{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 5000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "grad_norm": 0.3745611011981964,
      "learning_rate": 5e-06,
      "loss": 2.5464,
      "step": 50
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6350923180580139,
      "learning_rate": 1e-05,
      "loss": 2.5921,
      "step": 100
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1115145683288574,
      "learning_rate": 1.5e-05,
      "loss": 2.5699,
      "step": 150
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1577266454696655,
      "learning_rate": 2e-05,
      "loss": 2.5473,
      "step": 200
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.4771063327789307,
      "learning_rate": 2.5e-05,
      "loss": 2.5239,
      "step": 250
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.695399522781372,
      "learning_rate": 3e-05,
      "loss": 2.4753,
      "step": 300
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5696115493774414,
      "learning_rate": 3.5e-05,
      "loss": 2.4721,
      "step": 350
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8710404634475708,
      "learning_rate": 4e-05,
      "loss": 2.3697,
      "step": 400
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.613884210586548,
      "learning_rate": 4.5e-05,
      "loss": 2.2966,
      "step": 450
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.416072368621826,
      "learning_rate": 5e-05,
      "loss": 2.4018,
      "step": 500
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.746003270149231,
      "learning_rate": 4.99847706754774e-05,
      "loss": 2.2825,
      "step": 550
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.941234827041626,
      "learning_rate": 4.993910125649561e-05,
      "loss": 2.473,
      "step": 600
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.632983922958374,
      "learning_rate": 4.9863047384206835e-05,
      "loss": 2.2963,
      "step": 650
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.2639946937561035,
      "learning_rate": 4.975670171853926e-05,
      "loss": 2.2385,
      "step": 700
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5031816959381104,
      "learning_rate": 4.962019382530521e-05,
      "loss": 2.2266,
      "step": 750
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3354930877685547,
      "learning_rate": 4.9453690018345144e-05,
      "loss": 2.2189,
      "step": 800
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.2369980812072754,
      "learning_rate": 4.925739315689991e-05,
      "loss": 2.2916,
      "step": 850
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5904576778411865,
      "learning_rate": 4.9031542398457974e-05,
      "loss": 2.1422,
      "step": 900
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.333570957183838,
      "learning_rate": 4.877641290737884e-05,
      "loss": 2.3032,
      "step": 950
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.157676935195923,
      "learning_rate": 4.849231551964771e-05,
      "loss": 2.1895,
      "step": 1000
    },
    {
      "epoch": 0.21,
      "grad_norm": 8.637140274047852,
      "learning_rate": 4.817959636416969e-05,
      "loss": 2.3375,
      "step": 1050
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.910135507583618,
      "learning_rate": 4.783863644106502e-05,
      "loss": 2.261,
      "step": 1100
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8060033321380615,
      "learning_rate": 4.7469851157479177e-05,
      "loss": 2.198,
      "step": 1150
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5828884840011597,
      "learning_rate": 4.707368982147318e-05,
      "loss": 2.1015,
      "step": 1200
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.2127833366394043,
      "learning_rate": 4.665063509461097e-05,
      "loss": 2.2895,
      "step": 1250
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5719356536865234,
      "learning_rate": 4.620120240391065e-05,
      "loss": 2.3689,
      "step": 1300
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.7176144123077393,
      "learning_rate": 4.572593931387604e-05,
      "loss": 2.3083,
      "step": 1350
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.401768684387207,
      "learning_rate": 4.522542485937369e-05,
      "loss": 2.1516,
      "step": 1400
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.553105354309082,
      "learning_rate": 4.4700268840168045e-05,
      "loss": 2.3064,
      "step": 1450
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.074903964996338,
      "learning_rate": 4.415111107797445e-05,
      "loss": 2.1008,
      "step": 1500
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.7607364654541016,
      "learning_rate": 4.357862063693486e-05,
      "loss": 2.1576,
      "step": 1550
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.275789976119995,
      "learning_rate": 4.2983495008466276e-05,
      "loss": 2.2486,
      "step": 1600
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.713785409927368,
      "learning_rate": 4.2366459261474933e-05,
      "loss": 2.1851,
      "step": 1650
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.8428258895874023,
      "learning_rate": 4.172826515897146e-05,
      "loss": 2.2279,
      "step": 1700
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.787942409515381,
      "learning_rate": 4.1069690242163484e-05,
      "loss": 2.1374,
      "step": 1750
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.18039083480835,
      "learning_rate": 4.039153688314145e-05,
      "loss": 2.2694,
      "step": 1800
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7699658870697021,
      "learning_rate": 3.969463130731183e-05,
      "loss": 2.2034,
      "step": 1850
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5621023178100586,
      "learning_rate": 3.897982258676867e-05,
      "loss": 2.1479,
      "step": 1900
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.7159836292266846,
      "learning_rate": 3.824798160583012e-05,
      "loss": 2.066,
      "step": 1950
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.071038246154785,
      "learning_rate": 3.7500000000000003e-05,
      "loss": 2.3053,
      "step": 2000
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.285961866378784,
      "learning_rate": 3.673678906964727e-05,
      "loss": 2.3655,
      "step": 2050
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.2050671577453613,
      "learning_rate": 3.5959278669726935e-05,
      "loss": 2.386,
      "step": 2100
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8379002809524536,
      "learning_rate": 3.516841607689501e-05,
      "loss": 2.2381,
      "step": 2150
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.3484580516815186,
      "learning_rate": 3.436516483539781e-05,
      "loss": 2.2748,
      "step": 2200
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.6434378623962402,
      "learning_rate": 3.355050358314172e-05,
      "loss": 2.1108,
      "step": 2250
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.832470417022705,
      "learning_rate": 3.272542485937369e-05,
      "loss": 2.2859,
      "step": 2300
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.5317020416259766,
      "learning_rate": 3.1890933895424976e-05,
      "loss": 2.1841,
      "step": 2350
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.41705060005188,
      "learning_rate": 3.104804738999169e-05,
      "loss": 2.2283,
      "step": 2400
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.1975114345550537,
      "learning_rate": 3.0197792270443982e-05,
      "loss": 2.1952,
      "step": 2450
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.486776828765869,
      "learning_rate": 2.9341204441673266e-05,
      "loss": 2.289,
      "step": 2500
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.697965145111084,
      "learning_rate": 2.8479327524001636e-05,
      "loss": 2.2576,
      "step": 2550
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.713478684425354,
      "learning_rate": 2.761321158169134e-05,
      "loss": 2.1399,
      "step": 2600
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4894109964370728,
      "learning_rate": 2.674391184360313e-05,
      "loss": 2.0986,
      "step": 2650
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.0305914878845215,
      "learning_rate": 2.587248741756253e-05,
      "loss": 2.1002,
      "step": 2700
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.2819197177886963,
      "learning_rate": 2.5e-05,
      "loss": 2.2805,
      "step": 2750
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.552293539047241,
      "learning_rate": 2.4127512582437485e-05,
      "loss": 2.0637,
      "step": 2800
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.8868838548660278,
      "learning_rate": 2.3256088156396868e-05,
      "loss": 2.2437,
      "step": 2850
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.3621270656585693,
      "learning_rate": 2.238678841830867e-05,
      "loss": 2.1867,
      "step": 2900
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.550295829772949,
      "learning_rate": 2.1520672475998373e-05,
      "loss": 2.2582,
      "step": 2950
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.509620428085327,
      "learning_rate": 2.0658795558326743e-05,
      "loss": 2.144,
      "step": 3000
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.188652276992798,
      "learning_rate": 1.980220772955602e-05,
      "loss": 2.1568,
      "step": 3050
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.8035781383514404,
      "learning_rate": 1.895195261000831e-05,
      "loss": 2.1571,
      "step": 3100
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.5277318954467773,
      "learning_rate": 1.8109066104575023e-05,
      "loss": 2.1844,
      "step": 3150
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.469127655029297,
      "learning_rate": 1.7274575140626318e-05,
      "loss": 2.1792,
      "step": 3200
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.5568430423736572,
      "learning_rate": 1.6449496416858284e-05,
      "loss": 2.2361,
      "step": 3250
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.626446008682251,
      "learning_rate": 1.56348351646022e-05,
      "loss": 2.1619,
      "step": 3300
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.9157896041870117,
      "learning_rate": 1.4831583923104999e-05,
      "loss": 2.1609,
      "step": 3350
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.5856120586395264,
      "learning_rate": 1.4040721330273062e-05,
      "loss": 2.1897,
      "step": 3400
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.538245677947998,
      "learning_rate": 1.3263210930352737e-05,
      "loss": 2.0751,
      "step": 3450
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6384202241897583,
      "learning_rate": 1.2500000000000006e-05,
      "loss": 2.2173,
      "step": 3500
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.493316888809204,
      "learning_rate": 1.175201839416988e-05,
      "loss": 2.1041,
      "step": 3550
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.1629319190979004,
      "learning_rate": 1.1020177413231334e-05,
      "loss": 2.2025,
      "step": 3600
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.7992451190948486,
      "learning_rate": 1.0305368692688174e-05,
      "loss": 2.1657,
      "step": 3650
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.4246907234191895,
      "learning_rate": 9.608463116858542e-06,
      "loss": 2.1475,
      "step": 3700
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.659865617752075,
      "learning_rate": 8.930309757836517e-06,
      "loss": 2.1799,
      "step": 3750
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.3712964057922363,
      "learning_rate": 8.271734841028553e-06,
      "loss": 2.0888,
      "step": 3800
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6899696588516235,
      "learning_rate": 7.633540738525066e-06,
      "loss": 2.1238,
      "step": 3850
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6770470142364502,
      "learning_rate": 7.016504991533726e-06,
      "loss": 2.0053,
      "step": 3900
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.5809788703918457,
      "learning_rate": 6.421379363065142e-06,
      "loss": 2.1939,
      "step": 3950
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.4868805408477783,
      "learning_rate": 5.848888922025553e-06,
      "loss": 2.2744,
      "step": 4000
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.573330879211426,
      "learning_rate": 5.299731159831953e-06,
      "loss": 2.2459,
      "step": 4050
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.3828341960906982,
      "learning_rate": 4.7745751406263165e-06,
      "loss": 2.1207,
      "step": 4100
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.077352285385132,
      "learning_rate": 4.274060686123959e-06,
      "loss": 2.2318,
      "step": 4150
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.399592399597168,
      "learning_rate": 3.798797596089351e-06,
      "loss": 2.0679,
      "step": 4200
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.191920042037964,
      "learning_rate": 3.3493649053890326e-06,
      "loss": 2.1493,
      "step": 4250
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.8285448551177979,
      "learning_rate": 2.9263101785268254e-06,
      "loss": 2.2512,
      "step": 4300
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.347266912460327,
      "learning_rate": 2.5301488425208296e-06,
      "loss": 2.1985,
      "step": 4350
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.9266479015350342,
      "learning_rate": 2.1613635589349756e-06,
      "loss": 2.2455,
      "step": 4400
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.067202091217041,
      "learning_rate": 1.8204036358303173e-06,
      "loss": 2.1952,
      "step": 4450
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.765047788619995,
      "learning_rate": 1.5076844803522922e-06,
      "loss": 2.0983,
      "step": 4500
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.322967767715454,
      "learning_rate": 1.2235870926211619e-06,
      "loss": 2.1133,
      "step": 4550
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.262035608291626,
      "learning_rate": 9.684576015420278e-07,
      "loss": 2.1524,
      "step": 4600
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.373628854751587,
      "learning_rate": 7.426068431000882e-07,
      "loss": 2.2132,
      "step": 4650
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.6327779293060303,
      "learning_rate": 5.463099816548579e-07,
      "loss": 2.1586,
      "step": 4700
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.8346914052963257,
      "learning_rate": 3.7980617469479953e-07,
      "loss": 2.2988,
      "step": 4750
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.114342451095581,
      "learning_rate": 2.4329828146074095e-07,
      "loss": 2.2557,
      "step": 4800
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.1969785690307617,
      "learning_rate": 1.3695261579316777e-07,
      "loss": 2.1417,
      "step": 4850
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.7341632843017578,
      "learning_rate": 6.089874350439506e-08,
      "loss": 2.1895,
      "step": 4900
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.1261510848999023,
      "learning_rate": 1.522932452260595e-08,
      "loss": 2.1405,
      "step": 4950
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.2943620681762695,
      "learning_rate": 0.0,
      "loss": 2.1998,
      "step": 5000
    }
  ],
  "logging_steps": 50,
  "max_steps": 5000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 7.314072519030374e+16,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}