{
  "best_metric": 0.4106853980417199,
  "best_model_checkpoint": "ansilmbabl/cards-swin-tiny-patch4-window7-224-finetuned-v1/checkpoint-7340",
  "epoch": 10.0,
  "eval_steps": 500,
  "global_step": 7340,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 6.811989100817439e-07,
      "loss": 2.2703,
      "step": 10
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.3623978201634877e-06,
      "loss": 2.2502,
      "step": 20
    },
    {
      "epoch": 0.04,
      "learning_rate": 2.0435967302452317e-06,
      "loss": 2.2204,
      "step": 30
    },
    {
      "epoch": 0.05,
      "learning_rate": 2.7247956403269755e-06,
      "loss": 2.1483,
      "step": 40
    },
    {
      "epoch": 0.07,
      "learning_rate": 3.4059945504087197e-06,
      "loss": 2.0802,
      "step": 50
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.0871934604904634e-06,
      "loss": 2.0193,
      "step": 60
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.768392370572208e-06,
      "loss": 1.9472,
      "step": 70
    },
    {
      "epoch": 0.11,
      "learning_rate": 5.449591280653951e-06,
      "loss": 1.9118,
      "step": 80
    },
    {
      "epoch": 0.12,
      "learning_rate": 6.1307901907356956e-06,
      "loss": 1.8348,
      "step": 90
    },
    {
      "epoch": 0.14,
      "learning_rate": 6.811989100817439e-06,
      "loss": 1.8327,
      "step": 100
    },
    {
      "epoch": 0.15,
      "learning_rate": 7.493188010899182e-06,
      "loss": 1.7925,
      "step": 110
    },
    {
      "epoch": 0.16,
      "learning_rate": 8.174386920980927e-06,
      "loss": 1.7675,
      "step": 120
    },
    {
      "epoch": 0.18,
      "learning_rate": 8.85558583106267e-06,
      "loss": 1.7498,
      "step": 130
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.536784741144416e-06,
      "loss": 1.7557,
      "step": 140
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.0217983651226158e-05,
      "loss": 1.7356,
      "step": 150
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.0899182561307902e-05,
      "loss": 1.7307,
      "step": 160
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.1580381471389647e-05,
      "loss": 1.7072,
      "step": 170
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.2261580381471391e-05,
      "loss": 1.7258,
      "step": 180
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.2942779291553133e-05,
      "loss": 1.6938,
      "step": 190
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.3623978201634879e-05,
      "loss": 1.7356,
      "step": 200
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.4305177111716622e-05,
      "loss": 1.7092,
      "step": 210
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.4986376021798364e-05,
      "loss": 1.702,
      "step": 220
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.566757493188011e-05,
      "loss": 1.6801,
      "step": 230
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.6348773841961854e-05,
      "loss": 1.6869,
      "step": 240
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.7029972752043597e-05,
      "loss": 1.6762,
      "step": 250
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.771117166212534e-05,
      "loss": 1.6727,
      "step": 260
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.8392370572207085e-05,
      "loss": 1.6738,
      "step": 270
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.9073569482288832e-05,
      "loss": 1.6761,
      "step": 280
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.9754768392370573e-05,
      "loss": 1.6845,
      "step": 290
    },
    {
      "epoch": 0.41,
      "learning_rate": 2.0435967302452316e-05,
      "loss": 1.6963,
      "step": 300
    },
    {
      "epoch": 0.42,
      "learning_rate": 2.1117166212534063e-05,
      "loss": 1.6704,
      "step": 310
    },
    {
      "epoch": 0.44,
      "learning_rate": 2.1798365122615804e-05,
      "loss": 1.6535,
      "step": 320
    },
    {
      "epoch": 0.45,
      "learning_rate": 2.2479564032697548e-05,
      "loss": 1.6686,
      "step": 330
    },
    {
      "epoch": 0.46,
      "learning_rate": 2.3160762942779295e-05,
      "loss": 1.6805,
      "step": 340
    },
    {
      "epoch": 0.48,
      "learning_rate": 2.3841961852861035e-05,
      "loss": 1.6854,
      "step": 350
    },
    {
      "epoch": 0.49,
      "learning_rate": 2.4523160762942782e-05,
      "loss": 1.6753,
      "step": 360
    },
    {
      "epoch": 0.5,
      "learning_rate": 2.5204359673024526e-05,
      "loss": 1.6314,
      "step": 370
    },
    {
      "epoch": 0.52,
      "learning_rate": 2.5885558583106266e-05,
      "loss": 1.6485,
      "step": 380
    },
    {
      "epoch": 0.53,
      "learning_rate": 2.656675749318801e-05,
      "loss": 1.6729,
      "step": 390
    },
    {
      "epoch": 0.54,
      "learning_rate": 2.7247956403269757e-05,
      "loss": 1.6659,
      "step": 400
    },
    {
      "epoch": 0.56,
      "learning_rate": 2.79291553133515e-05,
      "loss": 1.659,
      "step": 410
    },
    {
      "epoch": 0.57,
      "learning_rate": 2.8610354223433245e-05,
      "loss": 1.6549,
      "step": 420
    },
    {
      "epoch": 0.59,
      "learning_rate": 2.929155313351499e-05,
      "loss": 1.6487,
      "step": 430
    },
    {
      "epoch": 0.6,
      "learning_rate": 2.997275204359673e-05,
      "loss": 1.6431,
      "step": 440
    },
    {
      "epoch": 0.61,
      "learning_rate": 3.065395095367847e-05,
      "loss": 1.6617,
      "step": 450
    },
    {
      "epoch": 0.63,
      "learning_rate": 3.133514986376022e-05,
      "loss": 1.6489,
      "step": 460
    },
    {
      "epoch": 0.64,
      "learning_rate": 3.201634877384197e-05,
      "loss": 1.6899,
      "step": 470
    },
    {
      "epoch": 0.65,
      "learning_rate": 3.269754768392371e-05,
      "loss": 1.6569,
      "step": 480
    },
    {
      "epoch": 0.67,
      "learning_rate": 3.337874659400545e-05,
      "loss": 1.656,
      "step": 490
    },
    {
      "epoch": 0.68,
      "learning_rate": 3.4059945504087195e-05,
      "loss": 1.6335,
      "step": 500
    },
    {
      "epoch": 0.69,
      "learning_rate": 3.474114441416894e-05,
      "loss": 1.6709,
      "step": 510
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.542234332425068e-05,
      "loss": 1.6408,
      "step": 520
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.610354223433243e-05,
      "loss": 1.6194,
      "step": 530
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.678474114441417e-05,
      "loss": 1.6722,
      "step": 540
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.746594005449591e-05,
      "loss": 1.6446,
      "step": 550
    },
    {
      "epoch": 0.76,
      "learning_rate": 3.8147138964577664e-05,
      "loss": 1.6414,
      "step": 560
    },
    {
      "epoch": 0.78,
      "learning_rate": 3.8828337874659405e-05,
      "loss": 1.6127,
      "step": 570
    },
    {
      "epoch": 0.79,
      "learning_rate": 3.9509536784741145e-05,
      "loss": 1.643,
      "step": 580
    },
    {
      "epoch": 0.8,
      "learning_rate": 4.019073569482289e-05,
      "loss": 1.6478,
      "step": 590
    },
    {
      "epoch": 0.82,
      "learning_rate": 4.087193460490463e-05,
      "loss": 1.6699,
      "step": 600
    },
    {
      "epoch": 0.83,
      "learning_rate": 4.155313351498637e-05,
      "loss": 1.6393,
      "step": 610
    },
    {
      "epoch": 0.84,
      "learning_rate": 4.223433242506813e-05,
      "loss": 1.6658,
      "step": 620
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.291553133514987e-05,
      "loss": 1.6263,
      "step": 630
    },
    {
      "epoch": 0.87,
      "learning_rate": 4.359673024523161e-05,
      "loss": 1.6329,
      "step": 640
    },
    {
      "epoch": 0.89,
      "learning_rate": 4.4277929155313355e-05,
      "loss": 1.6403,
      "step": 650
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.4959128065395095e-05,
      "loss": 1.6325,
      "step": 660
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.564032697547684e-05,
      "loss": 1.6576,
      "step": 670
    },
    {
      "epoch": 0.93,
      "learning_rate": 4.632152588555859e-05,
      "loss": 1.6405,
      "step": 680
    },
    {
      "epoch": 0.94,
      "learning_rate": 4.700272479564033e-05,
      "loss": 1.62,
      "step": 690
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.768392370572207e-05,
      "loss": 1.6096,
      "step": 700
    },
    {
      "epoch": 0.97,
      "learning_rate": 4.836512261580382e-05,
      "loss": 1.6386,
      "step": 710
    },
    {
      "epoch": 0.98,
      "learning_rate": 4.9046321525885565e-05,
      "loss": 1.6276,
      "step": 720
    },
    {
      "epoch": 0.99,
      "learning_rate": 4.9727520435967305e-05,
      "loss": 1.6417,
      "step": 730
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.3000425713069391,
      "eval_loss": 1.607536792755127,
      "eval_runtime": 340.2193,
      "eval_samples_per_second": 69.044,
      "eval_steps_per_second": 2.16,
      "step": 734
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.995458673932788e-05,
      "loss": 1.6447,
      "step": 740
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.987889797154103e-05,
      "loss": 1.6226,
      "step": 750
    },
    {
      "epoch": 1.04,
      "learning_rate": 4.9803209203754164e-05,
      "loss": 1.6242,
      "step": 760
    },
    {
      "epoch": 1.05,
      "learning_rate": 4.9727520435967305e-05,
      "loss": 1.6276,
      "step": 770
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.9651831668180446e-05,
      "loss": 1.6294,
      "step": 780
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.9576142900393586e-05,
      "loss": 1.6251,
      "step": 790
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.950045413260672e-05,
      "loss": 1.6196,
      "step": 800
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.942476536481986e-05,
      "loss": 1.6354,
      "step": 810
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.9349076597033e-05,
      "loss": 1.5853,
      "step": 820
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.927338782924614e-05,
      "loss": 1.6521,
      "step": 830
    },
    {
      "epoch": 1.14,
      "learning_rate": 4.9197699061459276e-05,
      "loss": 1.6335,
      "step": 840
    },
    {
      "epoch": 1.16,
      "learning_rate": 4.9122010293672424e-05,
      "loss": 1.6291,
      "step": 850
    },
    {
      "epoch": 1.17,
      "learning_rate": 4.9046321525885565e-05,
      "loss": 1.6032,
      "step": 860
    },
    {
      "epoch": 1.19,
      "learning_rate": 4.89706327580987e-05,
      "loss": 1.5978,
      "step": 870
    },
    {
      "epoch": 1.2,
      "learning_rate": 4.889494399031184e-05,
      "loss": 1.6251,
      "step": 880
    },
    {
      "epoch": 1.21,
      "learning_rate": 4.881925522252498e-05,
      "loss": 1.6434,
      "step": 890
    },
    {
      "epoch": 1.23,
      "learning_rate": 4.874356645473812e-05,
      "loss": 1.5967,
      "step": 900
    },
    {
      "epoch": 1.24,
      "learning_rate": 4.8667877686951255e-05,
      "loss": 1.6253,
      "step": 910
    },
    {
      "epoch": 1.25,
      "learning_rate": 4.85921889191644e-05,
      "loss": 1.596,
      "step": 920
    },
    {
      "epoch": 1.27,
      "learning_rate": 4.8516500151377536e-05,
      "loss": 1.6299,
      "step": 930
    },
    {
      "epoch": 1.28,
      "learning_rate": 4.844081138359068e-05,
      "loss": 1.6277,
      "step": 940
    },
    {
      "epoch": 1.29,
      "learning_rate": 4.836512261580382e-05,
      "loss": 1.6084,
      "step": 950
    },
    {
      "epoch": 1.31,
      "learning_rate": 4.828943384801696e-05,
      "loss": 1.6117,
      "step": 960
    },
    {
      "epoch": 1.32,
      "learning_rate": 4.821374508023009e-05,
      "loss": 1.6053,
      "step": 970
    },
    {
      "epoch": 1.34,
      "learning_rate": 4.813805631244323e-05,
      "loss": 1.5786,
      "step": 980
    },
    {
      "epoch": 1.35,
      "learning_rate": 4.8062367544656373e-05,
      "loss": 1.6168,
      "step": 990
    },
    {
      "epoch": 1.36,
      "learning_rate": 4.7986678776869514e-05,
      "loss": 1.6049,
      "step": 1000
    },
    {
      "epoch": 1.38,
      "learning_rate": 4.7910990009082655e-05,
      "loss": 1.6241,
      "step": 1010
    },
    {
      "epoch": 1.39,
      "learning_rate": 4.7835301241295796e-05,
      "loss": 1.6239,
      "step": 1020
    },
    {
      "epoch": 1.4,
      "learning_rate": 4.7759612473508936e-05,
      "loss": 1.6122,
      "step": 1030
    },
    {
      "epoch": 1.42,
      "learning_rate": 4.768392370572207e-05,
      "loss": 1.6283,
      "step": 1040
    },
    {
      "epoch": 1.43,
      "learning_rate": 4.760823493793522e-05,
      "loss": 1.6128,
      "step": 1050
    },
    {
      "epoch": 1.44,
      "learning_rate": 4.753254617014835e-05,
      "loss": 1.6277,
      "step": 1060
    },
    {
      "epoch": 1.46,
      "learning_rate": 4.745685740236149e-05,
      "loss": 1.5892,
      "step": 1070
    },
    {
      "epoch": 1.47,
      "learning_rate": 4.738116863457463e-05,
      "loss": 1.5945,
      "step": 1080
    },
    {
      "epoch": 1.49,
      "learning_rate": 4.7305479866787774e-05,
      "loss": 1.6154,
      "step": 1090
    },
    {
      "epoch": 1.5,
      "learning_rate": 4.722979109900091e-05,
      "loss": 1.6279,
      "step": 1100
    },
    {
      "epoch": 1.51,
      "learning_rate": 4.715410233121405e-05,
      "loss": 1.5854,
      "step": 1110
    },
    {
      "epoch": 1.53,
      "learning_rate": 4.707841356342719e-05,
      "loss": 1.5969,
      "step": 1120
    },
    {
      "epoch": 1.54,
      "learning_rate": 4.700272479564033e-05,
      "loss": 1.5948,
      "step": 1130
    },
    {
      "epoch": 1.55,
      "learning_rate": 4.692703602785347e-05,
      "loss": 1.5887,
      "step": 1140
    },
    {
      "epoch": 1.57,
      "learning_rate": 4.685134726006661e-05,
      "loss": 1.5696,
      "step": 1150
    },
    {
      "epoch": 1.58,
      "learning_rate": 4.677565849227975e-05,
      "loss": 1.6007,
      "step": 1160
    },
    {
      "epoch": 1.59,
      "learning_rate": 4.6699969724492886e-05,
      "loss": 1.6034,
      "step": 1170
    },
    {
      "epoch": 1.61,
      "learning_rate": 4.662428095670603e-05,
      "loss": 1.6126,
      "step": 1180
    },
    {
      "epoch": 1.62,
      "learning_rate": 4.654859218891917e-05,
      "loss": 1.588,
      "step": 1190
    },
    {
      "epoch": 1.63,
      "learning_rate": 4.647290342113231e-05,
      "loss": 1.5921,
      "step": 1200
    },
    {
      "epoch": 1.65,
      "learning_rate": 4.639721465334544e-05,
      "loss": 1.6069,
      "step": 1210
    },
    {
      "epoch": 1.66,
      "learning_rate": 4.632152588555859e-05,
      "loss": 1.5901,
      "step": 1220
    },
    {
      "epoch": 1.68,
      "learning_rate": 4.6245837117771723e-05,
      "loss": 1.5783,
      "step": 1230
    },
    {
      "epoch": 1.69,
      "learning_rate": 4.6170148349984864e-05,
      "loss": 1.5704,
      "step": 1240
    },
    {
      "epoch": 1.7,
      "learning_rate": 4.6094459582198005e-05,
      "loss": 1.5921,
      "step": 1250
    },
    {
      "epoch": 1.72,
      "learning_rate": 4.6018770814411146e-05,
      "loss": 1.5895,
      "step": 1260
    },
    {
      "epoch": 1.73,
      "learning_rate": 4.594308204662428e-05,
      "loss": 1.6095,
      "step": 1270
    },
    {
      "epoch": 1.74,
      "learning_rate": 4.586739327883742e-05,
      "loss": 1.6084,
      "step": 1280
    },
    {
      "epoch": 1.76,
      "learning_rate": 4.579170451105057e-05,
      "loss": 1.5739,
      "step": 1290
    },
    {
      "epoch": 1.77,
      "learning_rate": 4.57160157432637e-05,
      "loss": 1.5993,
      "step": 1300
    },
    {
      "epoch": 1.78,
      "learning_rate": 4.564032697547684e-05,
      "loss": 1.6168,
      "step": 1310
    },
    {
      "epoch": 1.8,
      "learning_rate": 4.556463820768998e-05,
      "loss": 1.5974,
      "step": 1320
    },
    {
      "epoch": 1.81,
      "learning_rate": 4.5488949439903124e-05,
      "loss": 1.597,
      "step": 1330
    },
    {
      "epoch": 1.83,
      "learning_rate": 4.541326067211626e-05,
      "loss": 1.5989,
      "step": 1340
    },
    {
      "epoch": 1.84,
      "learning_rate": 4.53375719043294e-05,
      "loss": 1.5919,
      "step": 1350
    },
    {
      "epoch": 1.85,
      "learning_rate": 4.526188313654254e-05,
      "loss": 1.5843,
      "step": 1360
    },
    {
      "epoch": 1.87,
      "learning_rate": 4.518619436875568e-05,
      "loss": 1.5793,
      "step": 1370
    },
    {
      "epoch": 1.88,
      "learning_rate": 4.5110505600968814e-05,
      "loss": 1.6142,
      "step": 1380
    },
    {
      "epoch": 1.89,
      "learning_rate": 4.503481683318196e-05,
      "loss": 1.5861,
      "step": 1390
    },
    {
      "epoch": 1.91,
      "learning_rate": 4.4959128065395095e-05,
      "loss": 1.5889,
      "step": 1400
    },
    {
      "epoch": 1.92,
      "learning_rate": 4.4883439297608236e-05,
      "loss": 1.5926,
      "step": 1410
    },
    {
      "epoch": 1.93,
      "learning_rate": 4.4807750529821377e-05,
      "loss": 1.6027,
      "step": 1420
    },
    {
      "epoch": 1.95,
      "learning_rate": 4.473206176203452e-05,
      "loss": 1.5771,
      "step": 1430
    },
    {
      "epoch": 1.96,
      "learning_rate": 4.465637299424766e-05,
      "loss": 1.5837,
      "step": 1440
    },
    {
      "epoch": 1.98,
      "learning_rate": 4.458068422646079e-05,
      "loss": 1.5849,
      "step": 1450
    },
    {
      "epoch": 1.99,
      "learning_rate": 4.450499545867394e-05,
      "loss": 1.577,
      "step": 1460
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.3355044699872286,
      "eval_loss": 1.5510696172714233,
      "eval_runtime": 347.9962,
      "eval_samples_per_second": 67.501,
      "eval_steps_per_second": 2.112,
      "step": 1468
    },
    {
      "epoch": 2.0,
      "learning_rate": 4.442930669088707e-05,
      "loss": 1.5973,
      "step": 1470
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.4353617923100214e-05,
      "loss": 1.5887,
      "step": 1480
    },
    {
      "epoch": 2.03,
      "learning_rate": 4.4277929155313355e-05,
      "loss": 1.5827,
      "step": 1490
    },
    {
      "epoch": 2.04,
      "learning_rate": 4.4202240387526495e-05,
      "loss": 1.5978,
      "step": 1500
    },
    {
      "epoch": 2.06,
      "learning_rate": 4.412655161973963e-05,
      "loss": 1.6029,
      "step": 1510
    },
    {
      "epoch": 2.07,
      "learning_rate": 4.405086285195277e-05,
      "loss": 1.5789,
      "step": 1520
    },
    {
      "epoch": 2.08,
      "learning_rate": 4.397517408416591e-05,
      "loss": 1.5892,
      "step": 1530
    },
    {
      "epoch": 2.1,
      "learning_rate": 4.389948531637905e-05,
      "loss": 1.6089,
      "step": 1540
    },
    {
      "epoch": 2.11,
      "learning_rate": 4.3823796548592185e-05,
      "loss": 1.5848,
      "step": 1550
    },
    {
      "epoch": 2.13,
      "learning_rate": 4.374810778080533e-05,
      "loss": 1.5827,
      "step": 1560
    },
    {
      "epoch": 2.14,
      "learning_rate": 4.367241901301847e-05,
      "loss": 1.5834,
      "step": 1570
    },
    {
      "epoch": 2.15,
      "learning_rate": 4.359673024523161e-05,
      "loss": 1.5795,
      "step": 1580
    },
    {
      "epoch": 2.17,
      "learning_rate": 4.352104147744475e-05,
      "loss": 1.5808,
      "step": 1590
    },
    {
      "epoch": 2.18,
      "learning_rate": 4.344535270965789e-05,
      "loss": 1.554,
      "step": 1600
    },
    {
      "epoch": 2.19,
      "learning_rate": 4.336966394187103e-05,
      "loss": 1.5745,
      "step": 1610
    },
    {
      "epoch": 2.21,
      "learning_rate": 4.3293975174084164e-05,
      "loss": 1.5763,
      "step": 1620
    },
    {
      "epoch": 2.22,
      "learning_rate": 4.321828640629731e-05,
      "loss": 1.577,
      "step": 1630
    },
    {
      "epoch": 2.23,
      "learning_rate": 4.3142597638510445e-05,
      "loss": 1.5459,
      "step": 1640
    },
    {
      "epoch": 2.25,
      "learning_rate": 4.3066908870723586e-05,
      "loss": 1.5638,
      "step": 1650
    },
    {
      "epoch": 2.26,
      "learning_rate": 4.2991220102936727e-05,
      "loss": 1.5855,
      "step": 1660
    },
    {
      "epoch": 2.28,
      "learning_rate": 4.291553133514987e-05,
      "loss": 1.5524,
      "step": 1670
    },
    {
      "epoch": 2.29,
      "learning_rate": 4.2839842567363e-05,
      "loss": 1.5847,
      "step": 1680
    },
    {
      "epoch": 2.3,
      "learning_rate": 4.276415379957615e-05,
      "loss": 1.5616,
      "step": 1690
    },
    {
      "epoch": 2.32,
      "learning_rate": 4.268846503178928e-05,
      "loss": 1.5656,
      "step": 1700
    },
    {
      "epoch": 2.33,
      "learning_rate": 4.261277626400242e-05,
      "loss": 1.5899,
      "step": 1710
    },
    {
      "epoch": 2.34,
      "learning_rate": 4.2537087496215564e-05,
      "loss": 1.57,
      "step": 1720
    },
    {
      "epoch": 2.36,
      "learning_rate": 4.2461398728428705e-05,
      "loss": 1.5747,
      "step": 1730
    },
    {
      "epoch": 2.37,
      "learning_rate": 4.2385709960641845e-05,
      "loss": 1.5566,
      "step": 1740
    },
    {
      "epoch": 2.38,
      "learning_rate": 4.231002119285498e-05,
      "loss": 1.5776,
      "step": 1750
    },
    {
      "epoch": 2.4,
      "learning_rate": 4.223433242506813e-05,
      "loss": 1.5886,
      "step": 1760
    },
    {
      "epoch": 2.41,
      "learning_rate": 4.215864365728126e-05,
      "loss": 1.5551,
      "step": 1770
    },
    {
      "epoch": 2.43,
      "learning_rate": 4.20829548894944e-05,
      "loss": 1.5521,
      "step": 1780
    },
    {
      "epoch": 2.44,
      "learning_rate": 4.200726612170754e-05,
      "loss": 1.56,
      "step": 1790
    },
    {
      "epoch": 2.45,
      "learning_rate": 4.193157735392068e-05,
      "loss": 1.6015,
      "step": 1800
    },
    {
      "epoch": 2.47,
      "learning_rate": 4.185588858613382e-05,
      "loss": 1.565,
      "step": 1810
    },
    {
      "epoch": 2.48,
      "learning_rate": 4.178019981834696e-05,
      "loss": 1.562,
      "step": 1820
    },
    {
      "epoch": 2.49,
      "learning_rate": 4.17045110505601e-05,
      "loss": 1.5519,
      "step": 1830
    },
    {
      "epoch": 2.51,
      "learning_rate": 4.162882228277324e-05,
      "loss": 1.5686,
      "step": 1840
    },
    {
      "epoch": 2.52,
      "learning_rate": 4.155313351498637e-05,
      "loss": 1.5754,
      "step": 1850
    },
    {
      "epoch": 2.53,
      "learning_rate": 4.147744474719952e-05,
      "loss": 1.5538,
      "step": 1860
    },
    {
      "epoch": 2.55,
      "learning_rate": 4.140175597941266e-05,
      "loss": 1.5892,
      "step": 1870
    },
    {
      "epoch": 2.56,
      "learning_rate": 4.1326067211625795e-05,
      "loss": 1.5632,
      "step": 1880
    },
    {
      "epoch": 2.57,
      "learning_rate": 4.1250378443838936e-05,
      "loss": 1.5655,
      "step": 1890
    },
    {
      "epoch": 2.59,
      "learning_rate": 4.1174689676052077e-05,
      "loss": 1.5464,
      "step": 1900
    },
    {
      "epoch": 2.6,
      "learning_rate": 4.109900090826522e-05,
      "loss": 1.5387,
      "step": 1910
    },
    {
      "epoch": 2.62,
      "learning_rate": 4.102331214047835e-05,
      "loss": 1.5949,
      "step": 1920
    },
    {
      "epoch": 2.63,
      "learning_rate": 4.09476233726915e-05,
      "loss": 1.5814,
      "step": 1930
    },
    {
      "epoch": 2.64,
      "learning_rate": 4.087193460490463e-05,
      "loss": 1.5473,
      "step": 1940
    },
    {
      "epoch": 2.66,
      "learning_rate": 4.079624583711777e-05,
      "loss": 1.5452,
      "step": 1950
    },
    {
      "epoch": 2.67,
      "learning_rate": 4.0720557069330914e-05,
      "loss": 1.5575,
      "step": 1960
    },
    {
      "epoch": 2.68,
      "learning_rate": 4.0644868301544055e-05,
      "loss": 1.5853,
      "step": 1970
    },
    {
      "epoch": 2.7,
      "learning_rate": 4.056917953375719e-05,
      "loss": 1.5795,
      "step": 1980
    },
    {
      "epoch": 2.71,
      "learning_rate": 4.049349076597033e-05,
      "loss": 1.5966,
      "step": 1990
    },
    {
      "epoch": 2.72,
      "learning_rate": 4.041780199818347e-05,
      "loss": 1.5672,
      "step": 2000
    },
    {
      "epoch": 2.74,
      "learning_rate": 4.034211323039661e-05,
      "loss": 1.5595,
      "step": 2010
    },
    {
      "epoch": 2.75,
      "learning_rate": 4.026642446260975e-05,
      "loss": 1.5377,
      "step": 2020
    },
    {
      "epoch": 2.77,
      "learning_rate": 4.019073569482289e-05,
      "loss": 1.557,
      "step": 2030
    },
    {
      "epoch": 2.78,
      "learning_rate": 4.011504692703603e-05,
      "loss": 1.5431,
      "step": 2040
    },
    {
      "epoch": 2.79,
      "learning_rate": 4.003935815924917e-05,
      "loss": 1.5525,
      "step": 2050
    },
    {
      "epoch": 2.81,
      "learning_rate": 3.996366939146231e-05,
      "loss": 1.5554,
      "step": 2060
    },
    {
      "epoch": 2.82,
      "learning_rate": 3.988798062367545e-05,
      "loss": 1.5562,
      "step": 2070
    },
    {
      "epoch": 2.83,
      "learning_rate": 3.981229185588859e-05,
      "loss": 1.5806,
      "step": 2080
    },
    {
      "epoch": 2.85,
      "learning_rate": 3.973660308810172e-05,
      "loss": 1.5614,
      "step": 2090
    },
    {
      "epoch": 2.86,
      "learning_rate": 3.966091432031487e-05,
      "loss": 1.5484,
      "step": 2100
    },
    {
      "epoch": 2.87,
      "learning_rate": 3.9585225552528004e-05,
      "loss": 1.5942,
      "step": 2110
    },
    {
      "epoch": 2.89,
      "learning_rate": 3.9509536784741145e-05,
      "loss": 1.5749,
      "step": 2120
    },
    {
      "epoch": 2.9,
      "learning_rate": 3.9433848016954286e-05,
      "loss": 1.57,
      "step": 2130
    },
    {
      "epoch": 2.92,
      "learning_rate": 3.9358159249167426e-05,
      "loss": 1.5517,
      "step": 2140
    },
    {
      "epoch": 2.93,
      "learning_rate": 3.928247048138057e-05,
      "loss": 1.5626,
      "step": 2150
    },
    {
      "epoch": 2.94,
      "learning_rate": 3.92067817135937e-05,
      "loss": 1.5631,
      "step": 2160
    },
    {
      "epoch": 2.96,
      "learning_rate": 3.913109294580685e-05,
      "loss": 1.5568,
      "step": 2170
    },
    {
      "epoch": 2.97,
      "learning_rate": 3.905540417801998e-05,
      "loss": 1.545,
      "step": 2180
    },
    {
      "epoch": 2.98,
      "learning_rate": 3.897971541023312e-05,
      "loss": 1.549,
      "step": 2190
    },
    {
      "epoch": 3.0,
      "learning_rate": 3.8904026642446264e-05,
      "loss": 1.5699,
      "step": 2200
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.3566624095359728,
      "eval_loss": 1.488677978515625,
      "eval_runtime": 350.239,
      "eval_samples_per_second": 67.068,
      "eval_steps_per_second": 2.099,
      "step": 2202
    },
    {
      "epoch": 3.01,
      "learning_rate": 3.8828337874659405e-05,
      "loss": 1.5528,
      "step": 2210
    },
    {
      "epoch": 3.02,
      "learning_rate": 3.875264910687254e-05,
      "loss": 1.5289,
      "step": 2220
    },
    {
      "epoch": 3.04,
      "learning_rate": 3.8676960339085686e-05,
      "loss": 1.525,
      "step": 2230
    },
    {
      "epoch": 3.05,
      "learning_rate": 3.860127157129882e-05,
      "loss": 1.5635,
      "step": 2240
    },
    {
      "epoch": 3.07,
      "learning_rate": 3.852558280351196e-05,
      "loss": 1.5532,
      "step": 2250
    },
    {
      "epoch": 3.08,
      "learning_rate": 3.84498940357251e-05,
      "loss": 1.5594,
      "step": 2260
    },
    {
      "epoch": 3.09,
      "learning_rate": 3.837420526793824e-05,
      "loss": 1.539,
      "step": 2270
    },
    {
      "epoch": 3.11,
      "learning_rate": 3.8298516500151376e-05,
      "loss": 1.5503,
      "step": 2280
    },
    {
      "epoch": 3.12,
      "learning_rate": 3.822282773236452e-05,
      "loss": 1.5445,
      "step": 2290
    },
    {
      "epoch": 3.13,
      "learning_rate": 3.8147138964577664e-05,
      "loss": 1.5538,
      "step": 2300
    },
    {
      "epoch": 3.15,
      "learning_rate": 3.80714501967908e-05,
      "loss": 1.5698,
      "step": 2310
    },
    {
      "epoch": 3.16,
      "learning_rate": 3.799576142900394e-05,
      "loss": 1.5584,
      "step": 2320
    },
    {
      "epoch": 3.17,
      "learning_rate": 3.792007266121708e-05,
      "loss": 1.5339,
      "step": 2330
    },
    {
      "epoch": 3.19,
      "learning_rate": 3.784438389343022e-05,
      "loss": 1.4933,
      "step": 2340
    },
    {
      "epoch": 3.2,
      "learning_rate": 3.7768695125643354e-05,
      "loss": 1.5552,
      "step": 2350
    },
    {
      "epoch": 3.22,
      "learning_rate": 3.7693006357856495e-05,
      "loss": 1.5268,
      "step": 2360
    },
    {
      "epoch": 3.23,
      "learning_rate": 3.7617317590069636e-05,
      "loss": 1.5464,
      "step": 2370
    },
    {
      "epoch": 3.24,
      "learning_rate": 3.7541628822282776e-05,
      "loss": 1.5324,
      "step": 2380
    },
    {
      "epoch": 3.26,
      "learning_rate": 3.746594005449591e-05,
      "loss": 1.5539,
      "step": 2390
    },
    {
      "epoch": 3.27,
      "learning_rate": 3.739025128670906e-05,
      "loss": 1.5386,
      "step": 2400
    },
    {
      "epoch": 3.28,
      "learning_rate": 3.731456251892219e-05,
      "loss": 1.5438,
      "step": 2410
    },
    {
      "epoch": 3.3,
      "learning_rate": 3.723887375113533e-05,
      "loss": 1.5428,
      "step": 2420
    },
    {
      "epoch": 3.31,
      "learning_rate": 3.716318498334847e-05,
      "loss": 1.5396,
      "step": 2430
    },
    {
      "epoch": 3.32,
      "learning_rate": 3.7087496215561614e-05,
      "loss": 1.5534,
      "step": 2440
    },
    {
      "epoch": 3.34,
      "learning_rate": 3.7011807447774755e-05,
      "loss": 1.5636,
      "step": 2450
    },
    {
      "epoch": 3.35,
      "learning_rate": 3.693611867998789e-05,
      "loss": 1.5493,
      "step": 2460
    },
    {
      "epoch": 3.37,
      "learning_rate": 3.6860429912201036e-05,
      "loss": 1.5372,
      "step": 2470
    },
    {
      "epoch": 3.38,
      "learning_rate": 3.678474114441417e-05,
      "loss": 1.543,
      "step": 2480
    },
    {
      "epoch": 3.39,
      "learning_rate": 3.670905237662731e-05,
      "loss": 1.5167,
      "step": 2490
    },
    {
      "epoch": 3.41,
      "learning_rate": 3.663336360884045e-05,
      "loss": 1.5618,
      "step": 2500
    },
    {
      "epoch": 3.42,
      "learning_rate": 3.655767484105359e-05,
      "loss": 1.5356,
      "step": 2510
    },
    {
      "epoch": 3.43,
      "learning_rate": 3.6481986073266726e-05,
      "loss": 1.534,
      "step": 2520
    },
    {
      "epoch": 3.45,
      "learning_rate": 3.640629730547987e-05,
      "loss": 1.5725,
      "step": 2530
    },
    {
      "epoch": 3.46,
      "learning_rate": 3.633060853769301e-05,
      "loss": 1.5523,
      "step": 2540
    },
    {
      "epoch": 3.47,
      "learning_rate": 3.625491976990615e-05,
      "loss": 1.5496,
      "step": 2550
    },
    {
      "epoch": 3.49,
      "learning_rate": 3.617923100211928e-05,
      "loss": 1.5423,
      "step": 2560
    },
    {
      "epoch": 3.5,
      "learning_rate": 3.610354223433243e-05,
      "loss": 1.5477,
      "step": 2570
    },
    {
      "epoch": 3.51,
      "learning_rate": 3.6027853466545564e-05,
      "loss": 1.5236,
      "step": 2580
    },
    {
      "epoch": 3.53,
      "learning_rate": 3.5952164698758704e-05,
      "loss": 1.5679,
      "step": 2590
    },
    {
      "epoch": 3.54,
      "learning_rate": 3.5876475930971845e-05,
      "loss": 1.5389,
      "step": 2600
    },
    {
      "epoch": 3.56,
      "learning_rate": 3.5800787163184986e-05,
      "loss": 1.5196,
      "step": 2610
    },
    {
      "epoch": 3.57,
      "learning_rate": 3.5725098395398126e-05,
      "loss": 1.5314,
      "step": 2620
    },
    {
      "epoch": 3.58,
      "learning_rate": 3.564940962761126e-05,
      "loss": 1.5327,
      "step": 2630
    },
    {
      "epoch": 3.6,
      "learning_rate": 3.557372085982441e-05,
      "loss": 1.5195,
      "step": 2640
    },
    {
      "epoch": 3.61,
      "learning_rate": 3.549803209203754e-05,
      "loss": 1.5231,
      "step": 2650
    },
    {
      "epoch": 3.62,
      "learning_rate": 3.542234332425068e-05,
      "loss": 1.5688,
      "step": 2660
    },
    {
      "epoch": 3.64,
      "learning_rate": 3.534665455646382e-05,
      "loss": 1.5211,
      "step": 2670
    },
    {
      "epoch": 3.65,
      "learning_rate": 3.5270965788676964e-05,
      "loss": 1.5293,
      "step": 2680
    },
    {
      "epoch": 3.66,
      "learning_rate": 3.51952770208901e-05,
      "loss": 1.5548,
      "step": 2690
    },
    {
      "epoch": 3.68,
      "learning_rate": 3.511958825310324e-05,
      "loss": 1.5444,
      "step": 2700
    },
    {
      "epoch": 3.69,
      "learning_rate": 3.504389948531638e-05,
      "loss": 1.5344,
      "step": 2710
    },
    {
      "epoch": 3.71,
      "learning_rate": 3.496821071752952e-05,
      "loss": 1.5309,
      "step": 2720
    },
    {
      "epoch": 3.72,
      "learning_rate": 3.489252194974266e-05,
      "loss": 1.5219,
      "step": 2730
    },
    {
      "epoch": 3.73,
      "learning_rate": 3.48168331819558e-05,
      "loss": 1.5609,
      "step": 2740
    },
    {
      "epoch": 3.75,
      "learning_rate": 3.474114441416894e-05,
      "loss": 1.5454,
      "step": 2750
    },
    {
      "epoch": 3.76,
      "learning_rate": 3.4665455646382076e-05,
      "loss": 1.558,
      "step": 2760
    },
    {
      "epoch": 3.77,
      "learning_rate": 3.458976687859522e-05,
      "loss": 1.5301,
      "step": 2770
    },
    {
      "epoch": 3.79,
      "learning_rate": 3.451407811080836e-05,
      "loss": 1.53,
      "step": 2780
    },
    {
      "epoch": 3.8,
      "learning_rate": 3.44383893430215e-05,
      "loss": 1.5468,
      "step": 2790
    },
    {
      "epoch": 3.81,
      "learning_rate": 3.436270057523463e-05,
      "loss": 1.561,
      "step": 2800
    },
    {
      "epoch": 3.83,
      "learning_rate": 3.428701180744778e-05,
      "loss": 1.5279,
      "step": 2810
    },
    {
      "epoch": 3.84,
      "learning_rate": 3.4211323039660914e-05,
      "loss": 1.5371,
      "step": 2820
    },
    {
      "epoch": 3.86,
      "learning_rate": 3.4135634271874054e-05,
      "loss": 1.5369,
      "step": 2830
    },
    {
      "epoch": 3.87,
      "learning_rate": 3.4059945504087195e-05,
      "loss": 1.5553,
      "step": 2840
    },
    {
      "epoch": 3.88,
      "learning_rate": 3.3984256736300336e-05,
      "loss": 1.529,
      "step": 2850
    },
    {
      "epoch": 3.9,
      "learning_rate": 3.390856796851347e-05,
      "loss": 1.515,
      "step": 2860
    },
    {
      "epoch": 3.91,
      "learning_rate": 3.383287920072662e-05,
      "loss": 1.5552,
      "step": 2870
    },
    {
      "epoch": 3.92,
      "learning_rate": 3.375719043293976e-05,
      "loss": 1.5491,
      "step": 2880
    },
    {
      "epoch": 3.94,
      "learning_rate": 3.368150166515289e-05,
      "loss": 1.5199,
      "step": 2890
    },
    {
      "epoch": 3.95,
      "learning_rate": 3.360581289736603e-05,
      "loss": 1.5396,
      "step": 2900
    },
    {
      "epoch": 3.96,
      "learning_rate": 3.353012412957917e-05,
      "loss": 1.5377,
      "step": 2910
    },
    {
      "epoch": 3.98,
      "learning_rate": 3.3454435361792314e-05,
      "loss": 1.5411,
      "step": 2920
    },
    {
      "epoch": 3.99,
      "learning_rate": 3.337874659400545e-05,
      "loss": 1.5361,
      "step": 2930
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.3685823754789272,
      "eval_loss": 1.4658845663070679,
      "eval_runtime": 348.8943,
      "eval_samples_per_second": 67.327,
      "eval_steps_per_second": 2.107,
      "step": 2936
    },
    {
      "epoch": 4.01,
      "learning_rate": 3.3303057826218595e-05,
      "loss": 1.5292,
      "step": 2940
    },
    {
      "epoch": 4.02,
      "learning_rate": 3.322736905843173e-05,
      "loss": 1.5278,
      "step": 2950
    },
    {
      "epoch": 4.03,
      "learning_rate": 3.315168029064487e-05,
      "loss": 1.5346,
      "step": 2960
    },
    {
      "epoch": 4.05,
      "learning_rate": 3.307599152285801e-05,
      "loss": 1.5354,
      "step": 2970
    },
    {
      "epoch": 4.06,
      "learning_rate": 3.300030275507115e-05,
      "loss": 1.5127,
      "step": 2980
    },
    {
      "epoch": 4.07,
      "learning_rate": 3.2924613987284285e-05,
      "loss": 1.5187,
      "step": 2990
    },
    {
      "epoch": 4.09,
      "learning_rate": 3.2848925219497426e-05,
      "loss": 1.5354,
      "step": 3000
    },
    {
      "epoch": 4.1,
      "learning_rate": 3.277323645171057e-05,
      "loss": 1.5551,
      "step": 3010
    },
    {
      "epoch": 4.11,
      "learning_rate": 3.269754768392371e-05,
      "loss": 1.5026,
      "step": 3020
    },
    {
      "epoch": 4.13,
      "learning_rate": 3.262185891613685e-05,
      "loss": 1.5196,
      "step": 3030
    },
    {
      "epoch": 4.14,
      "learning_rate": 3.254617014834999e-05,
      "loss": 1.5365,
      "step": 3040
    },
    {
      "epoch": 4.16,
      "learning_rate": 3.247048138056313e-05,
      "loss": 1.4862,
      "step": 3050
    },
    {
      "epoch": 4.17,
      "learning_rate": 3.2394792612776263e-05,
      "loss": 1.5044,
      "step": 3060
    },
    {
      "epoch": 4.18,
      "learning_rate": 3.2319103844989404e-05,
      "loss": 1.5064,
      "step": 3070
    },
    {
      "epoch": 4.2,
      "learning_rate": 3.2243415077202545e-05,
      "loss": 1.5176,
      "step": 3080
    },
    {
      "epoch": 4.21,
      "learning_rate": 3.2167726309415686e-05,
      "loss": 1.5206,
      "step": 3090
    },
    {
      "epoch": 4.22,
      "learning_rate": 3.209203754162882e-05,
      "loss": 1.5195,
      "step": 3100
    },
    {
      "epoch": 4.24,
      "learning_rate": 3.201634877384197e-05,
      "loss": 1.4858,
      "step": 3110
    },
    {
      "epoch": 4.25,
      "learning_rate": 3.19406600060551e-05,
      "loss": 1.4872,
      "step": 3120
    },
    {
      "epoch": 4.26,
      "learning_rate": 3.186497123826824e-05,
      "loss": 1.5452,
      "step": 3130
    },
    {
      "epoch": 4.28,
      "learning_rate": 3.178928247048138e-05,
      "loss": 1.5011,
      "step": 3140
    },
    {
      "epoch": 4.29,
      "learning_rate": 3.171359370269452e-05,
      "loss": 1.538,
      "step": 3150
    },
    {
      "epoch": 4.31,
      "learning_rate": 3.1637904934907664e-05,
      "loss": 1.5127,
      "step": 3160
    },
    {
      "epoch": 4.32,
      "learning_rate": 3.15622161671208e-05,
      "loss": 1.5213,
      "step": 3170
    },
    {
      "epoch": 4.33,
      "learning_rate": 3.1486527399333945e-05,
      "loss": 1.5411,
      "step": 3180
    },
    {
      "epoch": 4.35,
      "learning_rate": 3.141083863154708e-05,
      "loss": 1.4838,
      "step": 3190
    },
    {
      "epoch": 4.36,
      "learning_rate": 3.133514986376022e-05,
      "loss": 1.5379,
      "step": 3200
    },
    {
      "epoch": 4.37,
      "learning_rate": 3.125946109597336e-05,
      "loss": 1.5106,
      "step": 3210
    },
    {
      "epoch": 4.39,
      "learning_rate": 3.11837723281865e-05,
      "loss": 1.5474,
      "step": 3220
    },
    {
      "epoch": 4.4,
      "learning_rate": 3.1108083560399635e-05,
      "loss": 1.5255,
      "step": 3230
    },
    {
      "epoch": 4.41,
      "learning_rate": 3.1032394792612776e-05,
      "loss": 1.5114,
      "step": 3240
    },
    {
      "epoch": 4.43,
      "learning_rate": 3.095670602482592e-05,
      "loss": 1.5305,
      "step": 3250
    },
    {
      "epoch": 4.44,
      "learning_rate": 3.088101725703906e-05,
      "loss": 1.5169,
      "step": 3260
    },
    {
      "epoch": 4.46,
      "learning_rate": 3.080532848925219e-05,
      "loss": 1.5111,
      "step": 3270
    },
    {
      "epoch": 4.47,
      "learning_rate": 3.072963972146534e-05,
      "loss": 1.5406,
      "step": 3280
    },
    {
      "epoch": 4.48,
      "learning_rate": 3.065395095367847e-05,
      "loss": 1.54,
      "step": 3290
    },
    {
      "epoch": 4.5,
      "learning_rate": 3.0578262185891613e-05,
      "loss": 1.557,
      "step": 3300
    },
    {
      "epoch": 4.51,
      "learning_rate": 3.0502573418104758e-05,
      "loss": 1.5574,
      "step": 3310
    },
    {
      "epoch": 4.52,
      "learning_rate": 3.0426884650317895e-05,
      "loss": 1.4921,
      "step": 3320
    },
    {
      "epoch": 4.54,
      "learning_rate": 3.0351195882531036e-05,
      "loss": 1.5488,
      "step": 3330
    },
    {
      "epoch": 4.55,
      "learning_rate": 3.0275507114744173e-05,
      "loss": 1.5194,
      "step": 3340
    },
    {
      "epoch": 4.56,
      "learning_rate": 3.0199818346957314e-05,
      "loss": 1.5197,
      "step": 3350
    },
    {
      "epoch": 4.58,
      "learning_rate": 3.012412957917045e-05,
      "loss": 1.541,
      "step": 3360
    },
    {
      "epoch": 4.59,
      "learning_rate": 3.0048440811383595e-05,
      "loss": 1.5276,
      "step": 3370
    },
    {
      "epoch": 4.6,
      "learning_rate": 2.997275204359673e-05,
      "loss": 1.5185,
      "step": 3380
    },
    {
      "epoch": 4.62,
      "learning_rate": 2.9897063275809873e-05,
      "loss": 1.5042,
      "step": 3390
    },
    {
      "epoch": 4.63,
      "learning_rate": 2.982137450802301e-05,
      "loss": 1.498,
      "step": 3400
    },
    {
      "epoch": 4.65,
      "learning_rate": 2.974568574023615e-05,
      "loss": 1.5081,
      "step": 3410
    },
    {
      "epoch": 4.66,
      "learning_rate": 2.966999697244929e-05,
      "loss": 1.4646,
      "step": 3420
    },
    {
      "epoch": 4.67,
      "learning_rate": 2.959430820466243e-05,
      "loss": 1.5688,
      "step": 3430
    },
    {
      "epoch": 4.69,
      "learning_rate": 2.9518619436875566e-05,
      "loss": 1.5148,
      "step": 3440
    },
    {
      "epoch": 4.7,
      "learning_rate": 2.9442930669088707e-05,
      "loss": 1.4905,
      "step": 3450
    },
    {
      "epoch": 4.71,
      "learning_rate": 2.936724190130185e-05,
      "loss": 1.5279,
      "step": 3460
    },
    {
      "epoch": 4.73,
      "learning_rate": 2.929155313351499e-05,
      "loss": 1.5452,
      "step": 3470
    },
    {
      "epoch": 4.74,
      "learning_rate": 2.921586436572813e-05,
      "loss": 1.5129,
      "step": 3480
    },
    {
      "epoch": 4.75,
      "learning_rate": 2.9140175597941267e-05,
      "loss": 1.5482,
      "step": 3490
    },
    {
      "epoch": 4.77,
      "learning_rate": 2.9064486830154407e-05,
      "loss": 1.5003,
      "step": 3500
    },
    {
      "epoch": 4.78,
      "learning_rate": 2.8988798062367545e-05,
      "loss": 1.5157,
      "step": 3510
    },
    {
      "epoch": 4.8,
      "learning_rate": 2.8913109294580685e-05,
      "loss": 1.5311,
      "step": 3520
    },
    {
      "epoch": 4.81,
      "learning_rate": 2.8837420526793823e-05,
      "loss": 1.4755,
      "step": 3530
    },
    {
      "epoch": 4.82,
      "learning_rate": 2.8761731759006967e-05,
      "loss": 1.5102,
      "step": 3540
    },
    {
      "epoch": 4.84,
      "learning_rate": 2.86860429912201e-05,
      "loss": 1.5363,
      "step": 3550
    },
    {
      "epoch": 4.85,
      "learning_rate": 2.8610354223433245e-05,
      "loss": 1.5077,
      "step": 3560
    },
    {
      "epoch": 4.86,
      "learning_rate": 2.8534665455646382e-05,
      "loss": 1.5045,
      "step": 3570
    },
    {
      "epoch": 4.88,
      "learning_rate": 2.8458976687859523e-05,
      "loss": 1.5078,
      "step": 3580
    },
    {
      "epoch": 4.89,
      "learning_rate": 2.8383287920072667e-05,
      "loss": 1.5056,
      "step": 3590
    },
    {
      "epoch": 4.9,
      "learning_rate": 2.83075991522858e-05,
      "loss": 1.5183,
      "step": 3600
    },
    {
      "epoch": 4.92,
      "learning_rate": 2.8231910384498945e-05,
      "loss": 1.5242,
      "step": 3610
    },
    {
      "epoch": 4.93,
      "learning_rate": 2.8156221616712082e-05,
      "loss": 1.5083,
      "step": 3620
    },
    {
      "epoch": 4.95,
      "learning_rate": 2.8080532848925223e-05,
      "loss": 1.5076,
      "step": 3630
    },
    {
      "epoch": 4.96,
      "learning_rate": 2.800484408113836e-05,
      "loss": 1.5283,
      "step": 3640
    },
    {
      "epoch": 4.97,
      "learning_rate": 2.79291553133515e-05,
      "loss": 1.5135,
      "step": 3650
    },
    {
      "epoch": 4.99,
      "learning_rate": 2.785346654556464e-05,
      "loss": 1.5106,
      "step": 3660
    },
    {
      "epoch": 5.0,
      "learning_rate": 2.777777777777778e-05,
      "loss": 1.525,
      "step": 3670
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.392039165602384,
      "eval_loss": 1.416853427886963,
      "eval_runtime": 331.3993,
      "eval_samples_per_second": 70.881,
      "eval_steps_per_second": 2.218,
      "step": 3670
    },
    {
      "epoch": 5.01,
      "learning_rate": 2.7702089009990916e-05,
      "loss": 1.5369,
      "step": 3680
    },
    {
      "epoch": 5.03,
      "learning_rate": 2.762640024220406e-05,
      "loss": 1.5159,
      "step": 3690
    },
    {
      "epoch": 5.04,
      "learning_rate": 2.7550711474417194e-05,
      "loss": 1.5031,
      "step": 3700
    },
    {
      "epoch": 5.05,
      "learning_rate": 2.747502270663034e-05,
      "loss": 1.5172,
      "step": 3710
    },
    {
      "epoch": 5.07,
      "learning_rate": 2.7399333938843476e-05,
      "loss": 1.4728,
      "step": 3720
    },
    {
      "epoch": 5.08,
      "learning_rate": 2.7323645171056617e-05,
      "loss": 1.4821,
      "step": 3730
    },
    {
      "epoch": 5.1,
      "learning_rate": 2.7247956403269757e-05,
      "loss": 1.5144,
      "step": 3740
    },
    {
      "epoch": 5.11,
      "learning_rate": 2.7172267635482895e-05,
      "loss": 1.5246,
      "step": 3750
    },
    {
      "epoch": 5.12,
      "learning_rate": 2.709657886769604e-05,
      "loss": 1.5061,
      "step": 3760
    },
    {
      "epoch": 5.14,
      "learning_rate": 2.7020890099909173e-05,
      "loss": 1.4675,
      "step": 3770
    },
    {
      "epoch": 5.15,
      "learning_rate": 2.6945201332122317e-05,
      "loss": 1.4751,
      "step": 3780
    },
    {
      "epoch": 5.16,
      "learning_rate": 2.6869512564335454e-05,
      "loss": 1.5239,
      "step": 3790
    },
    {
      "epoch": 5.18,
      "learning_rate": 2.6793823796548595e-05,
      "loss": 1.5211,
      "step": 3800
    },
    {
      "epoch": 5.19,
      "learning_rate": 2.6718135028761732e-05,
      "loss": 1.5182,
      "step": 3810
    },
    {
      "epoch": 5.2,
      "learning_rate": 2.6642446260974873e-05,
      "loss": 1.4852,
      "step": 3820
    },
    {
      "epoch": 5.22,
      "learning_rate": 2.656675749318801e-05,
      "loss": 1.4999,
      "step": 3830
    },
    {
      "epoch": 5.23,
      "learning_rate": 2.649106872540115e-05,
      "loss": 1.5044,
      "step": 3840
    },
    {
      "epoch": 5.25,
      "learning_rate": 2.6415379957614288e-05,
      "loss": 1.4974,
      "step": 3850
    },
    {
      "epoch": 5.26,
      "learning_rate": 2.6339691189827432e-05,
      "loss": 1.5175,
      "step": 3860
    },
    {
      "epoch": 5.27,
      "learning_rate": 2.6264002422040566e-05,
      "loss": 1.5194,
      "step": 3870
    },
    {
      "epoch": 5.29,
      "learning_rate": 2.618831365425371e-05,
      "loss": 1.4728,
      "step": 3880
    },
    {
      "epoch": 5.3,
      "learning_rate": 2.611262488646685e-05,
      "loss": 1.481,
      "step": 3890
    },
    {
      "epoch": 5.31,
      "learning_rate": 2.603693611867999e-05,
      "loss": 1.5082,
      "step": 3900
    },
    {
      "epoch": 5.33,
      "learning_rate": 2.5961247350893132e-05,
      "loss": 1.5179,
      "step": 3910
    },
    {
      "epoch": 5.34,
      "learning_rate": 2.5885558583106266e-05,
      "loss": 1.4895,
      "step": 3920
    },
    {
      "epoch": 5.35,
      "learning_rate": 2.580986981531941e-05,
      "loss": 1.4849,
      "step": 3930
    },
    {
      "epoch": 5.37,
      "learning_rate": 2.5734181047532548e-05,
      "loss": 1.517,
      "step": 3940
    },
    {
      "epoch": 5.38,
      "learning_rate": 2.565849227974569e-05,
      "loss": 1.501,
      "step": 3950
    },
    {
      "epoch": 5.4,
      "learning_rate": 2.5582803511958826e-05,
      "loss": 1.5018,
      "step": 3960
    },
    {
      "epoch": 5.41,
      "learning_rate": 2.5507114744171967e-05,
      "loss": 1.4859,
      "step": 3970
    },
    {
      "epoch": 5.42,
      "learning_rate": 2.5431425976385104e-05,
      "loss": 1.482,
      "step": 3980
    },
    {
      "epoch": 5.44,
      "learning_rate": 2.5355737208598245e-05,
      "loss": 1.54,
      "step": 3990
    },
    {
      "epoch": 5.45,
      "learning_rate": 2.5280048440811382e-05,
      "loss": 1.5288,
      "step": 4000
    },
    {
      "epoch": 5.46,
      "learning_rate": 2.5204359673024526e-05,
      "loss": 1.5436,
      "step": 4010
    },
    {
      "epoch": 5.48,
      "learning_rate": 2.512867090523766e-05,
      "loss": 1.5277,
      "step": 4020
    },
    {
      "epoch": 5.49,
      "learning_rate": 2.5052982137450804e-05,
      "loss": 1.4873,
      "step": 4030
    },
    {
      "epoch": 5.5,
      "learning_rate": 2.497729336966394e-05,
      "loss": 1.506,
      "step": 4040
    },
    {
      "epoch": 5.52,
      "learning_rate": 2.4901604601877082e-05,
      "loss": 1.5304,
      "step": 4050
    },
    {
      "epoch": 5.53,
      "learning_rate": 2.4825915834090223e-05,
      "loss": 1.5295,
      "step": 4060
    },
    {
      "epoch": 5.54,
      "learning_rate": 2.475022706630336e-05,
      "loss": 1.4895,
      "step": 4070
    },
    {
      "epoch": 5.56,
      "learning_rate": 2.46745382985165e-05,
      "loss": 1.5188,
      "step": 4080
    },
    {
      "epoch": 5.57,
      "learning_rate": 2.4598849530729638e-05,
      "loss": 1.4842,
      "step": 4090
    },
    {
      "epoch": 5.59,
      "learning_rate": 2.4523160762942782e-05,
      "loss": 1.4947,
      "step": 4100
    },
    {
      "epoch": 5.6,
      "learning_rate": 2.444747199515592e-05,
      "loss": 1.4988,
      "step": 4110
    },
    {
      "epoch": 5.61,
      "learning_rate": 2.437178322736906e-05,
      "loss": 1.4615,
      "step": 4120
    },
    {
      "epoch": 5.63,
      "learning_rate": 2.42960944595822e-05,
      "loss": 1.4708,
      "step": 4130
    },
    {
      "epoch": 5.64,
      "learning_rate": 2.422040569179534e-05,
      "loss": 1.4658,
      "step": 4140
    },
    {
      "epoch": 5.65,
      "learning_rate": 2.414471692400848e-05,
      "loss": 1.475,
      "step": 4150
    },
    {
      "epoch": 5.67,
      "learning_rate": 2.4069028156221616e-05,
      "loss": 1.5239,
      "step": 4160
    },
    {
      "epoch": 5.68,
      "learning_rate": 2.3993339388434757e-05,
      "loss": 1.4691,
      "step": 4170
    },
    {
      "epoch": 5.69,
      "learning_rate": 2.3917650620647898e-05,
      "loss": 1.47,
      "step": 4180
    },
    {
      "epoch": 5.71,
      "learning_rate": 2.3841961852861035e-05,
      "loss": 1.4779,
      "step": 4190
    },
    {
      "epoch": 5.72,
      "learning_rate": 2.3766273085074176e-05,
      "loss": 1.5093,
      "step": 4200
    },
    {
      "epoch": 5.74,
      "learning_rate": 2.3690584317287317e-05,
      "loss": 1.5309,
      "step": 4210
    },
    {
      "epoch": 5.75,
      "learning_rate": 2.3614895549500454e-05,
      "loss": 1.4579,
      "step": 4220
    },
    {
      "epoch": 5.76,
      "learning_rate": 2.3539206781713595e-05,
      "loss": 1.4935,
      "step": 4230
    },
    {
      "epoch": 5.78,
      "learning_rate": 2.3463518013926735e-05,
      "loss": 1.5014,
      "step": 4240
    },
    {
      "epoch": 5.79,
      "learning_rate": 2.3387829246139876e-05,
      "loss": 1.5022,
      "step": 4250
    },
    {
      "epoch": 5.8,
      "learning_rate": 2.3312140478353013e-05,
      "loss": 1.4915,
      "step": 4260
    },
    {
      "epoch": 5.82,
      "learning_rate": 2.3236451710566154e-05,
      "loss": 1.5124,
      "step": 4270
    },
    {
      "epoch": 5.83,
      "learning_rate": 2.3160762942779295e-05,
      "loss": 1.5166,
      "step": 4280
    },
    {
      "epoch": 5.84,
      "learning_rate": 2.3085074174992432e-05,
      "loss": 1.499,
      "step": 4290
    },
    {
      "epoch": 5.86,
      "learning_rate": 2.3009385407205573e-05,
      "loss": 1.5097,
      "step": 4300
    },
    {
      "epoch": 5.87,
      "learning_rate": 2.293369663941871e-05,
      "loss": 1.4912,
      "step": 4310
    },
    {
      "epoch": 5.89,
      "learning_rate": 2.285800787163185e-05,
      "loss": 1.4932,
      "step": 4320
    },
    {
      "epoch": 5.9,
      "learning_rate": 2.278231910384499e-05,
      "loss": 1.4677,
      "step": 4330
    },
    {
      "epoch": 5.91,
      "learning_rate": 2.270663033605813e-05,
      "loss": 1.5223,
      "step": 4340
    },
    {
      "epoch": 5.93,
      "learning_rate": 2.263094156827127e-05,
      "loss": 1.496,
      "step": 4350
    },
    {
      "epoch": 5.94,
      "learning_rate": 2.2555252800484407e-05,
      "loss": 1.5019,
      "step": 4360
    },
    {
      "epoch": 5.95,
      "learning_rate": 2.2479564032697548e-05,
      "loss": 1.5447,
      "step": 4370
    },
    {
      "epoch": 5.97,
      "learning_rate": 2.2403875264910688e-05,
      "loss": 1.4918,
      "step": 4380
    },
    {
      "epoch": 5.98,
      "learning_rate": 2.232818649712383e-05,
      "loss": 1.5418,
      "step": 4390
    },
    {
      "epoch": 5.99,
      "learning_rate": 2.225249772933697e-05,
      "loss": 1.4744,
      "step": 4400
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.39565772669220944,
      "eval_loss": 1.4028806686401367,
      "eval_runtime": 359.9472,
      "eval_samples_per_second": 65.26,
      "eval_steps_per_second": 2.042,
      "step": 4404
    },
    {
      "epoch": 6.01,
      "learning_rate": 2.2176808961550107e-05,
      "loss": 1.4784,
      "step": 4410
    },
    {
      "epoch": 6.02,
      "learning_rate": 2.2101120193763248e-05,
      "loss": 1.4877,
      "step": 4420
    },
    {
      "epoch": 6.04,
      "learning_rate": 2.2025431425976385e-05,
      "loss": 1.5222,
      "step": 4430
    },
    {
      "epoch": 6.05,
      "learning_rate": 2.1949742658189526e-05,
      "loss": 1.5034,
      "step": 4440
    },
    {
      "epoch": 6.06,
      "learning_rate": 2.1874053890402666e-05,
      "loss": 1.5257,
      "step": 4450
    },
    {
      "epoch": 6.08,
      "learning_rate": 2.1798365122615804e-05,
      "loss": 1.5307,
      "step": 4460
    },
    {
      "epoch": 6.09,
      "learning_rate": 2.1722676354828945e-05,
      "loss": 1.4606,
      "step": 4470
    },
    {
      "epoch": 6.1,
      "learning_rate": 2.1646987587042082e-05,
      "loss": 1.4458,
      "step": 4480
    },
    {
      "epoch": 6.12,
      "learning_rate": 2.1571298819255223e-05,
      "loss": 1.4899,
      "step": 4490
    },
    {
      "epoch": 6.13,
      "learning_rate": 2.1495610051468363e-05,
      "loss": 1.4713,
      "step": 4500
    },
    {
      "epoch": 6.14,
      "learning_rate": 2.14199212836815e-05,
      "loss": 1.4526,
      "step": 4510
    },
    {
      "epoch": 6.16,
      "learning_rate": 2.134423251589464e-05,
      "loss": 1.4902,
      "step": 4520
    },
    {
      "epoch": 6.17,
      "learning_rate": 2.1268543748107782e-05,
      "loss": 1.4749,
      "step": 4530
    },
    {
      "epoch": 6.19,
      "learning_rate": 2.1192854980320923e-05,
      "loss": 1.4826,
      "step": 4540
    },
    {
      "epoch": 6.2,
      "learning_rate": 2.1117166212534063e-05,
      "loss": 1.4575,
      "step": 4550
    },
    {
      "epoch": 6.21,
      "learning_rate": 2.10414774447472e-05,
      "loss": 1.4361,
      "step": 4560
    },
    {
      "epoch": 6.23,
      "learning_rate": 2.096578867696034e-05,
      "loss": 1.4993,
      "step": 4570
    },
    {
      "epoch": 6.24,
      "learning_rate": 2.089009990917348e-05,
      "loss": 1.5065,
      "step": 4580
    },
    {
      "epoch": 6.25,
      "learning_rate": 2.081441114138662e-05,
      "loss": 1.4586,
      "step": 4590
    },
    {
      "epoch": 6.27,
      "learning_rate": 2.073872237359976e-05,
      "loss": 1.4808,
      "step": 4600
    },
    {
      "epoch": 6.28,
      "learning_rate": 2.0663033605812898e-05,
      "loss": 1.4414,
      "step": 4610
    },
    {
      "epoch": 6.29,
      "learning_rate": 2.0587344838026038e-05,
      "loss": 1.4614,
      "step": 4620
    },
    {
      "epoch": 6.31,
      "learning_rate": 2.0511656070239176e-05,
      "loss": 1.4883,
      "step": 4630
    },
    {
      "epoch": 6.32,
      "learning_rate": 2.0435967302452316e-05,
      "loss": 1.4935,
      "step": 4640
    },
    {
      "epoch": 6.34,
      "learning_rate": 2.0360278534665457e-05,
      "loss": 1.5119,
      "step": 4650
    },
    {
      "epoch": 6.35,
      "learning_rate": 2.0284589766878594e-05,
      "loss": 1.4969,
      "step": 4660
    },
    {
      "epoch": 6.36,
      "learning_rate": 2.0208900999091735e-05,
      "loss": 1.4248,
      "step": 4670
    },
    {
      "epoch": 6.38,
      "learning_rate": 2.0133212231304876e-05,
      "loss": 1.4878,
      "step": 4680
    },
    {
      "epoch": 6.39,
      "learning_rate": 2.0057523463518016e-05,
      "loss": 1.4804,
      "step": 4690
    },
    {
      "epoch": 6.4,
      "learning_rate": 1.9981834695731154e-05,
      "loss": 1.4599,
      "step": 4700
    },
    {
      "epoch": 6.42,
      "learning_rate": 1.9906145927944294e-05,
      "loss": 1.4887,
      "step": 4710
    },
    {
      "epoch": 6.43,
      "learning_rate": 1.9830457160157435e-05,
      "loss": 1.5,
      "step": 4720
    },
    {
      "epoch": 6.44,
      "learning_rate": 1.9754768392370573e-05,
      "loss": 1.4841,
      "step": 4730
    },
    {
      "epoch": 6.46,
      "learning_rate": 1.9679079624583713e-05,
      "loss": 1.4881,
      "step": 4740
    },
    {
      "epoch": 6.47,
      "learning_rate": 1.960339085679685e-05,
      "loss": 1.4421,
      "step": 4750
    },
    {
      "epoch": 6.49,
      "learning_rate": 1.952770208900999e-05,
      "loss": 1.5282,
      "step": 4760
    },
    {
      "epoch": 6.5,
      "learning_rate": 1.9452013321223132e-05,
      "loss": 1.4763,
      "step": 4770
    },
    {
      "epoch": 6.51,
      "learning_rate": 1.937632455343627e-05,
      "loss": 1.48,
      "step": 4780
    },
    {
      "epoch": 6.53,
      "learning_rate": 1.930063578564941e-05,
      "loss": 1.4858,
      "step": 4790
    },
    {
      "epoch": 6.54,
      "learning_rate": 1.922494701786255e-05,
      "loss": 1.4669,
      "step": 4800
    },
    {
      "epoch": 6.55,
      "learning_rate": 1.9149258250075688e-05,
      "loss": 1.4561,
      "step": 4810
    },
    {
      "epoch": 6.57,
      "learning_rate": 1.9073569482288832e-05,
      "loss": 1.4774,
      "step": 4820
    },
    {
      "epoch": 6.58,
      "learning_rate": 1.899788071450197e-05,
      "loss": 1.487,
      "step": 4830
    },
    {
      "epoch": 6.59,
      "learning_rate": 1.892219194671511e-05,
      "loss": 1.5055,
      "step": 4840
    },
    {
      "epoch": 6.61,
      "learning_rate": 1.8846503178928248e-05,
      "loss": 1.4942,
      "step": 4850
    },
    {
      "epoch": 6.62,
      "learning_rate": 1.8770814411141388e-05,
      "loss": 1.4704,
      "step": 4860
    },
    {
      "epoch": 6.63,
      "learning_rate": 1.869512564335453e-05,
      "loss": 1.4584,
      "step": 4870
    },
    {
      "epoch": 6.65,
      "learning_rate": 1.8619436875567666e-05,
      "loss": 1.5104,
      "step": 4880
    },
    {
      "epoch": 6.66,
      "learning_rate": 1.8543748107780807e-05,
      "loss": 1.5124,
      "step": 4890
    },
    {
      "epoch": 6.68,
      "learning_rate": 1.8468059339993944e-05,
      "loss": 1.5021,
      "step": 4900
    },
    {
      "epoch": 6.69,
      "learning_rate": 1.8392370572207085e-05,
      "loss": 1.4586,
      "step": 4910
    },
    {
      "epoch": 6.7,
      "learning_rate": 1.8316681804420226e-05,
      "loss": 1.5031,
      "step": 4920
    },
    {
      "epoch": 6.72,
      "learning_rate": 1.8240993036633363e-05,
      "loss": 1.4641,
      "step": 4930
    },
    {
      "epoch": 6.73,
      "learning_rate": 1.8165304268846504e-05,
      "loss": 1.4913,
      "step": 4940
    },
    {
      "epoch": 6.74,
      "learning_rate": 1.808961550105964e-05,
      "loss": 1.4752,
      "step": 4950
    },
    {
      "epoch": 6.76,
      "learning_rate": 1.8013926733272782e-05,
      "loss": 1.4776,
      "step": 4960
    },
    {
      "epoch": 6.77,
      "learning_rate": 1.7938237965485922e-05,
      "loss": 1.4875,
      "step": 4970
    },
    {
      "epoch": 6.78,
      "learning_rate": 1.7862549197699063e-05,
      "loss": 1.4853,
      "step": 4980
    },
    {
      "epoch": 6.8,
      "learning_rate": 1.7786860429912204e-05,
      "loss": 1.4577,
      "step": 4990
    },
    {
      "epoch": 6.81,
      "learning_rate": 1.771117166212534e-05,
      "loss": 1.4585,
      "step": 5000
    },
    {
      "epoch": 6.83,
      "learning_rate": 1.7635482894338482e-05,
      "loss": 1.4654,
      "step": 5010
    },
    {
      "epoch": 6.84,
      "learning_rate": 1.755979412655162e-05,
      "loss": 1.5003,
      "step": 5020
    },
    {
      "epoch": 6.85,
      "learning_rate": 1.748410535876476e-05,
      "loss": 1.5115,
      "step": 5030
    },
    {
      "epoch": 6.87,
      "learning_rate": 1.74084165909779e-05,
      "loss": 1.4893,
      "step": 5040
    },
    {
      "epoch": 6.88,
      "learning_rate": 1.7332727823191038e-05,
      "loss": 1.5016,
      "step": 5050
    },
    {
      "epoch": 6.89,
      "learning_rate": 1.725703905540418e-05,
      "loss": 1.4657,
      "step": 5060
    },
    {
      "epoch": 6.91,
      "learning_rate": 1.7181350287617316e-05,
      "loss": 1.4693,
      "step": 5070
    },
    {
      "epoch": 6.92,
      "learning_rate": 1.7105661519830457e-05,
      "loss": 1.465,
      "step": 5080
    },
    {
      "epoch": 6.93,
      "learning_rate": 1.7029972752043597e-05,
      "loss": 1.4474,
      "step": 5090
    },
    {
      "epoch": 6.95,
      "learning_rate": 1.6954283984256735e-05,
      "loss": 1.4942,
      "step": 5100
    },
    {
      "epoch": 6.96,
      "learning_rate": 1.687859521646988e-05,
      "loss": 1.4921,
      "step": 5110
    },
    {
      "epoch": 6.98,
      "learning_rate": 1.6802906448683016e-05,
      "loss": 1.4864,
      "step": 5120
    },
    {
      "epoch": 6.99,
      "learning_rate": 1.6727217680896157e-05,
      "loss": 1.4846,
      "step": 5130
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.40293742017879947,
      "eval_loss": 1.3961787223815918,
      "eval_runtime": 363.4809,
      "eval_samples_per_second": 64.625,
      "eval_steps_per_second": 2.022,
      "step": 5138
    },
    {
      "epoch": 7.0,
      "learning_rate": 1.6651528913109298e-05,
      "loss": 1.483,
      "step": 5140
    },
    {
      "epoch": 7.02,
      "learning_rate": 1.6575840145322435e-05,
      "loss": 1.4511,
      "step": 5150
    },
    {
      "epoch": 7.03,
      "learning_rate": 1.6500151377535576e-05,
      "loss": 1.4765,
      "step": 5160
    },
    {
      "epoch": 7.04,
      "learning_rate": 1.6424462609748713e-05,
      "loss": 1.461,
      "step": 5170
    },
    {
      "epoch": 7.06,
      "learning_rate": 1.6348773841961854e-05,
      "loss": 1.4666,
      "step": 5180
    },
    {
      "epoch": 7.07,
      "learning_rate": 1.6273085074174994e-05,
      "loss": 1.4539,
      "step": 5190
    },
    {
      "epoch": 7.08,
      "learning_rate": 1.6197396306388132e-05,
      "loss": 1.4603,
      "step": 5200
    },
    {
      "epoch": 7.1,
      "learning_rate": 1.6121707538601272e-05,
      "loss": 1.4775,
      "step": 5210
    },
    {
      "epoch": 7.11,
      "learning_rate": 1.604601877081441e-05,
      "loss": 1.4138,
      "step": 5220
    },
    {
      "epoch": 7.13,
      "learning_rate": 1.597033000302755e-05,
      "loss": 1.4511,
      "step": 5230
    },
    {
      "epoch": 7.14,
      "learning_rate": 1.589464123524069e-05,
      "loss": 1.4703,
      "step": 5240
    },
    {
      "epoch": 7.15,
      "learning_rate": 1.5818952467453832e-05,
      "loss": 1.4648,
      "step": 5250
    },
    {
      "epoch": 7.17,
      "learning_rate": 1.5743263699666973e-05,
      "loss": 1.4671,
      "step": 5260
    },
    {
      "epoch": 7.18,
      "learning_rate": 1.566757493188011e-05,
      "loss": 1.4808,
      "step": 5270
    },
    {
      "epoch": 7.19,
      "learning_rate": 1.559188616409325e-05,
      "loss": 1.4762,
      "step": 5280
    },
    {
      "epoch": 7.21,
      "learning_rate": 1.5516197396306388e-05,
      "loss": 1.4981,
      "step": 5290
    },
    {
      "epoch": 7.22,
      "learning_rate": 1.544050862851953e-05,
      "loss": 1.4441,
      "step": 5300
    },
    {
      "epoch": 7.23,
      "learning_rate": 1.536481986073267e-05,
      "loss": 1.4749,
      "step": 5310
    },
    {
      "epoch": 7.25,
      "learning_rate": 1.5289131092945807e-05,
      "loss": 1.4946,
      "step": 5320
    },
    {
      "epoch": 7.26,
      "learning_rate": 1.5213442325158947e-05,
      "loss": 1.4688,
      "step": 5330
    },
    {
      "epoch": 7.28,
      "learning_rate": 1.5137753557372086e-05,
      "loss": 1.4571,
      "step": 5340
    },
    {
      "epoch": 7.29,
      "learning_rate": 1.5062064789585225e-05,
      "loss": 1.4831,
      "step": 5350
    },
    {
      "epoch": 7.3,
      "learning_rate": 1.4986376021798364e-05,
      "loss": 1.4586,
      "step": 5360
    },
    {
      "epoch": 7.32,
      "learning_rate": 1.4910687254011505e-05,
      "loss": 1.4801,
      "step": 5370
    },
    {
      "epoch": 7.33,
      "learning_rate": 1.4834998486224644e-05,
      "loss": 1.4707,
      "step": 5380
    },
    {
      "epoch": 7.34,
      "learning_rate": 1.4759309718437783e-05,
      "loss": 1.4654,
      "step": 5390
    },
    {
      "epoch": 7.36,
      "learning_rate": 1.4683620950650926e-05,
      "loss": 1.4857,
      "step": 5400
    },
    {
      "epoch": 7.37,
      "learning_rate": 1.4607932182864065e-05,
      "loss": 1.472,
      "step": 5410
    },
    {
      "epoch": 7.38,
      "learning_rate": 1.4532243415077204e-05,
      "loss": 1.4684,
      "step": 5420
    },
    {
      "epoch": 7.4,
      "learning_rate": 1.4456554647290343e-05,
      "loss": 1.4817,
      "step": 5430
    },
    {
      "epoch": 7.41,
      "learning_rate": 1.4380865879503483e-05,
      "loss": 1.4945,
      "step": 5440
    },
    {
      "epoch": 7.43,
      "learning_rate": 1.4305177111716622e-05,
      "loss": 1.4482,
      "step": 5450
    },
    {
      "epoch": 7.44,
      "learning_rate": 1.4229488343929761e-05,
      "loss": 1.4632,
      "step": 5460
    },
    {
      "epoch": 7.45,
      "learning_rate": 1.41537995761429e-05,
      "loss": 1.4625,
      "step": 5470
    },
    {
      "epoch": 7.47,
      "learning_rate": 1.4078110808356041e-05,
      "loss": 1.4616,
      "step": 5480
    },
    {
      "epoch": 7.48,
      "learning_rate": 1.400242204056918e-05,
      "loss": 1.4855,
      "step": 5490
    },
    {
      "epoch": 7.49,
      "learning_rate": 1.392673327278232e-05,
      "loss": 1.443,
      "step": 5500
    },
    {
      "epoch": 7.51,
      "learning_rate": 1.3851044504995458e-05,
      "loss": 1.4805,
      "step": 5510
    },
    {
      "epoch": 7.52,
      "learning_rate": 1.3775355737208597e-05,
      "loss": 1.4757,
      "step": 5520
    },
    {
      "epoch": 7.53,
      "learning_rate": 1.3699666969421738e-05,
      "loss": 1.4493,
      "step": 5530
    },
    {
      "epoch": 7.55,
      "learning_rate": 1.3623978201634879e-05,
      "loss": 1.4438,
      "step": 5540
    },
    {
      "epoch": 7.56,
      "learning_rate": 1.354828943384802e-05,
      "loss": 1.4696,
      "step": 5550
    },
    {
      "epoch": 7.57,
      "learning_rate": 1.3472600666061158e-05,
      "loss": 1.4439,
      "step": 5560
    },
    {
      "epoch": 7.59,
      "learning_rate": 1.3396911898274297e-05,
      "loss": 1.4162,
      "step": 5570
    },
    {
      "epoch": 7.6,
      "learning_rate": 1.3321223130487436e-05,
      "loss": 1.486,
      "step": 5580
    },
    {
      "epoch": 7.62,
      "learning_rate": 1.3245534362700575e-05,
      "loss": 1.5021,
      "step": 5590
    },
    {
      "epoch": 7.63,
      "learning_rate": 1.3169845594913716e-05,
      "loss": 1.4685,
      "step": 5600
    },
    {
      "epoch": 7.64,
      "learning_rate": 1.3094156827126855e-05,
      "loss": 1.4651,
      "step": 5610
    },
    {
      "epoch": 7.66,
      "learning_rate": 1.3018468059339994e-05,
      "loss": 1.4672,
      "step": 5620
    },
    {
      "epoch": 7.67,
      "learning_rate": 1.2942779291553133e-05,
      "loss": 1.4606,
      "step": 5630
    },
    {
      "epoch": 7.68,
      "learning_rate": 1.2867090523766274e-05,
      "loss": 1.4651,
      "step": 5640
    },
    {
      "epoch": 7.7,
      "learning_rate": 1.2791401755979413e-05,
      "loss": 1.471,
      "step": 5650
    },
    {
      "epoch": 7.71,
      "learning_rate": 1.2715712988192552e-05,
      "loss": 1.5029,
      "step": 5660
    },
    {
      "epoch": 7.72,
      "learning_rate": 1.2640024220405691e-05,
      "loss": 1.435,
      "step": 5670
    },
    {
      "epoch": 7.74,
      "learning_rate": 1.256433545261883e-05,
      "loss": 1.4851,
      "step": 5680
    },
    {
      "epoch": 7.75,
      "learning_rate": 1.248864668483197e-05,
      "loss": 1.4494,
      "step": 5690
    },
    {
      "epoch": 7.77,
      "learning_rate": 1.2412957917045111e-05,
      "loss": 1.467,
      "step": 5700
    },
    {
      "epoch": 7.78,
      "learning_rate": 1.233726914925825e-05,
      "loss": 1.4781,
      "step": 5710
    },
    {
      "epoch": 7.79,
      "learning_rate": 1.2261580381471391e-05,
      "loss": 1.492,
      "step": 5720
    },
    {
      "epoch": 7.81,
      "learning_rate": 1.218589161368453e-05,
      "loss": 1.4414,
      "step": 5730
    },
    {
      "epoch": 7.82,
      "learning_rate": 1.211020284589767e-05,
      "loss": 1.4542,
      "step": 5740
    },
    {
      "epoch": 7.83,
      "learning_rate": 1.2034514078110808e-05,
      "loss": 1.4985,
      "step": 5750
    },
    {
      "epoch": 7.85,
      "learning_rate": 1.1958825310323949e-05,
      "loss": 1.457,
      "step": 5760
    },
    {
      "epoch": 7.86,
      "learning_rate": 1.1883136542537088e-05,
      "loss": 1.4695,
      "step": 5770
    },
    {
      "epoch": 7.87,
      "learning_rate": 1.1807447774750227e-05,
      "loss": 1.4525,
      "step": 5780
    },
    {
      "epoch": 7.89,
      "learning_rate": 1.1731759006963368e-05,
      "loss": 1.4464,
      "step": 5790
    },
    {
      "epoch": 7.9,
      "learning_rate": 1.1656070239176507e-05,
      "loss": 1.4726,
      "step": 5800
    },
    {
      "epoch": 7.92,
      "learning_rate": 1.1580381471389647e-05,
      "loss": 1.4199,
      "step": 5810
    },
    {
      "epoch": 7.93,
      "learning_rate": 1.1504692703602786e-05,
      "loss": 1.4717,
      "step": 5820
    },
    {
      "epoch": 7.94,
      "learning_rate": 1.1429003935815925e-05,
      "loss": 1.4747,
      "step": 5830
    },
    {
      "epoch": 7.96,
      "learning_rate": 1.1353315168029064e-05,
      "loss": 1.4479,
      "step": 5840
    },
    {
      "epoch": 7.97,
      "learning_rate": 1.1277626400242203e-05,
      "loss": 1.4486,
      "step": 5850
    },
    {
      "epoch": 7.98,
      "learning_rate": 1.1201937632455344e-05,
      "loss": 1.4418,
      "step": 5860
    },
    {
      "epoch": 8.0,
      "learning_rate": 1.1126248864668485e-05,
      "loss": 1.4729,
      "step": 5870
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.40263942103022565,
      "eval_loss": 1.3931641578674316,
      "eval_runtime": 345.2517,
      "eval_samples_per_second": 68.037,
      "eval_steps_per_second": 2.129,
      "step": 5872
    },
    {
      "epoch": 8.01,
      "learning_rate": 1.1050560096881624e-05,
      "loss": 1.4335,
      "step": 5880
    },
    {
      "epoch": 8.02,
      "learning_rate": 1.0974871329094763e-05,
      "loss": 1.4454,
      "step": 5890
    },
    {
      "epoch": 8.04,
      "learning_rate": 1.0899182561307902e-05,
      "loss": 1.4554,
      "step": 5900
    },
    {
      "epoch": 8.05,
      "learning_rate": 1.0823493793521041e-05,
      "loss": 1.4964,
      "step": 5910
    },
    {
      "epoch": 8.07,
      "learning_rate": 1.0747805025734182e-05,
      "loss": 1.476,
      "step": 5920
    },
    {
      "epoch": 8.08,
      "learning_rate": 1.067211625794732e-05,
      "loss": 1.4713,
      "step": 5930
    },
    {
      "epoch": 8.09,
      "learning_rate": 1.0596427490160461e-05,
      "loss": 1.4944,
      "step": 5940
    },
    {
      "epoch": 8.11,
      "learning_rate": 1.05207387223736e-05,
      "loss": 1.4448,
      "step": 5950
    },
    {
      "epoch": 8.12,
      "learning_rate": 1.044504995458674e-05,
      "loss": 1.44,
      "step": 5960
    },
    {
      "epoch": 8.13,
      "learning_rate": 1.036936118679988e-05,
      "loss": 1.4512,
      "step": 5970
    },
    {
      "epoch": 8.15,
      "learning_rate": 1.0293672419013019e-05,
      "loss": 1.4681,
      "step": 5980
    },
    {
      "epoch": 8.16,
      "learning_rate": 1.0217983651226158e-05,
      "loss": 1.4397,
      "step": 5990
    },
    {
      "epoch": 8.17,
      "learning_rate": 1.0142294883439297e-05,
      "loss": 1.4369,
      "step": 6000
    },
    {
      "epoch": 8.19,
      "learning_rate": 1.0066606115652438e-05,
      "loss": 1.4679,
      "step": 6010
    },
    {
      "epoch": 8.2,
      "learning_rate": 9.990917347865577e-06,
      "loss": 1.439,
      "step": 6020
    },
    {
      "epoch": 8.22,
      "learning_rate": 9.915228580078718e-06,
      "loss": 1.4614,
      "step": 6030
    },
    {
      "epoch": 8.23,
      "learning_rate": 9.839539812291857e-06,
      "loss": 1.4872,
      "step": 6040
    },
    {
      "epoch": 8.24,
      "learning_rate": 9.763851044504996e-06,
      "loss": 1.4459,
      "step": 6050
    },
    {
      "epoch": 8.26,
      "learning_rate": 9.688162276718135e-06,
      "loss": 1.4655,
      "step": 6060
    },
    {
      "epoch": 8.27,
      "learning_rate": 9.612473508931275e-06,
      "loss": 1.4264,
      "step": 6070
    },
    {
      "epoch": 8.28,
      "learning_rate": 9.536784741144416e-06,
      "loss": 1.4446,
      "step": 6080
    },
    {
      "epoch": 8.3,
      "learning_rate": 9.461095973357555e-06,
      "loss": 1.438,
      "step": 6090
    },
    {
      "epoch": 8.31,
      "learning_rate": 9.385407205570694e-06,
      "loss": 1.4489,
      "step": 6100
    },
    {
      "epoch": 8.32,
      "learning_rate": 9.309718437783833e-06,
      "loss": 1.4553,
      "step": 6110
    },
    {
      "epoch": 8.34,
      "learning_rate": 9.234029669996972e-06,
      "loss": 1.4156,
      "step": 6120
    },
    {
      "epoch": 8.35,
      "learning_rate": 9.158340902210113e-06,
      "loss": 1.4172,
      "step": 6130
    },
    {
      "epoch": 8.37,
      "learning_rate": 9.082652134423252e-06,
      "loss": 1.4247,
      "step": 6140
    },
    {
      "epoch": 8.38,
      "learning_rate": 9.006963366636391e-06,
      "loss": 1.4101,
      "step": 6150
    },
    {
      "epoch": 8.39,
      "learning_rate": 8.931274598849532e-06,
      "loss": 1.4653,
      "step": 6160
    },
    {
      "epoch": 8.41,
      "learning_rate": 8.85558583106267e-06,
      "loss": 1.4503,
      "step": 6170
    },
    {
      "epoch": 8.42,
      "learning_rate": 8.77989706327581e-06,
      "loss": 1.4485,
      "step": 6180
    },
    {
      "epoch": 8.43,
      "learning_rate": 8.70420829548895e-06,
      "loss": 1.4657,
      "step": 6190
    },
    {
      "epoch": 8.45,
      "learning_rate": 8.62851952770209e-06,
      "loss": 1.4497,
      "step": 6200
    },
    {
      "epoch": 8.46,
      "learning_rate": 8.552830759915228e-06,
      "loss": 1.4652,
      "step": 6210
    },
    {
      "epoch": 8.47,
      "learning_rate": 8.477141992128367e-06,
      "loss": 1.4186,
      "step": 6220
    },
    {
      "epoch": 8.49,
      "learning_rate": 8.401453224341508e-06,
      "loss": 1.4756,
      "step": 6230
    },
    {
      "epoch": 8.5,
      "learning_rate": 8.325764456554649e-06,
      "loss": 1.4448,
      "step": 6240
    },
    {
      "epoch": 8.51,
      "learning_rate": 8.250075688767788e-06,
      "loss": 1.4527,
      "step": 6250
    },
    {
      "epoch": 8.53,
      "learning_rate": 8.174386920980927e-06,
      "loss": 1.4815,
      "step": 6260
    },
    {
      "epoch": 8.54,
      "learning_rate": 8.098698153194066e-06,
      "loss": 1.4719,
      "step": 6270
    },
    {
      "epoch": 8.56,
      "learning_rate": 8.023009385407205e-06,
      "loss": 1.4552,
      "step": 6280
    },
    {
      "epoch": 8.57,
      "learning_rate": 7.947320617620346e-06,
      "loss": 1.4844,
      "step": 6290
    },
    {
      "epoch": 8.58,
      "learning_rate": 7.871631849833486e-06,
      "loss": 1.4661,
      "step": 6300
    },
    {
      "epoch": 8.6,
      "learning_rate": 7.795943082046625e-06,
      "loss": 1.4276,
      "step": 6310
    },
    {
      "epoch": 8.61,
      "learning_rate": 7.720254314259764e-06,
      "loss": 1.4753,
      "step": 6320
    },
    {
      "epoch": 8.62,
      "learning_rate": 7.644565546472903e-06,
      "loss": 1.471,
      "step": 6330
    },
    {
      "epoch": 8.64,
      "learning_rate": 7.568876778686043e-06,
      "loss": 1.4287,
      "step": 6340
    },
    {
      "epoch": 8.65,
      "learning_rate": 7.493188010899182e-06,
      "loss": 1.4716,
      "step": 6350
    },
    {
      "epoch": 8.66,
      "learning_rate": 7.417499243112322e-06,
      "loss": 1.4654,
      "step": 6360
    },
    {
      "epoch": 8.68,
      "learning_rate": 7.341810475325463e-06,
      "loss": 1.4302,
      "step": 6370
    },
    {
      "epoch": 8.69,
      "learning_rate": 7.266121707538602e-06,
      "loss": 1.4571,
      "step": 6380
    },
    {
      "epoch": 8.71,
      "learning_rate": 7.190432939751742e-06,
      "loss": 1.4737,
      "step": 6390
    },
    {
      "epoch": 8.72,
      "learning_rate": 7.114744171964881e-06,
      "loss": 1.4454,
      "step": 6400
    },
    {
      "epoch": 8.73,
      "learning_rate": 7.039055404178021e-06,
      "loss": 1.4239,
      "step": 6410
    },
    {
      "epoch": 8.75,
      "learning_rate": 6.96336663639116e-06,
      "loss": 1.4236,
      "step": 6420
    },
    {
      "epoch": 8.76,
      "learning_rate": 6.887677868604299e-06,
      "loss": 1.4426,
      "step": 6430
    },
    {
      "epoch": 8.77,
      "learning_rate": 6.811989100817439e-06,
      "loss": 1.4538,
      "step": 6440
    },
    {
      "epoch": 8.79,
      "learning_rate": 6.736300333030579e-06,
      "loss": 1.4653,
      "step": 6450
    },
    {
      "epoch": 8.8,
      "learning_rate": 6.660611565243718e-06,
      "loss": 1.4522,
      "step": 6460
    },
    {
      "epoch": 8.81,
      "learning_rate": 6.584922797456858e-06,
      "loss": 1.4433,
      "step": 6470
    },
    {
      "epoch": 8.83,
      "learning_rate": 6.509234029669997e-06,
      "loss": 1.4623,
      "step": 6480
    },
    {
      "epoch": 8.84,
      "learning_rate": 6.433545261883137e-06,
      "loss": 1.4426,
      "step": 6490
    },
    {
      "epoch": 8.86,
      "learning_rate": 6.357856494096276e-06,
      "loss": 1.4617,
      "step": 6500
    },
    {
      "epoch": 8.87,
      "learning_rate": 6.282167726309415e-06,
      "loss": 1.4359,
      "step": 6510
    },
    {
      "epoch": 8.88,
      "learning_rate": 6.206478958522556e-06,
      "loss": 1.4155,
      "step": 6520
    },
    {
      "epoch": 8.9,
      "learning_rate": 6.1307901907356956e-06,
      "loss": 1.4214,
      "step": 6530
    },
    {
      "epoch": 8.91,
      "learning_rate": 6.055101422948835e-06,
      "loss": 1.4762,
      "step": 6540
    },
    {
      "epoch": 8.92,
      "learning_rate": 5.9794126551619744e-06,
      "loss": 1.4665,
      "step": 6550
    },
    {
      "epoch": 8.94,
      "learning_rate": 5.9037238873751135e-06,
      "loss": 1.4446,
      "step": 6560
    },
    {
      "epoch": 8.95,
      "learning_rate": 5.828035119588253e-06,
      "loss": 1.4856,
      "step": 6570
    },
    {
      "epoch": 8.96,
      "learning_rate": 5.752346351801393e-06,
      "loss": 1.4426,
      "step": 6580
    },
    {
      "epoch": 8.98,
      "learning_rate": 5.676657584014532e-06,
      "loss": 1.4767,
      "step": 6590
    },
    {
      "epoch": 8.99,
      "learning_rate": 5.600968816227672e-06,
      "loss": 1.4416,
      "step": 6600
    },
    {
      "epoch": 9.0,
      "eval_accuracy": 0.40881226053639846,
      "eval_loss": 1.382081389427185,
      "eval_runtime": 335.0638,
      "eval_samples_per_second": 70.106,
      "eval_steps_per_second": 2.194,
      "step": 6606
    },
    {
      "epoch": 9.01,
      "learning_rate": 5.525280048440812e-06,
      "loss": 1.4128,
      "step": 6610
    },
    {
      "epoch": 9.02,
      "learning_rate": 5.449591280653951e-06,
      "loss": 1.4838,
      "step": 6620
    },
    {
      "epoch": 9.03,
      "learning_rate": 5.373902512867091e-06,
      "loss": 1.442,
      "step": 6630
    },
    {
      "epoch": 9.05,
      "learning_rate": 5.298213745080231e-06,
      "loss": 1.4423,
      "step": 6640
    },
    {
      "epoch": 9.06,
      "learning_rate": 5.22252497729337e-06,
      "loss": 1.4571,
      "step": 6650
    },
    {
      "epoch": 9.07,
      "learning_rate": 5.1468362095065096e-06,
      "loss": 1.4366,
      "step": 6660
    },
    {
      "epoch": 9.09,
      "learning_rate": 5.071147441719649e-06,
      "loss": 1.4392,
      "step": 6670
    },
    {
      "epoch": 9.1,
      "learning_rate": 4.9954586739327884e-06,
      "loss": 1.4108,
      "step": 6680
    },
    {
      "epoch": 9.11,
      "learning_rate": 4.919769906145928e-06,
      "loss": 1.4265,
      "step": 6690
    },
    {
      "epoch": 9.13,
      "learning_rate": 4.844081138359067e-06,
      "loss": 1.4451,
      "step": 6700
    },
    {
      "epoch": 9.14,
      "learning_rate": 4.768392370572208e-06,
      "loss": 1.4532,
      "step": 6710
    },
    {
      "epoch": 9.16,
      "learning_rate": 4.692703602785347e-06,
      "loss": 1.4238,
      "step": 6720
    },
    {
      "epoch": 9.17,
      "learning_rate": 4.617014834998486e-06,
      "loss": 1.4505,
      "step": 6730
    },
    {
      "epoch": 9.18,
      "learning_rate": 4.541326067211626e-06,
      "loss": 1.4523,
      "step": 6740
    },
    {
      "epoch": 9.2,
      "learning_rate": 4.465637299424766e-06,
      "loss": 1.4535,
      "step": 6750
    },
    {
      "epoch": 9.21,
      "learning_rate": 4.389948531637905e-06,
      "loss": 1.4517,
      "step": 6760
    },
    {
      "epoch": 9.22,
      "learning_rate": 4.314259763851045e-06,
      "loss": 1.4328,
      "step": 6770
    },
    {
      "epoch": 9.24,
      "learning_rate": 4.238570996064184e-06,
      "loss": 1.4395,
      "step": 6780
    },
    {
      "epoch": 9.25,
      "learning_rate": 4.162882228277324e-06,
      "loss": 1.3879,
      "step": 6790
    },
    {
      "epoch": 9.26,
      "learning_rate": 4.0871934604904634e-06,
      "loss": 1.4199,
      "step": 6800
    },
    {
      "epoch": 9.28,
      "learning_rate": 4.0115046927036024e-06,
      "loss": 1.467,
      "step": 6810
    },
    {
      "epoch": 9.29,
      "learning_rate": 3.935815924916743e-06,
      "loss": 1.4773,
      "step": 6820
    },
    {
      "epoch": 9.31,
      "learning_rate": 3.860127157129882e-06,
      "loss": 1.4594,
      "step": 6830
    },
    {
      "epoch": 9.32,
      "learning_rate": 3.7844383893430216e-06,
      "loss": 1.3987,
      "step": 6840
    },
    {
      "epoch": 9.33,
      "learning_rate": 3.708749621556161e-06,
      "loss": 1.455,
      "step": 6850
    },
    {
      "epoch": 9.35,
      "learning_rate": 3.633060853769301e-06,
      "loss": 1.4301,
      "step": 6860
    },
    {
      "epoch": 9.36,
      "learning_rate": 3.5573720859824404e-06,
      "loss": 1.4645,
      "step": 6870
    },
    {
      "epoch": 9.37,
      "learning_rate": 3.48168331819558e-06,
      "loss": 1.4648,
      "step": 6880
    },
    {
      "epoch": 9.39,
      "learning_rate": 3.4059945504087197e-06,
      "loss": 1.4308,
      "step": 6890
    },
    {
      "epoch": 9.4,
      "learning_rate": 3.330305782621859e-06,
      "loss": 1.398,
      "step": 6900
    },
    {
      "epoch": 9.41,
      "learning_rate": 3.2546170148349985e-06,
      "loss": 1.4166,
      "step": 6910
    },
    {
      "epoch": 9.43,
      "learning_rate": 3.178928247048138e-06,
      "loss": 1.4483,
      "step": 6920
    },
    {
      "epoch": 9.44,
      "learning_rate": 3.103239479261278e-06,
      "loss": 1.433,
      "step": 6930
    },
    {
      "epoch": 9.46,
      "learning_rate": 3.0275507114744173e-06,
      "loss": 1.457,
      "step": 6940
    },
    {
      "epoch": 9.47,
      "learning_rate": 2.9518619436875567e-06,
      "loss": 1.4638,
      "step": 6950
    },
    {
      "epoch": 9.48,
      "learning_rate": 2.8761731759006966e-06,
      "loss": 1.4256,
      "step": 6960
    },
    {
      "epoch": 9.5,
      "learning_rate": 2.800484408113836e-06,
      "loss": 1.43,
      "step": 6970
    },
    {
      "epoch": 9.51,
      "learning_rate": 2.7247956403269755e-06,
      "loss": 1.411,
      "step": 6980
    },
    {
      "epoch": 9.52,
      "learning_rate": 2.6491068725401153e-06,
      "loss": 1.4436,
      "step": 6990
    },
    {
      "epoch": 9.54,
      "learning_rate": 2.5734181047532548e-06,
      "loss": 1.4774,
      "step": 7000
    },
    {
      "epoch": 9.55,
      "learning_rate": 2.4977293369663942e-06,
      "loss": 1.4343,
      "step": 7010
    },
    {
      "epoch": 9.56,
      "learning_rate": 2.4220405691795337e-06,
      "loss": 1.4242,
      "step": 7020
    },
    {
      "epoch": 9.58,
      "learning_rate": 2.3463518013926735e-06,
      "loss": 1.4499,
      "step": 7030
    },
    {
      "epoch": 9.59,
      "learning_rate": 2.270663033605813e-06,
      "loss": 1.4385,
      "step": 7040
    },
    {
      "epoch": 9.6,
      "learning_rate": 2.1949742658189524e-06,
      "loss": 1.4419,
      "step": 7050
    },
    {
      "epoch": 9.62,
      "learning_rate": 2.119285498032092e-06,
      "loss": 1.4446,
      "step": 7060
    },
    {
      "epoch": 9.63,
      "learning_rate": 2.0435967302452317e-06,
      "loss": 1.4296,
      "step": 7070
    },
    {
      "epoch": 9.65,
      "learning_rate": 1.9679079624583716e-06,
      "loss": 1.454,
      "step": 7080
    },
    {
      "epoch": 9.66,
      "learning_rate": 1.8922191946715108e-06,
      "loss": 1.4488,
      "step": 7090
    },
    {
      "epoch": 9.67,
      "learning_rate": 1.8165304268846505e-06,
      "loss": 1.4558,
      "step": 7100
    },
    {
      "epoch": 9.69,
      "learning_rate": 1.74084165909779e-06,
      "loss": 1.4338,
      "step": 7110
    },
    {
      "epoch": 9.7,
      "learning_rate": 1.6651528913109296e-06,
      "loss": 1.4345,
      "step": 7120
    },
    {
      "epoch": 9.71,
      "learning_rate": 1.589464123524069e-06,
      "loss": 1.4213,
      "step": 7130
    },
    {
      "epoch": 9.73,
      "learning_rate": 1.5137753557372086e-06,
      "loss": 1.4392,
      "step": 7140
    },
    {
      "epoch": 9.74,
      "learning_rate": 1.4380865879503483e-06,
      "loss": 1.4382,
      "step": 7150
    },
    {
      "epoch": 9.75,
      "learning_rate": 1.3623978201634877e-06,
      "loss": 1.4347,
      "step": 7160
    },
    {
      "epoch": 9.77,
      "learning_rate": 1.2867090523766274e-06,
      "loss": 1.431,
      "step": 7170
    },
    {
      "epoch": 9.78,
      "learning_rate": 1.2110202845897668e-06,
      "loss": 1.4352,
      "step": 7180
    },
    {
      "epoch": 9.8,
      "learning_rate": 1.1353315168029065e-06,
      "loss": 1.4536,
      "step": 7190
    },
    {
      "epoch": 9.81,
      "learning_rate": 1.059642749016046e-06,
      "loss": 1.4236,
      "step": 7200
    },
    {
      "epoch": 9.82,
      "learning_rate": 9.839539812291858e-07,
      "loss": 1.4592,
      "step": 7210
    },
    {
      "epoch": 9.84,
      "learning_rate": 9.082652134423252e-07,
      "loss": 1.4574,
      "step": 7220
    },
    {
      "epoch": 9.85,
      "learning_rate": 8.325764456554648e-07,
      "loss": 1.443,
      "step": 7230
    },
    {
      "epoch": 9.86,
      "learning_rate": 7.568876778686043e-07,
      "loss": 1.4423,
      "step": 7240
    },
    {
      "epoch": 9.88,
      "learning_rate": 6.811989100817439e-07,
      "loss": 1.4055,
      "step": 7250
    },
    {
      "epoch": 9.89,
      "learning_rate": 6.055101422948834e-07,
      "loss": 1.4245,
      "step": 7260
    },
    {
      "epoch": 9.9,
      "learning_rate": 5.29821374508023e-07,
      "loss": 1.4609,
      "step": 7270
    },
    {
      "epoch": 9.92,
      "learning_rate": 4.541326067211626e-07,
      "loss": 1.4397,
      "step": 7280
    },
    {
      "epoch": 9.93,
      "learning_rate": 3.7844383893430216e-07,
      "loss": 1.4377,
      "step": 7290
    },
    {
      "epoch": 9.95,
      "learning_rate": 3.027550711474417e-07,
      "loss": 1.4288,
      "step": 7300
    },
    {
      "epoch": 9.96,
      "learning_rate": 2.270663033605813e-07,
      "loss": 1.4391,
      "step": 7310
    },
    {
      "epoch": 9.97,
      "learning_rate": 1.5137753557372085e-07,
      "loss": 1.4762,
      "step": 7320
    },
    {
      "epoch": 9.99,
      "learning_rate": 7.568876778686043e-08,
      "loss": 1.4183,
      "step": 7330
    },
    {
      "epoch": 10.0,
      "learning_rate": 0.0,
      "loss": 1.4255,
      "step": 7340
    },
    {
      "epoch": 10.0,
      "eval_accuracy": 0.4106853980417199,
      "eval_loss": 1.3763015270233154,
      "eval_runtime": 348.1925,
      "eval_samples_per_second": 67.463,
      "eval_steps_per_second": 2.111,
      "step": 7340
    },
    {
      "epoch": 10.0,
      "step": 7340,
      "total_flos": 2.335502900992665e+19,
      "train_loss": 1.529659651410677,
      "train_runtime": 20641.2908,
      "train_samples_per_second": 45.512,
      "train_steps_per_second": 0.356
    }
  ],
  "logging_steps": 10,
  "max_steps": 7340,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "save_steps": 500,
  "total_flos": 2.335502900992665e+19,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}