      "step": 22442
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.125381222474483e-06,
+      "loss": 0.9297,
+      "step": 22443
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1251646599709986e-06,
+      "loss": 0.9951,
+      "step": 22444
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1249482834782544e-06,
+      "loss": 0.9041,
+      "step": 22445
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.124732093000327e-06,
+      "loss": 0.9622,
+      "step": 22446
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1245160885412876e-06,
+      "loss": 0.9457,
+      "step": 22447
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1243002701052062e-06,
+      "loss": 0.9591,
+      "step": 22448
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.124084637696149e-06,
+      "loss": 1.0007,
+      "step": 22449
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1238691913181755e-06,
+      "loss": 0.9927,
+      "step": 22450
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.123653930975346e-06,
+      "loss": 0.9659,
+      "step": 22451
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1234388566717134e-06,
+      "loss": 0.9676,
+      "step": 22452
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.123223968411331e-06,
+      "loss": 0.9822,
+      "step": 22453
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.123009266198245e-06,
+      "loss": 0.9278,
+      "step": 22454
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1227947500365012e-06,
+      "loss": 0.9317,
+      "step": 22455
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.122580419930139e-06,
+      "loss": 1.0226,
+      "step": 22456
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1223662758831965e-06,
+      "loss": 1.0154,
+      "step": 22457
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1221523178997073e-06,
+      "loss": 0.9948,
+      "step": 22458
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.121938545983703e-06,
+      "loss": 0.9814,
+      "step": 22459
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1217249601392076e-06,
+      "loss": 0.9981,
+      "step": 22460
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1215115603702465e-06,
+      "loss": 0.8834,
+      "step": 22461
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1212983466808392e-06,
+      "loss": 0.9189,
+      "step": 22462
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1210853190750015e-06,
+      "loss": 1.0262,
+      "step": 22463
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1208724775567458e-06,
+      "loss": 1.0538,
+      "step": 22464
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1206598221300823e-06,
+      "loss": 1.0051,
+      "step": 22465
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1204473527990166e-06,
+      "loss": 0.8696,
+      "step": 22466
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1202350695675503e-06,
+      "loss": 0.988,
+      "step": 22467
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1200229724396837e-06,
+      "loss": 0.9613,
+      "step": 22468
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.119811061419409e-06,
+      "loss": 0.9915,
+      "step": 22469
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1195993365107203e-06,
+      "loss": 1.0111,
+      "step": 22470
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.119387797717606e-06,
+      "loss": 0.9865,
+      "step": 22471
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1191764450440503e-06,
+      "loss": 0.9349,
+      "step": 22472
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.118965278494034e-06,
+      "loss": 0.994,
+      "step": 22473
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.118754298071536e-06,
+      "loss": 0.851,
+      "step": 22474
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1185435037805282e-06,
+      "loss": 0.9473,
+      "step": 22475
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.118332895624984e-06,
+      "loss": 0.98,
+      "step": 22476
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1181224736088682e-06,
+      "loss": 0.9526,
+      "step": 22477
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.117912237736146e-06,
+      "loss": 0.9872,
+      "step": 22478
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1177021880107785e-06,
+      "loss": 0.8778,
+      "step": 22479
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1174923244367185e-06,
+      "loss": 0.9466,
+      "step": 22480
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.117282647017923e-06,
+      "loss": 0.9973,
+      "step": 22481
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1170731557583403e-06,
+      "loss": 0.9703,
+      "step": 22482
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.116863850661918e-06,
+      "loss": 0.9218,
+      "step": 22483
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.116654731732596e-06,
+      "loss": 0.9324,
+      "step": 22484
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.116445798974315e-06,
+      "loss": 1.0188,
+      "step": 22485
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1162370523910106e-06,
+      "loss": 0.8914,
+      "step": 22486
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.116028491986615e-06,
+      "loss": 0.9784,
+      "step": 22487
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.115820117765055e-06,
+      "loss": 1.0135,
+      "step": 22488
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1156119297302594e-06,
+      "loss": 1.0184,
+      "step": 22489
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1154039278861473e-06,
+      "loss": 0.9237,
+      "step": 22490
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1151961122366373e-06,
+      "loss": 1.0031,
+      "step": 22491
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.114988482785643e-06,
+      "loss": 0.9833,
+      "step": 22492
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.114781039537076e-06,
+      "loss": 0.9829,
+      "step": 22493
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.114573782494847e-06,
+      "loss": 0.9995,
+      "step": 22494
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1143667116628555e-06,
+      "loss": 0.9575,
+      "step": 22495
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.114159827045003e-06,
+      "loss": 0.9553,
+      "step": 22496
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.113953128645189e-06,
+      "loss": 0.9721,
+      "step": 22497
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.113746616467305e-06,
+      "loss": 1.0086,
+      "step": 22498
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1135402905152417e-06,
+      "loss": 0.9334,
+      "step": 22499
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.113334150792884e-06,
+      "loss": 0.9769,
+      "step": 22500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1131281973041185e-06,
+      "loss": 0.996,
+      "step": 22501
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.112922430052822e-06,
+      "loss": 1.0015,
+      "step": 22502
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.112716849042871e-06,
+      "loss": 0.9644,
+      "step": 22503
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.112511454278138e-06,
+      "loss": 0.9434,
+      "step": 22504
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1123062457624925e-06,
+      "loss": 0.9571,
+      "step": 22505
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1121012234998e-06,
+      "loss": 0.9119,
+      "step": 22506
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1118963874939195e-06,
+      "loss": 1.0291,
+      "step": 22507
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.111691737748715e-06,
+      "loss": 0.9717,
+      "step": 22508
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.111487274268037e-06,
+      "loss": 1.0314,
+      "step": 22509
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1112829970557395e-06,
+      "loss": 0.9704,
+      "step": 22510
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.111078906115667e-06,
+      "loss": 0.9433,
+      "step": 22511
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.110875001451669e-06,
+      "loss": 0.9019,
+      "step": 22512
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1106712830675827e-06,
+      "loss": 0.9573,
+      "step": 22513
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.110467750967247e-06,
+      "loss": 1.0514,
+      "step": 22514
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.110264405154493e-06,
+      "loss": 0.9185,
+      "step": 22515
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.110061245633156e-06,
+      "loss": 1.0006,
+      "step": 22516
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1098582724070596e-06,
+      "loss": 0.994,
+      "step": 22517
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.109655485480028e-06,
+      "loss": 0.9161,
+      "step": 22518
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1094528848558807e-06,
+      "loss": 0.9752,
+      "step": 22519
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1092504705384343e-06,
+      "loss": 1.0036,
+      "step": 22520
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.109048242531502e-06,
+      "loss": 0.9691,
+      "step": 22521
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1088462008388925e-06,
+      "loss": 1.0046,
+      "step": 22522
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1086443454644115e-06,
+      "loss": 1.0435,
+      "step": 22523
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.108442676411863e-06,
+      "loss": 0.924,
+      "step": 22524
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1082411936850436e-06,
+      "loss": 0.9577,
+      "step": 22525
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1080398972877496e-06,
+      "loss": 0.9965,
+      "step": 22526
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1078387872237727e-06,
+      "loss": 0.9786,
+      "step": 22527
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.107637863496901e-06,
+      "loss": 0.9337,
+      "step": 22528
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1074371261109195e-06,
+      "loss": 0.9697,
+      "step": 22529
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.10723657506961e-06,
+      "loss": 0.9843,
+      "step": 22530
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1070362103767494e-06,
+      "loss": 0.9481,
+      "step": 22531
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1068360320361126e-06,
+      "loss": 0.9427,
+      "step": 22532
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1066360400514697e-06,
+      "loss": 0.9647,
+      "step": 22533
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1064362344265876e-06,
+      "loss": 0.9433,
+      "step": 22534
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1062366151652313e-06,
+      "loss": 0.9934,
+      "step": 22535
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.10603718227116e-06,
+      "loss": 0.9285,
+      "step": 22536
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1058379357481307e-06,
+      "loss": 0.9288,
+      "step": 22537
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.105638875599896e-06,
+      "loss": 1.0029,
+      "step": 22538
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.105440001830207e-06,
+      "loss": 0.9498,
+      "step": 22539
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.105241314442809e-06,
+      "loss": 0.9734,
+      "step": 22540
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.105042813441444e-06,
+      "loss": 1.022,
+      "step": 22541
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1048444988298513e-06,
+      "loss": 1.0943,
+      "step": 22542
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1046463706117683e-06,
+      "loss": 1.0291,
+      "step": 22543
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.104448428790925e-06,
+      "loss": 0.9393,
+      "step": 22544
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.10425067337105e-06,
+      "loss": 1.0347,
+      "step": 22545
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.10405310435587e-06,
+      "loss": 0.9046,
+      "step": 22546
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1038557217491063e-06,
+      "loss": 1.0075,
+      "step": 22547
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1036585255544763e-06,
+      "loss": 1.068,
+      "step": 22548
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1034615157756934e-06,
+      "loss": 0.9422,
+      "step": 22549
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1032646924164714e-06,
+      "loss": 1.0018,
+      "step": 22550
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.103068055480515e-06,
+      "loss": 1.001,
+      "step": 22551
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1028716049715304e-06,
+      "loss": 0.9191,
+      "step": 22552
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1026753408932167e-06,
+      "loss": 0.9092,
+      "step": 22553
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.102479263249272e-06,
+      "loss": 1.0363,
+      "step": 22554
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.102283372043389e-06,
+      "loss": 0.9444,
+      "step": 22555
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.102087667279258e-06,
+      "loss": 1.0197,
+      "step": 22556
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.101892148960565e-06,
+      "loss": 0.9905,
+      "step": 22557
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1016968170909937e-06,
+      "loss": 1.0249,
+      "step": 22558
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1015016716742233e-06,
+      "loss": 0.9451,
+      "step": 22559
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.101306712713929e-06,
+      "loss": 0.9695,
+      "step": 22560
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1011119402137837e-06,
+      "loss": 0.9691,
+      "step": 22561
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1009173541774565e-06,
+      "loss": 0.9922,
+      "step": 22562
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.100722954608614e-06,
+      "loss": 0.9998,
+      "step": 22563
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1005287415109146e-06,
+      "loss": 0.9261,
+      "step": 22564
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1003347148880204e-06,
+      "loss": 0.9645,
+      "step": 22565
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.100140874743584e-06,
+      "loss": 1.0035,
+      "step": 22566
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.099947221081258e-06,
+      "loss": 0.9674,
+      "step": 22567
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.099753753904689e-06,
+      "loss": 0.9663,
+      "step": 22568
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0995604732175213e-06,
+      "loss": 1.0142,
+      "step": 22569
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0993673790233986e-06,
+      "loss": 0.9279,
+      "step": 22570
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0991744713259544e-06,
+      "loss": 1.0099,
+      "step": 22571
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0989817501288225e-06,
+      "loss": 1.0459,
+      "step": 22572
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.098789215435637e-06,
+      "loss": 1.0155,
+      "step": 22573
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0985968672500215e-06,
+      "loss": 0.9543,
+      "step": 22574
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0984047055755997e-06,
+      "loss": 0.9831,
+      "step": 22575
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0982127304159917e-06,
+      "loss": 1.003,
+      "step": 22576
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0980209417748146e-06,
+      "loss": 1.0111,
+      "step": 22577
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0978293396556796e-06,
+      "loss": 0.9727,
+      "step": 22578
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.097637924062196e-06,
+      "loss": 0.9944,
+      "step": 22579
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.09744669499797e-06,
+      "loss": 0.964,
+      "step": 22580
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0972556524666032e-06,
+      "loss": 1.0152,
+      "step": 22581
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.097064796471696e-06,
+      "loss": 0.976,
+      "step": 22582
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0968741270168416e-06,
+      "loss": 0.9873,
+      "step": 22583
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0966836441056315e-06,
+      "loss": 0.9854,
+      "step": 22584
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.096493347741655e-06,
+      "loss": 0.9913,
+      "step": 22585
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0963032379284967e-06,
+      "loss": 1.0281,
+      "step": 22586
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0961133146697366e-06,
+      "loss": 0.9682,
+      "step": 22587
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0959235779689527e-06,
+      "loss": 0.9997,
+      "step": 22588
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0957340278297204e-06,
+      "loss": 1.0054,
+      "step": 22589
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0955446642556078e-06,
+      "loss": 1.0203,
+      "step": 22590
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0953554872501825e-06,
+      "loss": 1.045,
+      "step": 22591
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0951664968170093e-06,
+      "loss": 0.9829,
+      "step": 22592
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.094977692959648e-06,
+      "loss": 0.9914,
+      "step": 22593
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0947890756816537e-06,
+      "loss": 0.9809,
+      "step": 22594
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.094600644986581e-06,
+      "loss": 1.0279,
+      "step": 22595
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.094412400877978e-06,
+      "loss": 0.9919,
+      "step": 22596
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0942243433593915e-06,
+      "loss": 0.9873,
+      "step": 22597
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0940364724343635e-06,
+      "loss": 0.946,
+      "step": 22598
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.093848788106432e-06,
+      "loss": 1.0175,
+      "step": 22599
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0936612903791357e-06,
+      "loss": 0.9775,
+      "step": 22600
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0934739792560023e-06,
+      "loss": 1.0347,
+      "step": 22601
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0932868547405616e-06,
+      "loss": 1.1095,
+      "step": 22602
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.09309991683634e-06,
+      "loss": 0.9124,
+      "step": 22603
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0929131655468573e-06,
+      "loss": 1.0412,
+      "step": 22604
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.092726600875631e-06,
+      "loss": 0.9447,
+      "step": 22605
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.092540222826177e-06,
+      "loss": 0.9619,
+      "step": 22606
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0923540314020042e-06,
+      "loss": 0.9826,
+      "step": 22607
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0921680266066216e-06,
+      "loss": 0.9527,
+      "step": 22608
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0919822084435317e-06,
+      "loss": 0.9431,
+      "step": 22609
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0917965769162354e-06,
+      "loss": 1.0268,
+      "step": 22610
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.091611132028229e-06,
+      "loss": 0.9915,
+      "step": 22611
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0914258737830055e-06,
+      "loss": 0.9821,
+      "step": 22612
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0912408021840547e-06,
+      "loss": 0.9829,
+      "step": 22613
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0910559172348634e-06,
+      "loss": 1.0034,
+      "step": 22614
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0908712189389137e-06,
+      "loss": 0.9316,
+      "step": 22615
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0906867072996852e-06,
+      "loss": 0.9856,
+      "step": 22616
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.090502382320653e-06,
+      "loss": 0.9351,
+      "step": 22617
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0903182440052883e-06,
+      "loss": 0.9772,
+      "step": 22618
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0901342923570625e-06,
+      "loss": 0.99,
+      "step": 22619
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0899505273794374e-06,
+      "loss": 0.9726,
+      "step": 22620
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.089766949075878e-06,
+      "loss": 0.9441,
+      "step": 22621
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.089583557449839e-06,
+      "loss": 1.0386,
+      "step": 22622
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.089400352504777e-06,
+      "loss": 1.0514,
+      "step": 22623
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0892173342441414e-06,
+      "loss": 0.954,
+      "step": 22624
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.089034502671381e-06,
+      "loss": 0.9064,
+      "step": 22625
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0888518577899397e-06,
+      "loss": 0.913,
+      "step": 22626
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.088669399603258e-06,
+      "loss": 0.9755,
+      "step": 22627
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.088487128114771e-06,
+      "loss": 1.0266,
+      "step": 22628
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0883050433279154e-06,
+      "loss": 1.0256,
+      "step": 22629
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0881231452461185e-06,
+      "loss": 0.9741,
+      "step": 22630
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.087941433872808e-06,
+      "loss": 0.9972,
+      "step": 22631
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0877599092114058e-06,
+      "loss": 0.9471,
+      "step": 22632
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.087578571265332e-06,
+      "loss": 0.9367,
+      "step": 22633
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0873974200380022e-06,
+      "loss": 1.0364,
+      "step": 22634
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.087216455532828e-06,
+      "loss": 0.9396,
+      "step": 22635
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.08703567775322e-06,
+      "loss": 1.0086,
+      "step": 22636
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0868550867025806e-06,
+      "loss": 1.0362,
+      "step": 22637
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0866746823843142e-06,
+      "loss": 0.9662,
+      "step": 22638
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0864944648018184e-06,
+      "loss": 0.9297,
+      "step": 22639
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0863144339584885e-06,
+      "loss": 0.9761,
+      "step": 22640
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0861345898577134e-06,
+      "loss": 0.9689,
+      "step": 22641
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.085954932502883e-06,
+      "loss": 1.0414,
+      "step": 22642
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0857754618973804e-06,
+      "loss": 0.9164,
+      "step": 22643
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0855961780445877e-06,
+      "loss": 0.9259,
+      "step": 22644
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0854170809478804e-06,
+      "loss": 1.0023,
+      "step": 22645
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.085238170610632e-06,
+      "loss": 0.9529,
+      "step": 22646
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0850594470362154e-06,
+      "loss": 0.9736,
+      "step": 22647
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.084880910227992e-06,
+      "loss": 0.9729,
+      "step": 22648
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.08470256018933e-06,
+      "loss": 0.9089,
+      "step": 22649
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.084524396923586e-06,
+      "loss": 1.0144,
+      "step": 22650
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.084346420434119e-06,
+      "loss": 0.9508,
+      "step": 22651
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0841686307242773e-06,
+      "loss": 0.9686,
+      "step": 22652
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.083991027797413e-06,
+      "loss": 1.003,
+      "step": 22653
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.083813611656871e-06,
+      "loss": 0.9896,
+      "step": 22654
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0836363823059914e-06,
+      "loss": 0.9879,
+      "step": 22655
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.083459339748115e-06,
+      "loss": 1.0492,
+      "step": 22656
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0832824839865762e-06,
+      "loss": 0.936,
+      "step": 22657
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0831058150247067e-06,
+      "loss": 0.9758,
+      "step": 22658
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.082929332865832e-06,
+      "loss": 1.0157,
+      "step": 22659
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.082753037513279e-06,
+      "loss": 0.9948,
+      "step": 22660
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.082576928970368e-06,
+      "loss": 0.9498,
+      "step": 22661
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.082401007240417e-06,
+      "loss": 1.0486,
+      "step": 22662
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.082225272326738e-06,
+      "loss": 0.9179,
+      "step": 22663
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.082049724232641e-06,
+      "loss": 1.0004,
+      "step": 22664
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.081874362961436e-06,
+      "loss": 0.9081,
+      "step": 22665
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.081699188516423e-06,
+      "loss": 0.9351,
+      "step": 22666
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0815242009009043e-06,
+      "loss": 0.9243,
+      "step": 22667
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0813494001181725e-06,
+      "loss": 0.9499,
+      "step": 22668
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.081174786171525e-06,
+      "loss": 1.0317,
+      "step": 22669
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0810003590642472e-06,
+      "loss": 0.9545,
+      "step": 22670
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.080826118799627e-06,
+      "loss": 0.9777,
+      "step": 22671
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0806520653809437e-06,
+      "loss": 0.9743,
+      "step": 22672
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0804781988114797e-06,
+      "loss": 1.0376,
+      "step": 22673
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0803045190945083e-06,
+      "loss": 1.0303,
+      "step": 22674
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.080131026233301e-06,
+      "loss": 0.9655,
+      "step": 22675
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.079957720231125e-06,
+      "loss": 0.9923,
+      "step": 22676
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0797846010912474e-06,
+      "loss": 1.0239,
+      "step": 22677
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0796116688169255e-06,
+      "loss": 0.9722,
+      "step": 22678
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.079438923411419e-06,
+      "loss": 1.0259,
+      "step": 22679
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0792663648779832e-06,
+      "loss": 0.984,
+      "step": 22680
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0790939932198663e-06,
+      "loss": 0.9884,
+      "step": 22681
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0789218084403166e-06,
+      "loss": 0.9546,
+      "step": 22682
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0787498105425764e-06,
+      "loss": 1.004,
+      "step": 22683
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0785779995298863e-06,
+      "loss": 0.9429,
+      "step": 22684
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.078406375405483e-06,
+      "loss": 0.9557,
+      "step": 22685
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0782349381725975e-06,
+      "loss": 0.9456,
+      "step": 22686
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.078063687834462e-06,
+      "loss": 0.8796,
+      "step": 22687
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0778926243943e-06,
+      "loss": 0.9329,
+      "step": 22688
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.077721747855335e-06,
+      "loss": 1.0051,
+      "step": 22689
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.077551058220785e-06,
+      "loss": 1.0303,
+      "step": 22690
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0773805554938665e-06,
+      "loss": 0.9593,
+      "step": 22691
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0772102396777896e-06,
+      "loss": 0.9809,
+      "step": 22692
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0770401107757634e-06,
+      "loss": 0.9895,
+      "step": 22693
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.076870168790992e-06,
+      "loss": 0.9948,
+      "step": 22694
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.076700413726678e-06,
+      "loss": 0.9737,
+      "step": 22695
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.076530845586018e-06,
+      "loss": 0.9684,
+      "step": 22696
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0763614643722067e-06,
+      "loss": 0.9849,
+      "step": 22697
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.076192270088433e-06,
+      "loss": 0.9741,
+      "step": 22698
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0760232627378864e-06,
+      "loss": 1.0061,
+      "step": 22699
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0758544423237483e-06,
+      "loss": 0.9909,
+      "step": 22700
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0756858088492006e-06,
+      "loss": 0.9321,
+      "step": 22701
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0755173623174188e-06,
+      "loss": 0.9853,
+      "step": 22702
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0753491027315755e-06,
+      "loss": 0.9252,
+      "step": 22703
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0751810300948406e-06,
+      "loss": 0.9738,
+      "step": 22704
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.075013144410382e-06,
+      "loss": 0.9361,
+      "step": 22705
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0748454456813584e-06,
+      "loss": 0.9944,
+      "step": 22706
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0746779339109315e-06,
+      "loss": 1.0085,
+      "step": 22707
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.074510609102256e-06,
+      "loss": 0.9377,
+      "step": 22708
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.074343471258483e-06,
+      "loss": 0.9024,
+      "step": 22709
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0741765203827614e-06,
+      "loss": 0.9307,
+      "step": 22710
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.074009756478237e-06,
+      "loss": 0.9836,
+      "step": 22711
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0738431795480498e-06,
+      "loss": 0.9717,
+      "step": 22712
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.073676789595338e-06,
+      "loss": 0.9569,
+      "step": 22713
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0735105866232356e-06,
+      "loss": 1.0139,
+      "step": 22714
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0733445706348743e-06,
+      "loss": 0.9554,
+      "step": 22715
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0731787416333805e-06,
+      "loss": 0.9416,
+      "step": 22716
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.073013099621877e-06,
+      "loss": 0.8795,
+      "step": 22717
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0728476446034863e-06,
+      "loss": 0.9972,
+      "step": 22718
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.072682376581322e-06,
+      "loss": 0.989,
+      "step": 22719
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0725172955585016e-06,
+      "loss": 1.0018,
+      "step": 22720
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0723524015381292e-06,
+      "loss": 0.9884,
+      "step": 22721
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0721876945233155e-06,
+      "loss": 0.9265,
+      "step": 22722
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0720231745171604e-06,
+      "loss": 0.944,
+      "step": 22723
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0718588415227647e-06,
+      "loss": 0.9612,
+      "step": 22724
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.071694695543223e-06,
+      "loss": 0.9808,
+      "step": 22725
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0715307365816263e-06,
+      "loss": 1.0634,
+      "step": 22726
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0713669646410652e-06,
+      "loss": 1.0061,
+      "step": 22727
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0712033797246234e-06,
+      "loss": 0.9998,
+      "step": 22728
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0710399818353823e-06,
+      "loss": 1.0458,
+      "step": 22729
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.07087677097642e-06,
+      "loss": 1.0305,
+      "step": 22730
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.070713747150811e-06,
+      "loss": 1.0011,
+      "step": 22731
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0705509103616265e-06,
+      "loss": 0.9277,
+      "step": 22732
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.070388260611932e-06,
+      "loss": 1.0384,
+      "step": 22733
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.070225797904794e-06,
+      "loss": 0.9602,
+      "step": 22734
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.070063522243271e-06,
+      "loss": 1.0415,
+      "step": 22735
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0699014336304202e-06,
+      "loss": 0.9715,
+      "step": 22736
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.069739532069295e-06,
+      "loss": 0.9191,
+      "step": 22737
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0695778175629444e-06,
+      "loss": 0.9797,
+      "step": 22738
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0694162901144156e-06,
+      "loss": 0.9321,
+      "step": 22739
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0692549497267502e-06,
+      "loss": 0.9779,
+      "step": 22740
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0690937964029892e-06,
+      "loss": 1.0095,
+      "step": 22741
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0689328301461666e-06,
+      "loss": 0.9276,
+      "step": 22742
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.068772050959315e-06,
+      "loss": 0.9346,
+      "step": 22743
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.068611458845461e-06,
+      "loss": 0.9657,
+      "step": 22744
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.068451053807633e-06,
+      "loss": 0.9502,
+      "step": 22745
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.068290835848852e-06,
+      "loss": 0.9388,
+      "step": 22746
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.068130804972134e-06,
+      "loss": 0.9989,
+      "step": 22747
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0679709611804946e-06,
+      "loss": 0.8818,
+      "step": 22748
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0678113044769446e-06,
+      "loss": 0.9345,
+      "step": 22749
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0676518348644917e-06,
+      "loss": 1.0136,
+      "step": 22750
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.067492552346139e-06,
+      "loss": 0.9289,
+      "step": 22751
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0673334569248875e-06,
+      "loss": 0.9947,
+      "step": 22752
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0671745486037345e-06,
+      "loss": 1.0321,
+      "step": 22753
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0670158273856734e-06,
+      "loss": 0.915,
+      "step": 22754
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0668572932736933e-06,
+      "loss": 0.9204,
+      "step": 22755
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.066698946270779e-06,
+      "loss": 0.9329,
+      "step": 22756
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.066540786379917e-06,
+      "loss": 0.9196,
+      "step": 22757
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.066382813604083e-06,
+      "loss": 0.9917,
+      "step": 22758
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.066225027946255e-06,
+      "loss": 0.9416,
+      "step": 22759
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0660674294094044e-06,
+      "loss": 1.0002,
+      "step": 22760
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.065910017996499e-06,
+      "loss": 0.9987,
+      "step": 22761
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.065752793710505e-06,
+      "loss": 0.9974,
+      "step": 22762
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.065595756554384e-06,
+      "loss": 0.9998,
+      "step": 22763
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0654389065310935e-06,
+      "loss": 1.0069,
+      "step": 22764
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0652822436435896e-06,
+      "loss": 0.922,
+      "step": 22765
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0651257678948203e-06,
+      "loss": 1.0299,
+      "step": 22766
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.064969479287735e-06,
+      "loss": 0.9395,
+      "step": 22767
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.064813377825279e-06,
+      "loss": 1.055,
+      "step": 22768
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0646574635103914e-06,
+      "loss": 0.9807,
+      "step": 22769
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0645017363460086e-06,
+      "loss": 0.9831,
+      "step": 22770
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0643461963350644e-06,
+      "loss": 0.9711,
+      "step": 22771
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0641908434804883e-06,
+      "loss": 0.9025,
+      "step": 22772
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0640356777852078e-06,
+      "loss": 0.91,
+      "step": 22773
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0638806992521446e-06,
+      "loss": 0.996,
+      "step": 22774
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.063725907884219e-06,
+      "loss": 0.982,
+      "step": 22775
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0635713036843452e-06,
+      "loss": 1.0336,
+      "step": 22776
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0634168866554384e-06,
+      "loss": 0.9693,
+      "step": 22777
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0632626568004045e-06,
+      "loss": 1.0456,
+      "step": 22778
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0631086141221492e-06,
+      "loss": 0.9713,
+      "step": 22779
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0629547586235755e-06,
+      "loss": 0.9806,
+      "step": 22780
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0628010903075804e-06,
+      "loss": 0.9889,
+      "step": 22781
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0626476091770576e-06,
+      "loss": 0.9507,
+      "step": 22782
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.062494315234901e-06,
+      "loss": 0.9908,
+      "step": 22783
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0623412084839956e-06,
+      "loss": 0.9345,
+      "step": 22784
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0621882889272273e-06,
+      "loss": 0.9948,
+      "step": 22785
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.062035556567476e-06,
+      "loss": 0.9571,
+      "step": 22786
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0618830114076172e-06,
+      "loss": 1.0121,
+      "step": 22787
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.061730653450527e-06,
+      "loss": 0.9591,
+      "step": 22788
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0615784826990728e-06,
+      "loss": 0.9527,
+      "step": 22789
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.061426499156123e-06,
+      "loss": 0.9644,
+      "step": 22790
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0612747028245387e-06,
+      "loss": 0.9813,
+      "step": 22791
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0611230937071813e-06,
+      "loss": 1.0018,
+      "step": 22792
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0609716718069053e-06,
+      "loss": 1.036,
+      "step": 22793
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.060820437126563e-06,
+      "loss": 1.0451,
+      "step": 22794
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0606693896690034e-06,
+      "loss": 0.9301,
+      "step": 22795
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0605185294370726e-06,
+      "loss": 1.003,
+      "step": 22796
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0603678564336106e-06,
+      "loss": 0.9598,
+      "step": 22797
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0602173706614572e-06,
+      "loss": 1.0025,
+      "step": 22798
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0600670721234453e-06,
+      "loss": 0.9983,
+      "step": 22799
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.059916960822408e-06,
+      "loss": 1.0658,
+      "step": 22800
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.059767036761172e-06,
+      "loss": 1.0074,
+      "step": 22801
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0596172999425625e-06,
+      "loss": 1.0492,
+      "step": 22802
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0594677503693983e-06,
+      "loss": 0.9793,
+      "step": 22803
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.059318388044498e-06,
+      "loss": 1.0013,
+      "step": 22804
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0591692129706733e-06,
+      "loss": 0.9823,
+      "step": 22805
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0590202251507355e-06,
+      "loss": 1.0523,
+      "step": 22806
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.058871424587491e-06,
+      "loss": 0.9721,
+      "step": 22807
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.058722811283742e-06,
+      "loss": 0.9601,
+      "step": 22808
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.058574385242288e-06,
+      "loss": 0.9952,
+      "step": 22809
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0584261464659264e-06,
+      "loss": 0.9701,
+      "step": 22810
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0582780949574477e-06,
+      "loss": 0.9547,
+      "step": 22811
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.058130230719643e-06,
+      "loss": 0.9873,
+      "step": 22812
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0579825537552948e-06,
+      "loss": 0.9403,
+      "step": 22813
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0578350640671864e-06,
+      "loss": 1.0008,
+      "step": 22814
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.057687761658096e-06,
+      "loss": 0.9493,
+      "step": 22815
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.057540646530798e-06,
+      "loss": 0.9069,
+      "step": 22816
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0573937186880625e-06,
+      "loss": 0.9714,
+      "step": 22817
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.057246978132659e-06,
+      "loss": 0.9447,
+      "step": 22818
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0571004248673524e-06,
+      "loss": 0.9278,
+      "step": 22819
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0569540588949e-06,
+      "loss": 0.9895,
+      "step": 22820
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0568078802180616e-06,
+      "loss": 0.9636,
+      "step": 22821
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.056661888839591e-06,
+      "loss": 1.0021,
+      "step": 22822
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.056516084762235e-06,
+      "loss": 0.9821,
+      "step": 22823
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0563704679887433e-06,
+      "loss": 1.064,
+      "step": 22824
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0562250385218565e-06,
+      "loss": 0.9887,
+      "step": 22825
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.056079796364318e-06,
+      "loss": 0.983,
+      "step": 22826
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0559347415188583e-06,
+      "loss": 0.9695,
+      "step": 22827
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0557898739882133e-06,
+      "loss": 0.9953,
+      "step": 22828
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.055645193775111e-06,
+      "loss": 1.0024,
+      "step": 22829
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.055500700882277e-06,
+      "loss": 1.0534,
+      "step": 22830
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0553563953124333e-06,
+      "loss": 0.909,
+      "step": 22831
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0552122770682973e-06,
+      "loss": 0.967,
+      "step": 22832
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0550683461525834e-06,
+      "loss": 0.871,
+      "step": 22833
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0549246025680047e-06,
+      "loss": 0.9529,
+      "step": 22834
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.054781046317267e-06,
+      "loss": 0.9532,
+      "step": 22835
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.054637677403074e-06,
+      "loss": 0.9786,
+      "step": 22836
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.05449449582813e-06,
+      "loss": 1.0069,
+      "step": 22837
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.054351501595128e-06,
+      "loss": 0.9531,
+      "step": 22838
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.054208694706763e-06,
+      "loss": 1.037,
+      "step": 22839
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.054066075165725e-06,
+      "loss": 0.9275,
+      "step": 22840
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0539236429747015e-06,
+      "loss": 0.9705,
+      "step": 22841
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.053781398136374e-06,
+      "loss": 1.0567,
+      "step": 22842
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.053639340653423e-06,
+      "loss": 1.062,
+      "step": 22843
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.053497470528524e-06,
+      "loss": 1.0157,
+      "step": 22844
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0533557877643497e-06,
+      "loss": 1.0664,
+      "step": 22845
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.053214292363568e-06,
+      "loss": 0.9634,
+      "step": 22846
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0530729843288446e-06,
+      "loss": 0.9794,
+      "step": 22847
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0529318636628418e-06,
+      "loss": 0.9206,
+      "step": 22848
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.052790930368217e-06,
+      "loss": 1.0376,
+      "step": 22849
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.052650184447627e-06,
+      "loss": 0.9169,
+      "step": 22850
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0525096259037205e-06,
+      "loss": 0.9732,
+      "step": 22851
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.052369254739146e-06,
+      "loss": 0.9266,
+      "step": 22852
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0522290709565486e-06,
+      "loss": 1.0518,
+      "step": 22853
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0520890745585676e-06,
+      "loss": 0.9981,
+      "step": 22854
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0519492655478415e-06,
+      "loss": 1.005,
+      "step": 22855
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.051809643927002e-06,
+      "loss": 0.9577,
+      "step": 22856
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.051670209698681e-06,
+      "loss": 1.0408,
+      "step": 22857
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0515309628655047e-06,
+      "loss": 1.0025,
+      "step": 22858
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0513919034300944e-06,
+      "loss": 0.9095,
+      "step": 22859
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0512530313950707e-06,
+      "loss": 1.0501,
+      "step": 22860
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0511143467630504e-06,
+      "loss": 0.995,
+      "step": 22861
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0509758495366452e-06,
+      "loss": 0.9746,
+      "step": 22862
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0508375397184625e-06,
+      "loss": 0.9745,
+      "step": 22863
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0506994173111097e-06,
+      "loss": 0.9644,
+      "step": 22864
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0505614823171873e-06,
+      "loss": 0.9597,
+      "step": 22865
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0504237347392944e-06,
+      "loss": 1.0119,
+      "step": 22866
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0502861745800254e-06,
+      "loss": 0.9169,
+      "step": 22867
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0501488018419715e-06,
+      "loss": 0.9343,
+      "step": 22868
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.05001161652772e-06,
+      "loss": 0.9758,
+      "step": 22869
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0498746186398547e-06,
+      "loss": 0.9802,
+      "step": 22870
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.049737808180957e-06,
+      "loss": 0.8382,
+      "step": 22871
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.049601185153605e-06,
+      "loss": 1.038,
+      "step": 22872
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0494647495603707e-06,
+      "loss": 0.8943,
+      "step": 22873
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0493285014038246e-06,
+      "loss": 1.0048,
+      "step": 22874
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.049192440686533e-06,
+      "loss": 1.0401,
+      "step": 22875
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0490565674110576e-06,
+      "loss": 1.0007,
+      "step": 22876
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.048920881579961e-06,
+      "loss": 0.9829,
+      "step": 22877
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.048785383195796e-06,
+      "loss": 0.8724,
+      "step": 22878
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.048650072261117e-06,
+      "loss": 1.0203,
+      "step": 22879
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.048514948778472e-06,
+      "loss": 0.966,
+      "step": 22880
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0483800127504065e-06,
+      "loss": 0.9535,
+      "step": 22881
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0482452641794605e-06,
+      "loss": 0.9844,
+      "step": 22882
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.048110703068176e-06,
+      "loss": 0.9896,
+      "step": 22883
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.047976329419085e-06,
+      "loss": 0.985,
+      "step": 22884
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0478421432347185e-06,
+      "loss": 0.9321,
+      "step": 22885
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.047708144517605e-06,
+      "loss": 0.9636,
+      "step": 22886
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0475743332702692e-06,
+      "loss": 0.9408,
+      "step": 22887
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.047440709495231e-06,
+      "loss": 0.9995,
+      "step": 22888
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0473072731950063e-06,
+      "loss": 0.9072,
+      "step": 22889
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0471740243721108e-06,
+      "loss": 0.9039,
+      "step": 22890
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0470409630290525e-06,
+      "loss": 1.0229,
+      "step": 22891
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0469080891683404e-06,
+      "loss": 0.9779,
+      "step": 22892
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.046775402792474e-06,
+      "loss": 0.9132,
+      "step": 22893
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.046642903903955e-06,
+      "loss": 0.9515,
+      "step": 22894
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.046510592505279e-06,
+      "loss": 1.0245,
+      "step": 22895
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0463784685989385e-06,
+      "loss": 0.9115,
+      "step": 22896
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0462465321874215e-06,
+      "loss": 1.0265,
+      "step": 22897
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.046114783273213e-06,
+      "loss": 0.9335,
+      "step": 22898
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0459832218587974e-06,
+      "loss": 0.9895,
+      "step": 22899
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0458518479466496e-06,
+      "loss": 0.9949,
+      "step": 22900
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0457206615392445e-06,
+      "loss": 1.0069,
+      "step": 22901
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0455896626390562e-06,
+      "loss": 0.9589,
+      "step": 22902
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0454588512485496e-06,
+      "loss": 0.9624,
+      "step": 22903
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.045328227370189e-06,
+      "loss": 1.0622,
+      "step": 22904
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.045197791006436e-06,
+      "loss": 0.8789,
+      "step": 22905
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0450675421597478e-06,
+      "loss": 1.0137,
+      "step": 22906
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0449374808325763e-06,
+      "loss": 1.0594,
+      "step": 22907
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0448076070273728e-06,
+      "loss": 1.0368,
+      "step": 22908
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.044677920746583e-06,
+      "loss": 0.9831,
+      "step": 22909
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0445484219926505e-06,
+      "loss": 1.0222,
+      "step": 22910
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0444191107680136e-06,
+      "loss": 0.9266,
+      "step": 22911
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.044289987075109e-06,
+      "loss": 0.9888,
+      "step": 22912
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.044161050916369e-06,
+      "loss": 0.976,
+      "step": 22913
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0440323022942223e-06,
+      "loss": 0.9736,
+      "step": 22914
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0439037412110933e-06,
+      "loss": 0.9965,
+      "step": 22915
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.043775367669404e-06,
+      "loss": 0.9549,
+      "step": 22916
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.043647181671573e-06,
+      "loss": 0.8719,
+      "step": 22917
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0435191832200154e-06,
+      "loss": 0.9208,
+      "step": 22918
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.043391372317141e-06,
+      "loss": 0.9913,
+      "step": 22919
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0432637489653573e-06,
+      "loss": 0.9948,
+      "step": 22920
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0431363131670703e-06,
+      "loss": 0.971,
+      "step": 22921
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0430090649246784e-06,
+      "loss": 1.0156,
+      "step": 22922
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0428820042405797e-06,
+      "loss": 0.9835,
+      "step": 22923
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.042755131117166e-06,
+      "loss": 0.9465,
+      "step": 22924
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0426284455568305e-06,
+      "loss": 0.985,
+      "step": 22925
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.042501947561956e-06,
+      "loss": 0.8984,
+      "step": 22926
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.042375637134927e-06,
+      "loss": 0.9807,
+      "step": 22927
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0422495142781232e-06,
+      "loss": 0.9409,
+      "step": 22928
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.04212357899392e-06,
+      "loss": 0.9693,
+      "step": 22929
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.041997831284688e-06,
+      "loss": 1.0125,
+      "step": 22930
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0418722711527986e-06,
+      "loss": 0.9352,
+      "step": 22931
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0417468986006153e-06,
+      "loss": 1.0373,
+      "step": 22932
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0416217136304995e-06,
+      "loss": 0.9406,
+      "step": 22933
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.041496716244811e-06,
+      "loss": 0.9814,
+      "step": 22934
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0413719064459014e-06,
+      "loss": 0.9796,
+      "step": 22935
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0412472842361253e-06,
+      "loss": 0.955,
+      "step": 22936
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.041122849617827e-06,
+      "loss": 0.8987,
+      "step": 22937
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0409986025933528e-06,
+      "loss": 0.9892,
+      "step": 22938
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0408745431650426e-06,
+      "loss": 0.9237,
+      "step": 22939
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0407506713352322e-06,
+      "loss": 1.0264,
+      "step": 22940
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.040626987106256e-06,
+      "loss": 0.9797,
+      "step": 22941
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0405034904804425e-06,
+      "loss": 0.9667,
+      "step": 22942
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.04038018146012e-06,
+      "loss": 0.9473,
+      "step": 22943
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.040257060047609e-06,
+      "loss": 0.9621,
+      "step": 22944
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0401341262452314e-06,
+      "loss": 0.9666,
+      "step": 22945
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0400113800553004e-06,
+      "loss": 1.0359,
+      "step": 22946
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0398888214801293e-06,
+      "loss": 0.8994,
+      "step": 22947
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0397664505220272e-06,
+      "loss": 0.9572,
+      "step": 22948
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0396442671832984e-06,
+      "loss": 0.98,
+      "step": 22949
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0395222714662448e-06,
+      "loss": 0.8853,
+      "step": 22950
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0394004633731637e-06,
+      "loss": 0.9434,
+      "step": 22951
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0392788429063496e-06,
+      "loss": 0.9993,
+      "step": 22952
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0391574100680944e-06,
+      "loss": 0.9823,
+      "step": 22953
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.039036164860685e-06,
+      "loss": 1.0131,
+      "step": 22954
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.038915107286406e-06,
+      "loss": 0.9205,
+      "step": 22955
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.038794237347536e-06,
+      "loss": 0.9981,
+      "step": 22956
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0386735550463534e-06,
+      "loss": 0.9066,
+      "step": 22957
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0385530603851313e-06,
+      "loss": 0.9716,
+      "step": 22958
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.038432753366139e-06,
+      "loss": 1.0091,
+      "step": 22959
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0383126339916427e-06,
+      "loss": 1.0354,
+      "step": 22960
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0381927022639048e-06,
+      "loss": 1.0546,
+      "step": 22961
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0380729581851847e-06,
+      "loss": 0.9144,
+      "step": 22962
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.037953401757739e-06,
+      "loss": 0.9723,
+      "step": 22963
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0378340329838177e-06,
+      "loss": 0.9798,
+      "step": 22964
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.037714851865672e-06,
+      "loss": 1.0457,
+      "step": 22965
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0375958584055435e-06,
+      "loss": 1.0256,
+      "step": 22966
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0374770526056766e-06,
+      "loss": 0.9894,
+      "step": 22967
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0373584344683084e-06,
+      "loss": 0.9926,
+      "step": 22968
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0372400039956734e-06,
+      "loss": 0.9381,
+      "step": 22969
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.037121761190001e-06,
+      "loss": 0.9666,
+      "step": 22970
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.037003706053521e-06,
+      "loss": 0.8981,
+      "step": 22971
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.036885838588455e-06,
+      "loss": 1.1074,
+      "step": 22972
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0367681587970237e-06,
+      "loss": 0.9295,
+      "step": 22973
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0366506666814447e-06,
+      "loss": 0.9293,
+      "step": 22974
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.036533362243931e-06,
+      "loss": 0.9156,
+      "step": 22975
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.036416245486692e-06,
+      "loss": 0.9846,
+      "step": 22976
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.036299316411933e-06,
+      "loss": 1.0089,
+      "step": 22977
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.036182575021857e-06,
+      "loss": 0.9718,
+      "step": 22978
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0360660213186646e-06,
+      "loss": 0.9961,
+      "step": 22979
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.03594965530455e-06,
+      "loss": 1.0495,
+      "step": 22980
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0358334769817035e-06,
+      "loss": 0.9916,
+      "step": 22981
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.035717486352317e-06,
+      "loss": 0.9983,
+      "step": 22982
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0356016834185735e-06,
+      "loss": 0.9377,
+      "step": 22983
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.035486068182654e-06,
+      "loss": 0.9326,
+      "step": 22984
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.035370640646737e-06,
+      "loss": 1.029,
+      "step": 22985
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.035255400812997e-06,
+      "loss": 0.9195,
+      "step": 22986
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0351403486836043e-06,
+      "loss": 1.0276,
+      "step": 22987
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0350254842607253e-06,
+      "loss": 0.9974,
+      "step": 22988
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0349108075465253e-06,
+      "loss": 0.9991,
+      "step": 22989
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.034796318543164e-06,
+      "loss": 0.9369,
+      "step": 22990
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.034682017252797e-06,
+      "loss": 0.9341,
+      "step": 22991
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.034567903677579e-06,
+      "loss": 1.0422,
+      "step": 22992
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0344539778196585e-06,
+      "loss": 0.9366,
+      "step": 22993
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.034340239681182e-06,
+      "loss": 1.0588,
+      "step": 22994
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0342266892642905e-06,
+      "loss": 0.891,
+      "step": 22995
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0341133265711247e-06,
+      "loss": 1.0572,
+      "step": 22996
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.034000151603819e-06,
+      "loss": 1.0354,
+      "step": 22997
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0338871643645064e-06,
+      "loss": 1.0124,
+      "step": 22998
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0337743648553145e-06,
+      "loss": 1.0131,
+      "step": 22999
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.033661753078367e-06,
+      "loss": 0.9658,
+      "step": 23000
+    },
+    {
+      "epoch": 0.97,
+      "eval_text_loss": 0.7514827251434326,
+      "eval_text_runtime": 1.4062,
+      "eval_text_samples_per_second": 52.623,
+      "eval_text_steps_per_second": 1.422,
+      "step": 23000
+    },
+    {
+      "epoch": 0.97,
+      "eval_svit_loss": 1.0696531534194946,
+      "eval_svit_runtime": 5.0281,
+      "eval_svit_samples_per_second": 135.438,
+      "eval_svit_steps_per_second": 2.188,
+      "step": 23000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.033549329035787e-06,
+      "loss": 0.9441,
+      "step": 23001
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.033437092729691e-06,
+      "loss": 1.0487,
+      "step": 23002
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0333250441621944e-06,
+      "loss": 0.9382,
+      "step": 23003
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0332131833354055e-06,
+      "loss": 0.993,
+      "step": 23004
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0331015102514344e-06,
+      "loss": 0.9534,
+      "step": 23005
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0329900249123825e-06,
+      "loss": 0.952,
+      "step": 23006
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0328787273203513e-06,
+      "loss": 0.9256,
+      "step": 23007
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0327676174774348e-06,
+      "loss": 1.0072,
+      "step": 23008
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0326566953857297e-06,
+      "loss": 0.9021,
+      "step": 23009
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0325459610473222e-06,
+      "loss": 0.9122,
+      "step": 23010
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0324354144643e-06,
+      "loss": 0.9341,
+      "step": 23011
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0323250556387445e-06,
+      "loss": 1.002,
+      "step": 23012
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0322148845727344e-06,
+      "loss": 0.9798,
+      "step": 23013
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.032104901268346e-06,
+      "loss": 0.9506,
+      "step": 23014
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.03199510572765e-06,
+      "loss": 1.0105,
+      "step": 23015
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.031885497952716e-06,
+      "loss": 0.9417,
+      "step": 23016
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0317760779456065e-06,
+      "loss": 0.9833,
+      "step": 23017
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0316668457083845e-06,
+      "loss": 0.938,
+      "step": 23018
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0315578012431072e-06,
+      "loss": 0.9476,
+      "step": 23019
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0314489445518282e-06,
+      "loss": 1.0025,
+      "step": 23020
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0313402756365986e-06,
+      "loss": 0.9089,
+      "step": 23021
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0312317944994645e-06,
+      "loss": 1.0543,
+      "step": 23022
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0311235011424705e-06,
+      "loss": 1.0258,
+      "step": 23023
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.031015395567655e-06,
+      "loss": 1.0152,
+      "step": 23024
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0309074777770556e-06,
+      "loss": 0.9783,
+      "step": 23025
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.030799747772704e-06,
+      "loss": 1.0571,
+      "step": 23026
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0306922055566314e-06,
+      "loss": 1.0159,
+      "step": 23027
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0305848511308614e-06,
+      "loss": 0.9956,
+      "step": 23028
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0304776844974186e-06,
+      "loss": 1.0108,
+      "step": 23029
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.030370705658319e-06,
+      "loss": 1.0341,
+      "step": 23030
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.030263914615579e-06,
+      "loss": 1.0169,
+      "step": 23031
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0301573113712117e-06,
+      "loss": 0.9796,
+      "step": 23032
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0300508959272217e-06,
+      "loss": 0.954,
+      "step": 23033
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0299446682856175e-06,
+      "loss": 1.0098,
+      "step": 23034
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0298386284483963e-06,
+      "loss": 1.0326,
+      "step": 23035
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.029732776417559e-06,
+      "loss": 0.9651,
+      "step": 23036
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.029627112195097e-06,
+      "loss": 1.046,
+      "step": 23037
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.029521635783002e-06,
+      "loss": 0.9512,
+      "step": 23038
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.02941634718326e-06,
+      "loss": 1.0191,
+      "step": 23039
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.029311246397855e-06,
+      "loss": 1.0264,
+      "step": 23040
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.029206333428766e-06,
+      "loss": 0.988,
+      "step": 23041
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0291016082779705e-06,
+      "loss": 0.9962,
+      "step": 23042
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.028997070947441e-06,
+      "loss": 1.0133,
+      "step": 23043
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.028892721439144e-06,
+      "loss": 0.9985,
+      "step": 23044
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0287885597550495e-06,
+      "loss": 1.0188,
+      "step": 23045
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0286845858971164e-06,
+      "loss": 0.9775,
+      "step": 23046
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0285807998673036e-06,
+      "loss": 0.9919,
+      "step": 23047
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.028477201667567e-06,
+      "loss": 0.9623,
+      "step": 23048
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0283737912998584e-06,
+      "loss": 1.0577,
+      "step": 23049
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.028270568766123e-06,
+      "loss": 1.0546,
+      "step": 23050
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0281675340683086e-06,
+      "loss": 0.9364,
+      "step": 23051
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0280646872083545e-06,
+      "loss": 0.9457,
+      "step": 23052
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0279620281881987e-06,
+      "loss": 1.0035,
+      "step": 23053
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0278595570097737e-06,
+      "loss": 1.0544,
+      "step": 23054
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.027757273675011e-06,
+      "loss": 0.941,
+      "step": 23055
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0276551781858356e-06,
+      "loss": 0.8459,
+      "step": 23056
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.027553270544173e-06,
+      "loss": 1.0241,
+      "step": 23057
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.027451550751941e-06,
+      "loss": 1.0161,
+      "step": 23058
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0273500188110565e-06,
+      "loss": 1.0344,
+      "step": 23059
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.027248674723432e-06,
+      "loss": 0.9905,
+      "step": 23060
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.027147518490976e-06,
+      "loss": 0.9986,
+      "step": 23061
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0270465501155945e-06,
+      "loss": 0.942,
+      "step": 23062
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.02694576959919e-06,
+      "loss": 0.9798,
+      "step": 23063
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0268451769436593e-06,
+      "loss": 1.0102,
+      "step": 23064
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0267447721508983e-06,
+      "loss": 0.9899,
+      "step": 23065
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0266445552227978e-06,
+      "loss": 0.9344,
+      "step": 23066
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0265445261612462e-06,
+      "loss": 0.9862,
+      "step": 23067
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0264446849681274e-06,
+      "loss": 0.9505,
+      "step": 23068
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0263450316453217e-06,
+      "loss": 1.0141,
+      "step": 23069
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0262455661947067e-06,
+      "loss": 1.0656,
+      "step": 23070
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0261462886181563e-06,
+      "loss": 1.0522,
+      "step": 23071
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0260471989175405e-06,
+      "loss": 1.078,
+      "step": 23072
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0259482970947242e-06,
+      "loss": 0.9382,
+      "step": 23073
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0258495831515737e-06,
+      "loss": 0.9723,
+      "step": 23074
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0257510570899455e-06,
+      "loss": 0.999,
+      "step": 23075
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.025652718911696e-06,
+      "loss": 0.992,
+      "step": 23076
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0255545686186785e-06,
+      "loss": 0.995,
+      "step": 23077
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0254566062127423e-06,
+      "loss": 0.94,
+      "step": 23078
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.025358831695731e-06,
+      "loss": 0.9725,
+      "step": 23079
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.025261245069488e-06,
+      "loss": 0.9505,
+      "step": 23080
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0251638463358494e-06,
+      "loss": 1.0274,
+      "step": 23081
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.025066635496652e-06,
+      "loss": 0.9904,
+      "step": 23082
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0249696125537266e-06,
+      "loss": 1.0326,
+      "step": 23083
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0248727775088996e-06,
+      "loss": 1.0277,
+      "step": 23084
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.024776130363997e-06,
+      "loss": 0.9903,
+      "step": 23085
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.024679671120837e-06,
+      "loss": 0.9572,
+      "step": 23086
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0245833997812385e-06,
+      "loss": 0.9662,
+      "step": 23087
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0244873163470137e-06,
+      "loss": 0.9344,
+      "step": 23088
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0243914208199735e-06,
+      "loss": 0.9746,
+      "step": 23089
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.024295713201924e-06,
+      "loss": 1.0395,
+      "step": 23090
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0242001934946666e-06,
+      "loss": 0.9475,
+      "step": 23091
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.024104861700003e-06,
+      "loss": 0.9896,
+      "step": 23092
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.024009717819727e-06,
+      "loss": 1.0133,
+      "step": 23093
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0239147618556323e-06,
+      "loss": 1.0308,
+      "step": 23094
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.023819993809506e-06,
+      "loss": 1.0075,
+      "step": 23095
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0237254136831354e-06,
+      "loss": 0.9433,
+      "step": 23096
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0236310214782998e-06,
+      "loss": 0.9576,
+      "step": 23097
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0235368171967787e-06,
+      "loss": 0.9331,
+      "step": 23098
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0234428008403454e-06,
+      "loss": 0.9872,
+      "step": 23099
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.023348972410773e-06,
+      "loss": 0.9973,
+      "step": 23100
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.023255331909826e-06,
+      "loss": 0.8671,
+      "step": 23101
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0231618793392704e-06,
+      "loss": 0.9869,
+      "step": 23102
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.023068614700866e-06,
+      "loss": 1.0228,
+      "step": 23103
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0229755379963704e-06,
+      "loss": 0.9882,
+      "step": 23104
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0228826492275344e-06,
+      "loss": 0.9137,
+      "step": 23105
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.02278994839611e-06,
+      "loss": 0.9237,
+      "step": 23106
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0226974355038434e-06,
+      "loss": 0.9456,
+      "step": 23107
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0226051105524773e-06,
+      "loss": 0.9107,
+      "step": 23108
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.022512973543749e-06,
+      "loss": 0.9995,
+      "step": 23109
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.022421024479395e-06,
+      "loss": 0.9044,
+      "step": 23110
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0223292633611482e-06,
+      "loss": 0.9709,
+      "step": 23111
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.022237690190736e-06,
+      "loss": 0.9443,
+      "step": 23112
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0221463049698844e-06,
+      "loss": 1.0378,
+      "step": 23113
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.022055107700313e-06,
+      "loss": 0.9079,
+      "step": 23114
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0219640983837422e-06,
+      "loss": 1.0008,
+      "step": 23115
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0218732770218844e-06,
+      "loss": 0.9067,
+      "step": 23116
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.021782643616451e-06,
+      "loss": 0.9976,
+      "step": 23117
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0216921981691486e-06,
+      "loss": 0.9725,
+      "step": 23118
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0216019406816825e-06,
+      "loss": 0.9234,
+      "step": 23119
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0215118711557516e-06,
+      "loss": 1.0367,
+      "step": 23120
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0214219895930534e-06,
+      "loss": 0.9787,
+      "step": 23121
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0213322959952798e-06,
+      "loss": 1.0425,
+      "step": 23122
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.021242790364121e-06,
+      "loss": 0.9378,
+      "step": 23123
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.021153472701263e-06,
+      "loss": 1.0147,
+      "step": 23124
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0210643430083877e-06,
+      "loss": 1.0057,
+      "step": 23125
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0209754012871753e-06,
+      "loss": 1.1448,
+      "step": 23126
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0208866475393e-06,
+      "loss": 0.916,
+      "step": 23127
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0207980817664343e-06,
+      "loss": 1.0009,
+      "step": 23128
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0207097039702463e-06,
+      "loss": 0.9754,
+      "step": 23129
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0206215141524012e-06,
+      "loss": 0.9312,
+      "step": 23130
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.020533512314559e-06,
+      "loss": 0.9673,
+      "step": 23131
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.020445698458378e-06,
+      "loss": 0.9104,
+      "step": 23132
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0203580725855133e-06,
+      "loss": 0.9678,
+      "step": 23133
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0202706346976146e-06,
+      "loss": 0.9948,
+      "step": 23134
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0201833847963283e-06,
+      "loss": 0.9908,
+      "step": 23135
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.020096322883298e-06,
+      "loss": 0.9917,
+      "step": 23136
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0200094489601666e-06,
+      "loss": 0.971,
+      "step": 23137
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0199227630285658e-06,
+      "loss": 0.9655,
+      "step": 23138
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0198362650901318e-06,
+      "loss": 1.0015,
+      "step": 23139
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.019749955146494e-06,
+      "loss": 0.9802,
+      "step": 23140
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.019663833199276e-06,
+      "loss": 1.0438,
+      "step": 23141
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0195778992501014e-06,
+      "loss": 0.9806,
+      "step": 23142
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0194921533005897e-06,
+      "loss": 0.9926,
+      "step": 23143
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0194065953523547e-06,
+      "loss": 0.967,
+      "step": 23144
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0193212254070083e-06,
+      "loss": 1.012,
+      "step": 23145
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0192360434661596e-06,
+      "loss": 0.9568,
+      "step": 23146
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0191510495314127e-06,
+      "loss": 0.9444,
+      "step": 23147
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0190662436043677e-06,
+      "loss": 0.8755,
+      "step": 23148
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0189816256866225e-06,
+      "loss": 0.9383,
+      "step": 23149
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0188971957797713e-06,
+      "loss": 0.9934,
+      "step": 23150
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.018812953885405e-06,
+      "loss": 1.0452,
+      "step": 23151
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0187289000051094e-06,
+      "loss": 0.9587,
+      "step": 23152
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0186450341404687e-06,
+      "loss": 0.9734,
+      "step": 23153
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.018561356293062e-06,
+      "loss": 0.984,
+      "step": 23154
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.018477866464467e-06,
+      "loss": 0.9357,
+      "step": 23155
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0183945646562534e-06,
+      "loss": 0.9676,
+      "step": 23156
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.018311450869993e-06,
+      "loss": 1.0388,
+      "step": 23157
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.018228525107251e-06,
+      "loss": 0.9966,
+      "step": 23158
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.018145787369589e-06,
+      "loss": 0.9748,
+      "step": 23159
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.018063237658564e-06,
+      "loss": 0.9197,
+      "step": 23160
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.017980875975734e-06,
+      "loss": 0.9715,
+      "step": 23161
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0178987023226483e-06,
+      "loss": 1.0147,
+      "step": 23162
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.017816716700856e-06,
+      "loss": 0.9488,
+      "step": 23163
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0177349191119e-06,
+      "loss": 0.9548,
+      "step": 23164
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.017653309557323e-06,
+      "loss": 0.9466,
+      "step": 23165
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0175718880386608e-06,
+      "loss": 1.0026,
+      "step": 23166
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.017490654557447e-06,
+      "loss": 0.9563,
+      "step": 23167
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.017409609115213e-06,
+      "loss": 0.981,
+      "step": 23168
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0173287517134854e-06,
+      "loss": 1.077,
+      "step": 23169
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.017248082353786e-06,
+      "loss": 0.9039,
+      "step": 23170
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0171676010376354e-06,
+      "loss": 0.9594,
+      "step": 23171
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.017087307766549e-06,
+      "loss": 0.9331,
+      "step": 23172
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0170072025420384e-06,
+      "loss": 0.8909,
+      "step": 23173
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0169272853656153e-06,
+      "loss": 1.0063,
+      "step": 23174
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0168475562387824e-06,
+      "loss": 0.984,
+      "step": 23175
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0167680151630433e-06,
+      "loss": 1.042,
+      "step": 23176
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0166886621398954e-06,
+      "loss": 0.9822,
+      "step": 23177
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.016609497170834e-06,
+      "loss": 1.0331,
+      "step": 23178
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0165305202573496e-06,
+      "loss": 1.0843,
+      "step": 23179
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0164517314009304e-06,
+      "loss": 0.939,
+      "step": 23180
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0163731306030612e-06,
+      "loss": 0.9478,
+      "step": 23181
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0162947178652207e-06,
+      "loss": 1.0306,
+      "step": 23182
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0162164931888874e-06,
+      "loss": 0.9459,
+      "step": 23183
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.016138456575534e-06,
+      "loss": 0.9197,
+      "step": 23184
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.016060608026631e-06,
+      "loss": 0.9785,
+      "step": 23185
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0159829475436456e-06,
+      "loss": 0.9808,
+      "step": 23186
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0159054751280385e-06,
+      "loss": 1.0165,
+      "step": 23187
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0158281907812717e-06,
+      "loss": 1.0233,
+      "step": 23188
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0157510945047987e-06,
+      "loss": 0.9893,
+      "step": 23189
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.015674186300073e-06,
+      "loss": 0.987,
+      "step": 23190
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.015597466168543e-06,
+      "loss": 1.0142,
+      "step": 23191
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0155209341116533e-06,
+      "loss": 0.9857,
+      "step": 23192
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0154445901308475e-06,
+      "loss": 0.9857,
+      "step": 23193
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0153684342275603e-06,
+      "loss": 0.9933,
+      "step": 23194
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0152924664032304e-06,
+      "loss": 0.9199,
+      "step": 23195
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.015216686659285e-06,
+      "loss": 0.9721,
+      "step": 23196
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.015141094997154e-06,
+      "loss": 0.9519,
+      "step": 23197
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.01506569141826e-06,
+      "loss": 1.0086,
+      "step": 23198
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0149904759240236e-06,
+      "loss": 0.9633,
+      "step": 23199
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.014915448515862e-06,
+      "loss": 1.0317,
+      "step": 23200
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.014840609195189e-06,
+      "loss": 0.9727,
+      "step": 23201
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.014765957963413e-06,
+      "loss": 0.9861,
+      "step": 23202
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.014691494821941e-06,
+      "loss": 0.9491,
+      "step": 23203
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0146172197721748e-06,
+      "loss": 0.9422,
+      "step": 23204
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0145431328155154e-06,
+      "loss": 0.8875,
+      "step": 23205
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.014469233953356e-06,
+      "loss": 1.0216,
+      "step": 23206
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0143955231870906e-06,
+      "loss": 0.988,
+      "step": 23207
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0143220005181064e-06,
+      "loss": 0.866,
+      "step": 23208
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.014248665947789e-06,
+      "loss": 0.8928,
+      "step": 23209
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0141755194775207e-06,
+      "loss": 1.0241,
+      "step": 23210
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0141025611086762e-06,
+      "loss": 0.9302,
+      "step": 23211
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0140297908426334e-06,
+      "loss": 0.947,
+      "step": 23212
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0139572086807615e-06,
+      "loss": 1.1019,
+      "step": 23213
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0138848146244273e-06,
+      "loss": 1.0435,
+      "step": 23214
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0138126086749943e-06,
+      "loss": 0.9933,
+      "step": 23215
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.013740590833824e-06,
+      "loss": 0.9472,
+      "step": 23216
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0136687611022724e-06,
+      "loss": 0.9448,
+      "step": 23217
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.013597119481692e-06,
+      "loss": 1.0306,
+      "step": 23218
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.013525665973433e-06,
+      "loss": 0.8887,
+      "step": 23219
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0134544005788403e-06,
+      "loss": 1.0125,
+      "step": 23220
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0133833232992574e-06,
+      "loss": 0.953,
+      "step": 23221
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0133124341360236e-06,
+      "loss": 0.9526,
+      "step": 23222
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.013241733090472e-06,
+      "loss": 0.9788,
+      "step": 23223
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0131712201639367e-06,
+      "loss": 1.0461,
+      "step": 23224
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.013100895357745e-06,
+      "loss": 0.9985,
+      "step": 23225
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0130307586732214e-06,
+      "loss": 1.0455,
+      "step": 23226
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0129608101116885e-06,
+      "loss": 0.9963,
+      "step": 23227
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0128910496744617e-06,
+      "loss": 1.0349,
+      "step": 23228
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0128214773628565e-06,
+      "loss": 0.9384,
+      "step": 23229
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.012752093178182e-06,
+      "loss": 0.9528,
+      "step": 23230
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0126828971217473e-06,
+      "loss": 0.9749,
+      "step": 23231
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0126138891948547e-06,
+      "loss": 0.9944,
+      "step": 23232
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0125450693988036e-06,
+      "loss": 1.0095,
+      "step": 23233
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.012476437734892e-06,
+      "loss": 0.9888,
+      "step": 23234
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0124079942044098e-06,
+      "loss": 0.9517,
+      "step": 23235
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.01233973880865e-06,
+      "loss": 0.9784,
+      "step": 23236
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0122716715488947e-06,
+      "loss": 0.9994,
+      "step": 23237
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0122037924264283e-06,
+      "loss": 1.0584,
+      "step": 23238
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.012136101442529e-06,
+      "loss": 1.0117,
+      "step": 23239
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0120685985984724e-06,
+      "loss": 0.9906,
+      "step": 23240
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0120012838955286e-06,
+      "loss": 0.9447,
+      "step": 23241
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0119341573349663e-06,
+      "loss": 0.9292,
+      "step": 23242
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.01186721891805e-06,
+      "loss": 0.9896,
+      "step": 23243
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0118004686460413e-06,
+      "loss": 1.0014,
+      "step": 23244
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.011733906520196e-06,
+      "loss": 0.9748,
+      "step": 23245
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.011667532541769e-06,
+      "loss": 0.9605,
+      "step": 23246
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.011601346712011e-06,
+      "loss": 0.9958,
+      "step": 23247
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0115353490321677e-06,
+      "loss": 0.9374,
+      "step": 23248
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0114695395034834e-06,
+      "loss": 0.9708,
+      "step": 23249
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0114039181271963e-06,
+      "loss": 0.9689,
+      "step": 23250
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.011338484904544e-06,
+      "loss": 0.9598,
+      "step": 23251
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0112732398367577e-06,
+      "loss": 0.9822,
+      "step": 23252
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0112081829250677e-06,
+      "loss": 0.9264,
+      "step": 23253
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.011143314170698e-06,
+      "loss": 1.0426,
+      "step": 23254
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.011078633574871e-06,
+      "loss": 0.9633,
+      "step": 23255
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0110141411388065e-06,
+      "loss": 0.838,
+      "step": 23256
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0109498368637183e-06,
+      "loss": 0.9712,
+      "step": 23257
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.010885720750817e-06,
+      "loss": 0.9123,
+      "step": 23258
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.010821792801311e-06,
+      "loss": 0.9982,
+      "step": 23259
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0107580530164054e-06,
+      "loss": 0.9166,
+      "step": 23260
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.010694501397298e-06,
+      "loss": 1.0291,
+      "step": 23261
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0106311379451893e-06,
+      "loss": 1.0125,
+      "step": 23262
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.010567962661271e-06,
+      "loss": 0.9425,
+      "step": 23263
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0105049755467335e-06,
+      "loss": 1.0307,
+      "step": 23264
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.010442176602764e-06,
+      "loss": 0.9673,
+      "step": 23265
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.010379565830544e-06,
+      "loss": 0.9532,
+      "step": 23266
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.010317143231254e-06,
+      "loss": 0.9872,
+      "step": 23267
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0102549088060687e-06,
+      "loss": 0.8934,
+      "step": 23268
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.010192862556162e-06,
+      "loss": 0.9723,
+      "step": 23269
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.010131004482702e-06,
+      "loss": 1.0197,
+      "step": 23270
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.010069334586854e-06,
+      "loss": 0.9844,
+      "step": 23271
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0100078528697787e-06,
+      "loss": 0.9876,
+      "step": 23272
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0099465593326354e-06,
+      "loss": 0.9451,
+      "step": 23273
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.009885453976577e-06,
+      "loss": 0.9451,
+      "step": 23274
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.009824536802757e-06,
+      "loss": 0.9834,
+      "step": 23275
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0097638078123206e-06,
+      "loss": 0.9605,
+      "step": 23276
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0097032670064143e-06,
+      "loss": 1.0397,
+      "step": 23277
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0096429143861753e-06,
+      "loss": 1.0238,
+      "step": 23278
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0095827499527433e-06,
+      "loss": 0.9904,
+      "step": 23279
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.00952277370725e-06,
+      "loss": 0.8861,
+      "step": 23280
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0094629856508253e-06,
+      "loss": 1.0093,
+      "step": 23281
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0094033857845965e-06,
+      "loss": 0.9461,
+      "step": 23282
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0093439741096857e-06,
+      "loss": 0.939,
+      "step": 23283
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.009284750627211e-06,
+      "loss": 0.9383,
+      "step": 23284
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.009225715338289e-06,
+      "loss": 1.0398,
+      "step": 23285
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0091668682440313e-06,
+      "loss": 0.944,
+      "step": 23286
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0091082093455475e-06,
+      "loss": 0.9453,
+      "step": 23287
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0090497386439402e-06,
+      "loss": 0.998,
+      "step": 23288
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0089914561403136e-06,
+      "loss": 0.9708,
+      "step": 23289
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0089333618357633e-06,
+      "loss": 1.0073,
+      "step": 23290
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0088754557313854e-06,
+      "loss": 0.9414,
+      "step": 23291
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0088177378282695e-06,
+      "loss": 0.9856,
+      "step": 23292
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.008760208127503e-06,
+      "loss": 0.9826,
+      "step": 23293
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.00870286663017e-06,
+      "loss": 1.0334,
+      "step": 23294
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0086457133373506e-06,
+      "loss": 0.8992,
+      "step": 23295
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.00858874825012e-06,
+      "loss": 0.9727,
+      "step": 23296
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.008531971369554e-06,
+      "loss": 0.9971,
+      "step": 23297
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.00847538269672e-06,
+      "loss": 1.0848,
+      "step": 23298
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.008418982232683e-06,
+      "loss": 0.9209,
+      "step": 23299
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.008362769978508e-06,
+      "loss": 1.0107,
+      "step": 23300
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0083067459352537e-06,
+      "loss": 0.9177,
+      "step": 23301
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0082509101039728e-06,
+      "loss": 0.9656,
+      "step": 23302
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0081952624857197e-06,
+      "loss": 0.9467,
+      "step": 23303
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0081398030815404e-06,
+      "loss": 0.9416,
+      "step": 23304
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.008084531892483e-06,
+      "loss": 0.9308,
+      "step": 23305
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.008029448919584e-06,
+      "loss": 0.9573,
+      "step": 23306
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0079745541638843e-06,
+      "loss": 1.0222,
+      "step": 23307
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0079198476264173e-06,
+      "loss": 0.9808,
+      "step": 23308
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.007865329308213e-06,
+      "loss": 1.0267,
+      "step": 23309
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.007810999210299e-06,
+      "loss": 0.8916,
+      "step": 23310
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.007756857333698e-06,
+      "loss": 0.9923,
+      "step": 23311
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0077029036794304e-06,
+      "loss": 0.9368,
+      "step": 23312
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0076491382485133e-06,
+      "loss": 1.0009,
+      "step": 23313
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0075955610419564e-06,
+      "loss": 0.9583,
+      "step": 23314
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0075421720607724e-06,
+      "loss": 1.0276,
+      "step": 23315
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0074889713059654e-06,
+      "loss": 1.0364,
+      "step": 23316
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0074359587785375e-06,
+      "loss": 1.028,
+      "step": 23317
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.007383134479487e-06,
+      "loss": 0.9831,
+      "step": 23318
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0073304984098108e-06,
+      "loss": 0.9596,
+      "step": 23319
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0072780505704983e-06,
+      "loss": 1.011,
+      "step": 23320
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0072257909625384e-06,
+      "loss": 1.0645,
+      "step": 23321
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0071737195869147e-06,
+      "loss": 1.0241,
+      "step": 23322
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.007121836444609e-06,
+      "loss": 0.9838,
+      "step": 23323
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.007070141536598e-06,
+      "loss": 1.0324,
+      "step": 23324
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0070186348638568e-06,
+      "loss": 0.9554,
+      "step": 23325
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.006967316427353e-06,
+      "loss": 0.9627,
+      "step": 23326
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0069161862280563e-06,
+      "loss": 0.9558,
+      "step": 23327
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.006865244266928e-06,
+      "loss": 1.0226,
+      "step": 23328
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0068144905449275e-06,
+      "loss": 0.9625,
+      "step": 23329
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.006763925063011e-06,
+      "loss": 1.0212,
+      "step": 23330
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0067135478221326e-06,
+      "loss": 1.0397,
+      "step": 23331
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.00666335882324e-06,
+      "loss": 1.0391,
+      "step": 23332
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0066133580672777e-06,
+      "loss": 0.9472,
+      "step": 23333
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0065635455551892e-06,
+      "loss": 0.9896,
+      "step": 23334
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0065139212879126e-06,
+      "loss": 0.9902,
+      "step": 23335
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.006464485266381e-06,
+      "loss": 0.9536,
+      "step": 23336
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0064152374915282e-06,
+      "loss": 1.0021,
+      "step": 23337
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.006366177964279e-06,
+      "loss": 1.0328,
+      "step": 23338
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.00631730668556e-06,
+      "loss": 0.9644,
+      "step": 23339
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0062686236562904e-06,
+      "loss": 1.015,
+      "step": 23340
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0062201288773878e-06,
+      "loss": 1.0278,
+      "step": 23341
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0061718223497653e-06,
+      "loss": 1.0969,
+      "step": 23342
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0061237040743332e-06,
+      "loss": 1.023,
+      "step": 23343
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.006075774051998e-06,
+      "loss": 1.011,
+      "step": 23344
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.006028032283661e-06,
+      "loss": 0.94,
+      "step": 23345
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0059804787702245e-06,
+      "loss": 0.9252,
+      "step": 23346
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.005933113512582e-06,
+      "loss": 0.9777,
+      "step": 23347
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.005885936511626e-06,
+      "loss": 0.9276,
+      "step": 23348
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.005838947768245e-06,
+      "loss": 0.9754,
+      "step": 23349
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0057921472833257e-06,
+      "loss": 1.0009,
+      "step": 23350
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.005745535057747e-06,
+      "loss": 0.9629,
+      "step": 23351
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0056991110923897e-06,
+      "loss": 0.9837,
+      "step": 23352
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.005652875388127e-06,
+      "loss": 0.9604,
+      "step": 23353
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.00560682794583e-06,
+      "loss": 1.022,
+      "step": 23354
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0055609687663647e-06,
+      "loss": 0.9939,
+      "step": 23355
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0055152978505977e-06,
+      "loss": 0.9886,
+      "step": 23356
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.005469815199387e-06,
+      "loss": 0.9351,
+      "step": 23357
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0054245208135903e-06,
+      "loss": 1.0218,
+      "step": 23358
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.00537941469406e-06,
+      "loss": 1.0264,
+      "step": 23359
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.005334496841648e-06,
+      "loss": 0.9823,
+      "step": 23360
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0052897672571972e-06,
+      "loss": 0.9568,
+      "step": 23361
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0052452259415524e-06,
+      "loss": 0.9625,
+      "step": 23362
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.005200872895552e-06,
+      "loss": 0.9251,
+      "step": 23363
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.005156708120032e-06,
+      "loss": 0.996,
+      "step": 23364
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.005112731615823e-06,
+      "loss": 0.994,
+      "step": 23365
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.005068943383755e-06,
+      "loss": 0.975,
+      "step": 23366
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0050253434246515e-06,
+      "loss": 1.0414,
+      "step": 23367
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.004981931739334e-06,
+      "loss": 0.9186,
+      "step": 23368
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.004938708328621e-06,
+      "loss": 0.9838,
+      "step": 23369
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.004895673193326e-06,
+      "loss": 0.9995,
+      "step": 23370
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0048528263342606e-06,
+      "loss": 1.0863,
+      "step": 23371
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0048101677522305e-06,
+      "loss": 0.9783,
+      "step": 23372
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0047676974480392e-06,
+      "loss": 1.0055,
+      "step": 23373
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.004725415422488e-06,
+      "loss": 0.9496,
+      "step": 23374
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.004683321676374e-06,
+      "loss": 0.997,
+      "step": 23375
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.004641416210488e-06,
+      "loss": 0.9662,
+      "step": 23376
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.004599699025619e-06,
+      "loss": 0.9722,
+      "step": 23377
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.004558170122556e-06,
+      "loss": 0.972,
+      "step": 23378
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.004516829502078e-06,
+      "loss": 0.9412,
+      "step": 23379
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0044756771649667e-06,
+      "loss": 0.9792,
+      "step": 23380
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0044347131119945e-06,
+      "loss": 0.9145,
+      "step": 23381
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0043939373439337e-06,
+      "loss": 0.9966,
+      "step": 23382
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0043533498615544e-06,
+      "loss": 0.9787,
+      "step": 23383
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.004312950665618e-06,
+      "loss": 1.0086,
+      "step": 23384
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0042727397568887e-06,
+      "loss": 0.8953,
+      "step": 23385
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.004232717136121e-06,
+      "loss": 0.9429,
+      "step": 23386
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0041928828040715e-06,
+      "loss": 0.9255,
+      "step": 23387
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.004153236761488e-06,
+      "loss": 0.9626,
+      "step": 23388
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0041137790091193e-06,
+      "loss": 0.9265,
+      "step": 23389
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0040745095477078e-06,
+      "loss": 1.0302,
+      "step": 23390
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.004035428377993e-06,
+      "loss": 1.0057,
+      "step": 23391
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.003996535500712e-06,
+      "loss": 1.0519,
+      "step": 23392
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0039578309165967e-06,
+      "loss": 0.8811,
+      "step": 23393
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.003919314626375e-06,
+      "loss": 1.015,
+      "step": 23394
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0038809866307743e-06,
+      "loss": 0.9671,
+      "step": 23395
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0038428469305173e-06,
+      "loss": 0.9397,
+      "step": 23396
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.003804895526319e-06,
+      "loss": 0.9237,
+      "step": 23397
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.003767132418898e-06,
+      "loss": 0.9495,
+      "step": 23398
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0037295576089633e-06,
+      "loss": 0.9495,
+      "step": 23399
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0036921710972238e-06,
+      "loss": 1.012,
+      "step": 23400
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0036549728843826e-06,
+      "loss": 0.9592,
+      "step": 23401
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.003617962971142e-06,
+      "loss": 0.9918,
+      "step": 23402
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0035811413581974e-06,
+      "loss": 0.9952,
+      "step": 23403
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0035445080462447e-06,
+      "loss": 0.9793,
+      "step": 23404
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.003508063035972e-06,
+      "loss": 1.0085,
+      "step": 23405
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.003471806328066e-06,
+      "loss": 0.96,
+      "step": 23406
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0034357379232097e-06,
+      "loss": 0.9638,
+      "step": 23407
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.003399857822084e-06,
+      "loss": 0.9765,
+      "step": 23408
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0033641660253623e-06,
+      "loss": 0.8441,
+      "step": 23409
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0033286625337192e-06,
+      "loss": 0.908,
+      "step": 23410
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0032933473478223e-06,
+      "loss": 1.078,
+      "step": 23411
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.003258220468336e-06,
+      "loss": 1.0296,
+      "step": 23412
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0032232818959243e-06,
+      "loss": 0.9064,
+      "step": 23413
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.003188531631244e-06,
+      "loss": 1.0278,
+      "step": 23414
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0031539696749492e-06,
+      "loss": 0.9701,
+      "step": 23415
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.003119596027692e-06,
+      "loss": 0.9912,
+      "step": 23416
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.003085410690119e-06,
+      "loss": 0.9311,
+      "step": 23417
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0030514136628746e-06,
+      "loss": 1.0142,
+      "step": 23418
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.003017604946599e-06,
+      "loss": 0.9475,
+      "step": 23419
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0029839845419288e-06,
+      "loss": 1.0039,
+      "step": 23420
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0029505524494974e-06,
+      "loss": 0.9365,
+      "step": 23421
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.002917308669936e-06,
+      "loss": 0.916,
+      "step": 23422
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0028842532038697e-06,
+      "loss": 0.9421,
+      "step": 23423
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0028513860519197e-06,
+      "loss": 1.0316,
+      "step": 23424
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.002818707214708e-06,
+      "loss": 0.9933,
+      "step": 23425
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.002786216692847e-06,
+      "loss": 0.9344,
+      "step": 23426
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0027539144869512e-06,
+      "loss": 0.9677,
+      "step": 23427
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0027218005976294e-06,
+      "loss": 1.0079,
+      "step": 23428
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.002689875025483e-06,
+      "loss": 0.9826,
+      "step": 23429
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0026581377711176e-06,
+      "loss": 0.9805,
+      "step": 23430
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0026265888351283e-06,
+      "loss": 1.0104,
+      "step": 23431
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0025952282181107e-06,
+      "loss": 1.0744,
+      "step": 23432
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.002564055920655e-06,
+      "loss": 0.9261,
+      "step": 23433
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.002533071943349e-06,
+      "loss": 1.0196,
+      "step": 23434
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0025022762867754e-06,
+      "loss": 0.9421,
+      "step": 23435
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0024716689515145e-06,
+      "loss": 1.0152,
+      "step": 23436
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0024412499381433e-06,
+      "loss": 1.01,
+      "step": 23437
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.002411019247235e-06,
+      "loss": 0.9138,
+      "step": 23438
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.002380976879358e-06,
+      "loss": 0.9636,
+      "step": 23439
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.002351122835078e-06,
+      "loss": 1.0405,
+      "step": 23440
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.00232145711496e-06,
+      "loss": 1.0192,
+      "step": 23441
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0022919797195597e-06,
+      "loss": 1.0374,
+      "step": 23442
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0022626906494336e-06,
+      "loss": 0.9993,
+      "step": 23443
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0022335899051337e-06,
+      "loss": 0.9684,
+      "step": 23444
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0022046774872087e-06,
+      "loss": 0.9542,
+      "step": 23445
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0021759533962018e-06,
+      "loss": 0.9556,
+      "step": 23446
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0021474176326542e-06,
+      "loss": 0.9285,
+      "step": 23447
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.002119070197105e-06,
+      "loss": 0.9413,
+      "step": 23448
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0020909110900866e-06,
+      "loss": 0.9059,
+      "step": 23449
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0020629403121296e-06,
+      "loss": 1.0295,
+      "step": 23450
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.002035157863761e-06,
+      "loss": 1.0576,
+      "step": 23451
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.002007563745505e-06,
+      "loss": 0.9958,
+      "step": 23452
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0019801579578816e-06,
+      "loss": 1.013,
+      "step": 23453
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.001952940501405e-06,
+      "loss": 0.9396,
+      "step": 23454
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.001925911376589e-06,
+      "loss": 1.0186,
+      "step": 23455
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0018990705839437e-06,
+      "loss": 0.9845,
+      "step": 23456
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0018724181239725e-06,
+      "loss": 0.9948,
+      "step": 23457
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0018459539971803e-06,
+      "loss": 0.9893,
+      "step": 23458
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0018196782040626e-06,
+      "loss": 0.9723,
+      "step": 23459
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.001793590745117e-06,
+      "loss": 0.998,
+      "step": 23460
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.001767691620833e-06,
+      "loss": 1.1274,
+      "step": 23461
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0017419808316996e-06,
+      "loss": 1.0528,
+      "step": 23462
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0017164583782002e-06,
+      "loss": 1.0251,
+      "step": 23463
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0016911242608165e-06,
+      "loss": 0.9021,
+      "step": 23464
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0016659784800244e-06,
+      "loss": 0.9488,
+      "step": 23465
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0016410210362992e-06,
+      "loss": 1.0163,
+      "step": 23466
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0016162519301097e-06,
+      "loss": 0.9464,
+      "step": 23467
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0015916711619243e-06,
+      "loss": 0.916,
+      "step": 23468
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0015672787322038e-06,
+      "loss": 1.0416,
+      "step": 23469
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.001543074641409e-06,
+      "loss": 0.9653,
+      "step": 23470
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.001519058889995e-06,
+      "loss": 1.0167,
+      "step": 23471
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0014952314784144e-06,
+      "loss": 0.9653,
+      "step": 23472
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.001471592407117e-06,
+      "loss": 0.9352,
+      "step": 23473
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0014481416765464e-06,
+      "loss": 1.0544,
+      "step": 23474
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.001424879287146e-06,
+      "loss": 1.0138,
+      "step": 23475
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.001401805239354e-06,
+      "loss": 1.0352,
+      "step": 23476
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.001378919533603e-06,
+      "loss": 0.9265,
+      "step": 23477
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0013562221703267e-06,
+      "loss": 0.99,
+      "step": 23478
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0013337131499503e-06,
+      "loss": 0.903,
+      "step": 23479
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.001311392472899e-06,
+      "loss": 0.9564,
+      "step": 23480
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.001289260139593e-06,
+      "loss": 0.9583,
+      "step": 23481
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0012673161504494e-06,
+      "loss": 0.9922,
+      "step": 23482
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0012455605058825e-06,
+      "loss": 1.0187,
+      "step": 23483
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0012239932062996e-06,
+      "loss": 0.8943,
+      "step": 23484
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.001202614252109e-06,
+      "loss": 0.9827,
+      "step": 23485
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0011814236437133e-06,
+      "loss": 1.0345,
+      "step": 23486
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0011604213815107e-06,
+      "loss": 0.936,
+      "step": 23487
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0011396074658973e-06,
+      "loss": 0.9281,
+      "step": 23488
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0011189818972657e-06,
+      "loss": 1.0062,
+      "step": 23489
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0010985446760034e-06,
+      "loss": 1.0223,
+      "step": 23490
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0010782958024967e-06,
+      "loss": 0.9323,
+      "step": 23491
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.001058235277126e-06,
+      "loss": 0.9765,
+      "step": 23492
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0010383631002687e-06,
+      "loss": 0.8968,
+      "step": 23493
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0010186792723e-06,
+      "loss": 0.9338,
+      "step": 23494
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0009991837935908e-06,
+      "loss": 0.9674,
+      "step": 23495
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0009798766645073e-06,
+      "loss": 1.0093,
+      "step": 23496
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0009607578854152e-06,
+      "loss": 0.9829,
+      "step": 23497
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000941827456672e-06,
+      "loss": 0.9924,
+      "step": 23498
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000923085378637e-06,
+      "loss": 1.0174,
+      "step": 23499
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0009045316516608e-06,
+      "loss": 0.9222,
+      "step": 23500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000886166276095e-06,
+      "loss": 1.0086,
+      "step": 23501
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0008679892522833e-06,
+      "loss": 0.945,
+      "step": 23502
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.00085000058057e-06,
+      "loss": 0.9271,
+      "step": 23503
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0008322002612935e-06,
+      "loss": 0.9541,
+      "step": 23504
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0008145882947886e-06,
+      "loss": 1.0516,
+      "step": 23505
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0007971646813875e-06,
+      "loss": 0.9845,
+      "step": 23506
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0007799294214183e-06,
+      "loss": 0.9316,
+      "step": 23507
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000762882515205e-06,
+      "loss": 1.043,
+      "step": 23508
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0007460239630696e-06,
+      "loss": 0.9756,
+      "step": 23509
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0007293537653297e-06,
+      "loss": 0.8675,
+      "step": 23510
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0007128719222994e-06,
+      "loss": 0.9944,
+      "step": 23511
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000696578434287e-06,
+      "loss": 0.9696,
+      "step": 23512
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000680473301603e-06,
+      "loss": 0.9606,
+      "step": 23513
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0006645565245475e-06,
+      "loss": 0.9746,
+      "step": 23514
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000648828103423e-06,
+      "loss": 0.981,
+      "step": 23515
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000633288038524e-06,
+      "loss": 0.9917,
+      "step": 23516
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000617936330144e-06,
+      "loss": 0.9783,
+      "step": 23517
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000602772978572e-06,
+      "loss": 0.9752,
+      "step": 23518
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0005877979840927e-06,
+      "loss": 0.9647,
+      "step": 23519
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0005730113469904e-06,
+      "loss": 0.9396,
+      "step": 23520
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0005584130675417e-06,
+      "loss": 1.0448,
+      "step": 23521
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000544003146021e-06,
+      "loss": 0.8756,
+      "step": 23522
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000529781582702e-06,
+      "loss": 0.9538,
+      "step": 23523
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0005157483778518e-06,
+      "loss": 0.9989,
+      "step": 23524
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0005019035317342e-06,
+      "loss": 0.9815,
+      "step": 23525
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.00048824704461e-06,
+      "loss": 1.0096,
+      "step": 23526
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000474778916737e-06,
+      "loss": 0.9806,
+      "step": 23527
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0004614991483688e-06,
+      "loss": 0.9914,
+      "step": 23528
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0004484077397536e-06,
+      "loss": 0.916,
+      "step": 23529
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000435504691142e-06,
+      "loss": 1.0544,
+      "step": 23530
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000422790002773e-06,
+      "loss": 0.9886,
+      "step": 23531
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000410263674889e-06,
+      "loss": 0.9521,
+      "step": 23532
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0003979257077248e-06,
+      "loss": 0.9777,
+      "step": 23533
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0003857761015123e-06,
+      "loss": 1.0325,
+      "step": 23534
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000373814856481e-06,
+      "loss": 0.9232,
+      "step": 23535
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0003620419728565e-06,
+      "loss": 0.9914,
+      "step": 23536
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0003504574508592e-06,
+      "loss": 0.9372,
+      "step": 23537
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000339061290709e-06,
+      "loss": 0.9908,
+      "step": 23538
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.00032785349262e-06,
+      "loss": 1.0173,
+      "step": 23539
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0003168340568037e-06,
+      "loss": 1.0298,
+      "step": 23540
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0003060029834668e-06,
+      "loss": 1.0388,
+      "step": 23541
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0002953602728126e-06,
+      "loss": 0.9958,
+      "step": 23542
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000284905925044e-06,
+      "loss": 0.9884,
+      "step": 23543
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000274639940355e-06,
+      "loss": 0.8914,
+      "step": 23544
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0002645623189423e-06,
+      "loss": 0.9916,
+      "step": 23545
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000254673060993e-06,
+      "loss": 0.9404,
+      "step": 23546
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0002449721666952e-06,
+      "loss": 0.8876,
+      "step": 23547
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0002354596362296e-06,
+      "loss": 0.9283,
+      "step": 23548
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0002261354697776e-06,
+      "loss": 1.0391,
+      "step": 23549
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000216999667514e-06,
+      "loss": 1.07,
+      "step": 23550
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.00020805222961e-06,
+      "loss": 0.9743,
+      "step": 23551
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000199293156234e-06,
+      "loss": 0.9685,
+      "step": 23552
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000190722447554e-06,
+      "loss": 0.9683,
+      "step": 23553
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0001823401037283e-06,
+      "loss": 1.0293,
+      "step": 23554
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000174146124916e-06,
+      "loss": 0.9607,
+      "step": 23555
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0001661405112707e-06,
+      "loss": 0.9534,
+      "step": 23556
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000158323262944e-06,
+      "loss": 0.9857,
+      "step": 23557
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000150694380083e-06,
+      "loss": 0.9823,
+      "step": 23558
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0001432538628308e-06,
+      "loss": 0.9711,
+      "step": 23559
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0001360017113283e-06,
+      "loss": 1.0065,
+      "step": 23560
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0001289379257117e-06,
+      "loss": 1.0498,
+      "step": 23561
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000122062506115e-06,
+      "loss": 1.0211,
+      "step": 23562
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0001153754526672e-06,
+      "loss": 0.9295,
+      "step": 23563
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000108876765493e-06,
+      "loss": 1.0089,
+      "step": 23564
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0001025664447163e-06,
+      "loss": 0.9781,
+      "step": 23565
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000096444490455e-06,
+      "loss": 0.9773,
+      "step": 23566
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000905109028247e-06,
+      "loss": 0.9672,
+      "step": 23567
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000084765681938e-06,
+      "loss": 0.9736,
+      "step": 23568
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000792088279032e-06,
+      "loss": 0.9688,
+      "step": 23569
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000073840340823e-06,
+      "loss": 1.0168,
+      "step": 23570
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000686602208e-06,
+      "loss": 1.0,
+      "step": 23571
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000063668467932e-06,
+      "loss": 0.937,
+      "step": 23572
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000058865082312e-06,
+      "loss": 0.9535,
+      "step": 23573
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000054250064032e-06,
+      "loss": 0.99,
+      "step": 23574
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000498234131775e-06,
+      "loss": 0.9488,
+      "step": 23575
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000045585129832e-06,
+      "loss": 0.9628,
+      "step": 23576
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000415352140756e-06,
+      "loss": 0.9675,
+      "step": 23577
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000376736659857e-06,
+      "loss": 1.0027,
+      "step": 23578
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000340004856336e-06,
+      "loss": 0.9619,
+      "step": 23579
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000305156730894e-06,
+      "loss": 0.9636,
+      "step": 23580
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000272192284178e-06,
+      "loss": 0.9498,
+      "step": 23581
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000241111516805e-06,
+      "loss": 0.9768,
+      "step": 23582
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000211914429383e-06,
+      "loss": 1.0178,
+      "step": 23583
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000184601022445e-06,
+      "loss": 0.9505,
+      "step": 23584
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000015917129651e-06,
+      "loss": 0.9989,
+      "step": 23585
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000013562525205e-06,
+      "loss": 0.9585,
+      "step": 23586
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000113962889515e-06,
+      "loss": 0.9139,
+      "step": 23587
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000009418420932e-06,
+      "loss": 0.9198,
+      "step": 23588
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000076289211823e-06,
+      "loss": 1.01,
+      "step": 23589
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000060277897374e-06,
+      "loss": 0.9285,
+      "step": 23590
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000046150266263e-06,
+      "loss": 1.0297,
+      "step": 23591
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000003390631876e-06,
+      "loss": 1.0032,
+      "step": 23592
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000023546055106e-06,
+      "loss": 1.0237,
+      "step": 23593
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000001506947548e-06,
+      "loss": 0.9214,
+      "step": 23594
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000000847658005e-06,
+      "loss": 0.9656,
+      "step": 23595
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000003767368937e-06,
+      "loss": 0.9009,
+      "step": 23596
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000000941842238e-06,
+      "loss": 1.0026,
+      "step": 23597
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2e-06,
+      "loss": 1.0206,
+      "step": 23598
+    },
+    {
+      "epoch": 1.0,
+      "step": 23598,
+      "total_flos": 8.286827072173638e+19,
+      "train_loss": 1.063429206137395,
+      "train_runtime": 76734.7094,
+      "train_samples_per_second": 39.363,
+      "train_steps_per_second": 0.308
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 23598,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 100000,
+  "total_flos": 8.286827072173638e+19,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}