{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 20.0,
  "eval_steps": 500,
  "global_step": 371900,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.05,
      "grad_norm": 1.0875967741012573,
      "learning_rate": 3.125e-06,
      "loss": 7.5939,
      "step": 1000
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8077093362808228,
      "learning_rate": 6.25e-06,
      "loss": 5.8309,
      "step": 2000
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8327828049659729,
      "learning_rate": 9.375000000000001e-06,
      "loss": 5.3685,
      "step": 3000
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9203261137008667,
      "learning_rate": 1.25e-05,
      "loss": 5.1581,
      "step": 4000
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9855815768241882,
      "learning_rate": 1.5625e-05,
      "loss": 5.0109,
      "step": 5000
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1521941423416138,
      "learning_rate": 1.8750000000000002e-05,
      "loss": 4.8668,
      "step": 6000
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1727802753448486,
      "learning_rate": 2.1875e-05,
      "loss": 4.759,
      "step": 7000
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1063674688339233,
      "learning_rate": 2.5e-05,
      "loss": 4.6561,
      "step": 8000
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.178877592086792,
      "learning_rate": 2.8125000000000003e-05,
      "loss": 4.5671,
      "step": 9000
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0879851579666138,
      "learning_rate": 3.125e-05,
      "loss": 4.4872,
      "step": 10000
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1763601303100586,
      "learning_rate": 3.4371875e-05,
      "loss": 4.4182,
      "step": 11000
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0679717063903809,
      "learning_rate": 3.7496875e-05,
      "loss": 4.3542,
      "step": 12000
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.051241159439087,
      "learning_rate": 4.061875e-05,
      "loss": 4.2985,
      "step": 13000
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0657079219818115,
      "learning_rate": 4.374375e-05,
      "loss": 4.2397,
      "step": 14000
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0396409034729004,
      "learning_rate": 4.686875e-05,
      "loss": 4.1923,
      "step": 15000
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9875408411026001,
      "learning_rate": 4.999375e-05,
      "loss": 4.1418,
      "step": 16000
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.003056526184082,
      "learning_rate": 5.3115625000000005e-05,
      "loss": 4.091,
      "step": 17000
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9668901562690735,
      "learning_rate": 5.62375e-05,
      "loss": 4.0553,
      "step": 18000
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.3076893208292497,
      "eval_loss": 4.2751569747924805,
      "eval_runtime": 153.8934,
      "eval_samples_per_second": 376.371,
      "eval_steps_per_second": 5.887,
      "step": 18595
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.0316466093063354,
      "learning_rate": 5.93625e-05,
      "loss": 4.0151,
      "step": 19000
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.0109944343566895,
      "learning_rate": 6.24875e-05,
      "loss": 3.9733,
      "step": 20000
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.0080300569534302,
      "learning_rate": 6.56125e-05,
      "loss": 3.9328,
      "step": 21000
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.0754659175872803,
      "learning_rate": 6.873125000000001e-05,
      "loss": 3.8969,
      "step": 22000
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.018548607826233,
      "learning_rate": 7.185625e-05,
      "loss": 3.8657,
      "step": 23000
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9804821014404297,
      "learning_rate": 7.4975e-05,
      "loss": 3.8338,
      "step": 24000
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0100795030593872,
      "learning_rate": 7.81e-05,
      "loss": 3.7995,
      "step": 25000
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9405222535133362,
      "learning_rate": 8.1221875e-05,
      "loss": 3.773,
      "step": 26000
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9475222826004028,
      "learning_rate": 8.434687500000001e-05,
      "loss": 3.751,
      "step": 27000
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.890032172203064,
      "learning_rate": 8.746875e-05,
      "loss": 3.7279,
      "step": 28000
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9058054685592651,
      "learning_rate": 9.059375e-05,
      "loss": 3.7031,
      "step": 29000
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8877508044242859,
      "learning_rate": 9.3715625e-05,
      "loss": 3.6828,
      "step": 30000
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.8418898582458496,
      "learning_rate": 9.684062500000001e-05,
      "loss": 3.6672,
      "step": 31000
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.8882458806037903,
      "learning_rate": 9.9965625e-05,
      "loss": 3.6463,
      "step": 32000
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.8538274168968201,
      "learning_rate": 9.970903206825538e-05,
      "loss": 3.6331,
      "step": 33000
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.8449269533157349,
      "learning_rate": 9.941512209473375e-05,
      "loss": 3.6114,
      "step": 34000
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.8731977343559265,
      "learning_rate": 9.912150632538982e-05,
      "loss": 3.5938,
      "step": 35000
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.8085331916809082,
      "learning_rate": 9.88273021476905e-05,
      "loss": 3.5792,
      "step": 36000
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.8539865612983704,
      "learning_rate": 9.853309796999118e-05,
      "loss": 3.5587,
      "step": 37000
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.3633328691293394,
      "eval_loss": 3.7501370906829834,
      "eval_runtime": 155.089,
      "eval_samples_per_second": 373.47,
      "eval_steps_per_second": 5.842,
      "step": 37190
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.8294937610626221,
      "learning_rate": 9.823889379229185e-05,
      "loss": 3.5228,
      "step": 38000
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.7889467477798462,
      "learning_rate": 9.794468961459253e-05,
      "loss": 3.5132,
      "step": 39000
    },
    {
      "epoch": 2.15,
      "grad_norm": 0.7929906249046326,
      "learning_rate": 9.765048543689321e-05,
      "loss": 3.495,
      "step": 40000
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.8031061887741089,
      "learning_rate": 9.735657546337158e-05,
      "loss": 3.4887,
      "step": 41000
    },
    {
      "epoch": 2.26,
      "grad_norm": 0.8079201579093933,
      "learning_rate": 9.706266548984995e-05,
      "loss": 3.4837,
      "step": 42000
    },
    {
      "epoch": 2.31,
      "grad_norm": 0.8008145689964294,
      "learning_rate": 9.676846131215064e-05,
      "loss": 3.471,
      "step": 43000
    },
    {
      "epoch": 2.37,
      "grad_norm": 0.7869082689285278,
      "learning_rate": 9.647425713445132e-05,
      "loss": 3.4643,
      "step": 44000
    },
    {
      "epoch": 2.42,
      "grad_norm": 0.8000088930130005,
      "learning_rate": 9.618005295675198e-05,
      "loss": 3.4542,
      "step": 45000
    },
    {
      "epoch": 2.47,
      "grad_norm": 0.7834064960479736,
      "learning_rate": 9.588614298323037e-05,
      "loss": 3.4491,
      "step": 46000
    },
    {
      "epoch": 2.53,
      "grad_norm": 0.7652507424354553,
      "learning_rate": 9.559193880553105e-05,
      "loss": 3.4387,
      "step": 47000
    },
    {
      "epoch": 2.58,
      "grad_norm": 0.8015010356903076,
      "learning_rate": 9.529802883200942e-05,
      "loss": 3.4331,
      "step": 48000
    },
    {
      "epoch": 2.64,
      "grad_norm": 0.8380176424980164,
      "learning_rate": 9.500382465431009e-05,
      "loss": 3.4261,
      "step": 49000
    },
    {
      "epoch": 2.69,
      "grad_norm": 0.7744593024253845,
      "learning_rate": 9.470962047661077e-05,
      "loss": 3.4181,
      "step": 50000
    },
    {
      "epoch": 2.74,
      "grad_norm": 0.7821061015129089,
      "learning_rate": 9.441541629891145e-05,
      "loss": 3.412,
      "step": 51000
    },
    {
      "epoch": 2.8,
      "grad_norm": 0.7764970660209656,
      "learning_rate": 9.412150632538983e-05,
      "loss": 3.4048,
      "step": 52000
    },
    {
      "epoch": 2.85,
      "grad_norm": 0.7456904649734497,
      "learning_rate": 9.38273021476905e-05,
      "loss": 3.3998,
      "step": 53000
    },
    {
      "epoch": 2.9,
      "grad_norm": 0.7432152032852173,
      "learning_rate": 9.353339217416887e-05,
      "loss": 3.394,
      "step": 54000
    },
    {
      "epoch": 2.96,
      "grad_norm": 0.7415820956230164,
      "learning_rate": 9.323948220064726e-05,
      "loss": 3.3865,
      "step": 55000
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.3815021224736739,
      "eval_loss": 3.5883822441101074,
      "eval_runtime": 154.6982,
      "eval_samples_per_second": 374.413,
      "eval_steps_per_second": 5.857,
      "step": 55785
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.7807851433753967,
      "learning_rate": 9.294527802294794e-05,
      "loss": 3.3755,
      "step": 56000
    },
    {
      "epoch": 3.07,
      "grad_norm": 0.7891095876693726,
      "learning_rate": 9.26510738452486e-05,
      "loss": 3.3311,
      "step": 57000
    },
    {
      "epoch": 3.12,
      "grad_norm": 0.7911957502365112,
      "learning_rate": 9.235686966754929e-05,
      "loss": 3.3342,
      "step": 58000
    },
    {
      "epoch": 3.17,
      "grad_norm": 0.7487536072731018,
      "learning_rate": 9.206295969402766e-05,
      "loss": 3.3276,
      "step": 59000
    },
    {
      "epoch": 3.23,
      "grad_norm": 0.7860606908798218,
      "learning_rate": 9.176875551632832e-05,
      "loss": 3.3254,
      "step": 60000
    },
    {
      "epoch": 3.28,
      "grad_norm": 0.7669396996498108,
      "learning_rate": 9.147484554280671e-05,
      "loss": 3.3237,
      "step": 61000
    },
    {
      "epoch": 3.33,
      "grad_norm": 0.7570671439170837,
      "learning_rate": 9.118064136510739e-05,
      "loss": 3.3181,
      "step": 62000
    },
    {
      "epoch": 3.39,
      "grad_norm": 0.764175295829773,
      "learning_rate": 9.088643718740807e-05,
      "loss": 3.3194,
      "step": 63000
    },
    {
      "epoch": 3.44,
      "grad_norm": 0.723421573638916,
      "learning_rate": 9.059252721388644e-05,
      "loss": 3.3182,
      "step": 64000
    },
    {
      "epoch": 3.5,
      "grad_norm": 0.7632123231887817,
      "learning_rate": 9.029861724036481e-05,
      "loss": 3.3158,
      "step": 65000
    },
    {
      "epoch": 3.55,
      "grad_norm": 0.7372169494628906,
      "learning_rate": 9.00044130626655e-05,
      "loss": 3.3099,
      "step": 66000
    },
    {
      "epoch": 3.6,
      "grad_norm": 0.7337743639945984,
      "learning_rate": 8.971050308914388e-05,
      "loss": 3.3064,
      "step": 67000
    },
    {
      "epoch": 3.66,
      "grad_norm": 0.7390604019165039,
      "learning_rate": 8.941629891144455e-05,
      "loss": 3.3011,
      "step": 68000
    },
    {
      "epoch": 3.71,
      "grad_norm": 0.7338905334472656,
      "learning_rate": 8.912238893792292e-05,
      "loss": 3.3003,
      "step": 69000
    },
    {
      "epoch": 3.76,
      "grad_norm": 0.7679952383041382,
      "learning_rate": 8.88281847602236e-05,
      "loss": 3.2998,
      "step": 70000
    },
    {
      "epoch": 3.82,
      "grad_norm": 0.7354390025138855,
      "learning_rate": 8.853398058252428e-05,
      "loss": 3.2903,
      "step": 71000
    },
    {
      "epoch": 3.87,
      "grad_norm": 0.769103467464447,
      "learning_rate": 8.823977640482495e-05,
      "loss": 3.2916,
      "step": 72000
    },
    {
      "epoch": 3.93,
      "grad_norm": 0.7130835652351379,
      "learning_rate": 8.794616063548102e-05,
      "loss": 3.2913,
      "step": 73000
    },
    {
      "epoch": 3.98,
      "grad_norm": 0.7574236989021301,
      "learning_rate": 8.76519564577817e-05,
      "loss": 3.2906,
      "step": 74000
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.389078925897793,
      "eval_loss": 3.505403757095337,
      "eval_runtime": 154.7175,
      "eval_samples_per_second": 374.366,
      "eval_steps_per_second": 5.856,
      "step": 74380
    },
    {
      "epoch": 4.03,
      "grad_norm": 0.7582383155822754,
      "learning_rate": 8.735775228008238e-05,
      "loss": 3.2494,
      "step": 75000
    },
    {
      "epoch": 4.09,
      "grad_norm": 0.7709288597106934,
      "learning_rate": 8.706354810238306e-05,
      "loss": 3.2266,
      "step": 76000
    },
    {
      "epoch": 4.14,
      "grad_norm": 0.7542001605033875,
      "learning_rate": 8.676934392468373e-05,
      "loss": 3.2335,
      "step": 77000
    },
    {
      "epoch": 4.19,
      "grad_norm": 0.7459376454353333,
      "learning_rate": 8.647513974698441e-05,
      "loss": 3.2389,
      "step": 78000
    },
    {
      "epoch": 4.25,
      "grad_norm": 0.7790886163711548,
      "learning_rate": 8.61812297734628e-05,
      "loss": 3.2343,
      "step": 79000
    },
    {
      "epoch": 4.3,
      "grad_norm": 0.7570230960845947,
      "learning_rate": 8.588702559576346e-05,
      "loss": 3.236,
      "step": 80000
    },
    {
      "epoch": 4.36,
      "grad_norm": 0.7441847324371338,
      "learning_rate": 8.559311562224183e-05,
      "loss": 3.2345,
      "step": 81000
    },
    {
      "epoch": 4.41,
      "grad_norm": 0.7472123503684998,
      "learning_rate": 8.529891144454252e-05,
      "loss": 3.2338,
      "step": 82000
    },
    {
      "epoch": 4.46,
      "grad_norm": 0.8000940680503845,
      "learning_rate": 8.50050014710209e-05,
      "loss": 3.2329,
      "step": 83000
    },
    {
      "epoch": 4.52,
      "grad_norm": 0.763556957244873,
      "learning_rate": 8.471079729332157e-05,
      "loss": 3.2309,
      "step": 84000
    },
    {
      "epoch": 4.57,
      "grad_norm": 0.7414514422416687,
      "learning_rate": 8.441659311562225e-05,
      "loss": 3.2346,
      "step": 85000
    },
    {
      "epoch": 4.62,
      "grad_norm": 0.7634936571121216,
      "learning_rate": 8.412238893792293e-05,
      "loss": 3.228,
      "step": 86000
    },
    {
      "epoch": 4.68,
      "grad_norm": 0.7772691249847412,
      "learning_rate": 8.38284789644013e-05,
      "loss": 3.2234,
      "step": 87000
    },
    {
      "epoch": 4.73,
      "grad_norm": 0.7693132758140564,
      "learning_rate": 8.353427478670197e-05,
      "loss": 3.2299,
      "step": 88000
    },
    {
      "epoch": 4.79,
      "grad_norm": 0.7822285294532776,
      "learning_rate": 8.324007060900265e-05,
      "loss": 3.2258,
      "step": 89000
    },
    {
      "epoch": 4.84,
      "grad_norm": 0.7286272644996643,
      "learning_rate": 8.294616063548103e-05,
      "loss": 3.2252,
      "step": 90000
    },
    {
      "epoch": 4.89,
      "grad_norm": 0.7409440279006958,
      "learning_rate": 8.265195645778171e-05,
      "loss": 3.2183,
      "step": 91000
    },
    {
      "epoch": 4.95,
      "grad_norm": 0.7559691667556763,
      "learning_rate": 8.235775228008238e-05,
      "loss": 3.2202,
      "step": 92000
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.39429668627542264,
      "eval_loss": 3.4769208431243896,
      "eval_runtime": 155.051,
      "eval_samples_per_second": 373.561,
      "eval_steps_per_second": 5.843,
      "step": 92975
    },
    {
      "epoch": 5.0,
      "grad_norm": 0.7476261854171753,
      "learning_rate": 8.206384230656075e-05,
      "loss": 3.2131,
      "step": 93000
    },
    {
      "epoch": 5.06,
      "grad_norm": 0.7305045127868652,
      "learning_rate": 8.176993233303914e-05,
      "loss": 3.1643,
      "step": 94000
    },
    {
      "epoch": 5.11,
      "grad_norm": 0.758934497833252,
      "learning_rate": 8.147572815533982e-05,
      "loss": 3.1671,
      "step": 95000
    },
    {
      "epoch": 5.16,
      "grad_norm": 0.7448575496673584,
      "learning_rate": 8.118152397764049e-05,
      "loss": 3.1674,
      "step": 96000
    },
    {
      "epoch": 5.22,
      "grad_norm": 0.7979652881622314,
      "learning_rate": 8.088731979994117e-05,
      "loss": 3.1695,
      "step": 97000
    },
    {
      "epoch": 5.27,
      "grad_norm": 0.7594915628433228,
      "learning_rate": 8.059311562224185e-05,
      "loss": 3.1731,
      "step": 98000
    },
    {
      "epoch": 5.32,
      "grad_norm": 0.7524967789649963,
      "learning_rate": 8.02992056487202e-05,
      "loss": 3.1724,
      "step": 99000
    },
    {
      "epoch": 5.38,
      "grad_norm": 0.7075839042663574,
      "learning_rate": 8.000500147102089e-05,
      "loss": 3.1744,
      "step": 100000
    },
    {
      "epoch": 5.43,
      "grad_norm": 0.7107969522476196,
      "learning_rate": 7.971109149749927e-05,
      "loss": 3.1691,
      "step": 101000
    },
    {
      "epoch": 5.49,
      "grad_norm": 0.7338487505912781,
      "learning_rate": 7.941688731979995e-05,
      "loss": 3.1714,
      "step": 102000
    },
    {
      "epoch": 5.54,
      "grad_norm": 0.7563539743423462,
      "learning_rate": 7.912268314210062e-05,
      "loss": 3.1702,
      "step": 103000
    },
    {
      "epoch": 5.59,
      "grad_norm": 0.7359240055084229,
      "learning_rate": 7.882877316857899e-05,
      "loss": 3.1703,
      "step": 104000
    },
    {
      "epoch": 5.65,
      "grad_norm": 0.7448987364768982,
      "learning_rate": 7.853456899087967e-05,
      "loss": 3.172,
      "step": 105000
    },
    {
      "epoch": 5.7,
      "grad_norm": 0.7606428265571594,
      "learning_rate": 7.824036481318035e-05,
      "loss": 3.1726,
      "step": 106000
    },
    {
      "epoch": 5.75,
      "grad_norm": 0.7235444784164429,
      "learning_rate": 7.794616063548102e-05,
      "loss": 3.1733,
      "step": 107000
    },
    {
      "epoch": 5.81,
      "grad_norm": 0.7561464905738831,
      "learning_rate": 7.76522506619594e-05,
      "loss": 3.1696,
      "step": 108000
    },
    {
      "epoch": 5.86,
      "grad_norm": 0.7387487292289734,
      "learning_rate": 7.735804648426009e-05,
      "loss": 3.1705,
      "step": 109000
    },
    {
      "epoch": 5.92,
      "grad_norm": 0.7437654137611389,
      "learning_rate": 7.706443071491616e-05,
      "loss": 3.1642,
      "step": 110000
    },
    {
      "epoch": 5.97,
      "grad_norm": 0.7192903161048889,
      "learning_rate": 7.677022653721683e-05,
      "loss": 3.1699,
      "step": 111000
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.3977428802964349,
      "eval_loss": 3.4425718784332275,
      "eval_runtime": 154.7206,
      "eval_samples_per_second": 374.359,
      "eval_steps_per_second": 5.856,
      "step": 111570
    },
    {
      "epoch": 6.02,
      "grad_norm": 0.7664169073104858,
      "learning_rate": 7.647602235951751e-05,
      "loss": 3.1431,
      "step": 112000
    },
    {
      "epoch": 6.08,
      "grad_norm": 0.7941229343414307,
      "learning_rate": 7.618181818181819e-05,
      "loss": 3.1107,
      "step": 113000
    },
    {
      "epoch": 6.13,
      "grad_norm": 0.7529132962226868,
      "learning_rate": 7.588761400411886e-05,
      "loss": 3.1141,
      "step": 114000
    },
    {
      "epoch": 6.18,
      "grad_norm": 0.7258307933807373,
      "learning_rate": 7.559370403059724e-05,
      "loss": 3.1197,
      "step": 115000
    },
    {
      "epoch": 6.24,
      "grad_norm": 0.7766565680503845,
      "learning_rate": 7.529949985289791e-05,
      "loss": 3.1211,
      "step": 116000
    },
    {
      "epoch": 6.29,
      "grad_norm": 0.7603769302368164,
      "learning_rate": 7.5005884083554e-05,
      "loss": 3.1226,
      "step": 117000
    },
    {
      "epoch": 6.35,
      "grad_norm": 0.7358362674713135,
      "learning_rate": 7.471167990585468e-05,
      "loss": 3.1224,
      "step": 118000
    },
    {
      "epoch": 6.4,
      "grad_norm": 0.7426795363426208,
      "learning_rate": 7.441747572815534e-05,
      "loss": 3.1258,
      "step": 119000
    },
    {
      "epoch": 6.45,
      "grad_norm": 0.760974645614624,
      "learning_rate": 7.412327155045603e-05,
      "loss": 3.1284,
      "step": 120000
    },
    {
      "epoch": 6.51,
      "grad_norm": 0.7280532717704773,
      "learning_rate": 7.38293615769344e-05,
      "loss": 3.1269,
      "step": 121000
    },
    {
      "epoch": 6.56,
      "grad_norm": 0.7262037396430969,
      "learning_rate": 7.353515739923508e-05,
      "loss": 3.1293,
      "step": 122000
    },
    {
      "epoch": 6.61,
      "grad_norm": 0.7488662600517273,
      "learning_rate": 7.324124742571345e-05,
      "loss": 3.1244,
      "step": 123000
    },
    {
      "epoch": 6.67,
      "grad_norm": 0.7436386346817017,
      "learning_rate": 7.294704324801413e-05,
      "loss": 3.1273,
      "step": 124000
    },
    {
      "epoch": 6.72,
      "grad_norm": 0.7284784913063049,
      "learning_rate": 7.265283907031481e-05,
      "loss": 3.1301,
      "step": 125000
    },
    {
      "epoch": 6.78,
      "grad_norm": 0.7190147638320923,
      "learning_rate": 7.235863489261548e-05,
      "loss": 3.1281,
      "step": 126000
    },
    {
      "epoch": 6.83,
      "grad_norm": 0.7366654276847839,
      "learning_rate": 7.206472491909385e-05,
      "loss": 3.1286,
      "step": 127000
    },
    {
      "epoch": 6.88,
      "grad_norm": 0.7567622065544128,
      "learning_rate": 7.177081494557223e-05,
      "loss": 3.1262,
      "step": 128000
    },
    {
      "epoch": 6.94,
      "grad_norm": 0.7190493941307068,
      "learning_rate": 7.147661076787291e-05,
      "loss": 3.1309,
      "step": 129000
    },
    {
      "epoch": 6.99,
      "grad_norm": 0.7200164794921875,
      "learning_rate": 7.11824065901736e-05,
      "loss": 3.1232,
      "step": 130000
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.399363026131729,
      "eval_loss": 3.447763442993164,
      "eval_runtime": 155.0052,
      "eval_samples_per_second": 373.671,
      "eval_steps_per_second": 5.845,
      "step": 130165
    },
    {
      "epoch": 7.04,
      "grad_norm": 0.7389765977859497,
      "learning_rate": 7.088820241247426e-05,
      "loss": 3.0769,
      "step": 131000
    },
    {
      "epoch": 7.1,
      "grad_norm": 0.7703937888145447,
      "learning_rate": 7.059429243895263e-05,
      "loss": 3.0708,
      "step": 132000
    },
    {
      "epoch": 7.15,
      "grad_norm": 0.7399336099624634,
      "learning_rate": 7.030008826125331e-05,
      "loss": 3.0763,
      "step": 133000
    },
    {
      "epoch": 7.21,
      "grad_norm": 0.7627936005592346,
      "learning_rate": 7.000588408355398e-05,
      "loss": 3.0798,
      "step": 134000
    },
    {
      "epoch": 7.26,
      "grad_norm": 0.7476488351821899,
      "learning_rate": 6.971167990585466e-05,
      "loss": 3.0854,
      "step": 135000
    },
    {
      "epoch": 7.31,
      "grad_norm": 0.7593668103218079,
      "learning_rate": 6.941776993233305e-05,
      "loss": 3.0837,
      "step": 136000
    },
    {
      "epoch": 7.37,
      "grad_norm": 0.7753123044967651,
      "learning_rate": 6.912356575463373e-05,
      "loss": 3.0842,
      "step": 137000
    },
    {
      "epoch": 7.42,
      "grad_norm": 0.7626084089279175,
      "learning_rate": 6.882965578111209e-05,
      "loss": 3.0883,
      "step": 138000
    },
    {
      "epoch": 7.48,
      "grad_norm": 0.7539852857589722,
      "learning_rate": 6.853545160341277e-05,
      "loss": 3.0873,
      "step": 139000
    },
    {
      "epoch": 7.53,
      "grad_norm": 0.7760306000709534,
      "learning_rate": 6.824154162989115e-05,
      "loss": 3.0932,
      "step": 140000
    },
    {
      "epoch": 7.58,
      "grad_norm": 0.7804057598114014,
      "learning_rate": 6.794733745219183e-05,
      "loss": 3.0908,
      "step": 141000
    },
    {
      "epoch": 7.64,
      "grad_norm": 0.7768170833587646,
      "learning_rate": 6.76531332744925e-05,
      "loss": 3.0889,
      "step": 142000
    },
    {
      "epoch": 7.69,
      "grad_norm": 0.7796506881713867,
      "learning_rate": 6.735892909679318e-05,
      "loss": 3.0864,
      "step": 143000
    },
    {
      "epoch": 7.74,
      "grad_norm": 0.7837468981742859,
      "learning_rate": 6.706472491909386e-05,
      "loss": 3.0948,
      "step": 144000
    },
    {
      "epoch": 7.8,
      "grad_norm": 0.7560127377510071,
      "learning_rate": 6.677081494557223e-05,
      "loss": 3.0946,
      "step": 145000
    },
    {
      "epoch": 7.85,
      "grad_norm": 0.7457463145256042,
      "learning_rate": 6.64769049720506e-05,
      "loss": 3.092,
      "step": 146000
    },
    {
      "epoch": 7.91,
      "grad_norm": 0.7344046235084534,
      "learning_rate": 6.618270079435128e-05,
      "loss": 3.0904,
      "step": 147000
    },
    {
      "epoch": 7.96,
      "grad_norm": 0.7419469356536865,
      "learning_rate": 6.588849661665197e-05,
      "loss": 3.091,
      "step": 148000
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.4014489882469057,
      "eval_loss": 3.42431378364563,
      "eval_runtime": 154.6897,
      "eval_samples_per_second": 374.433,
      "eval_steps_per_second": 5.857,
      "step": 148760
    },
    {
      "epoch": 8.01,
      "grad_norm": 0.7680034041404724,
      "learning_rate": 6.559458664313034e-05,
      "loss": 3.0783,
      "step": 149000
    },
    {
      "epoch": 8.07,
      "grad_norm": 0.7723122239112854,
      "learning_rate": 6.5300382465431e-05,
      "loss": 3.0382,
      "step": 150000
    },
    {
      "epoch": 8.12,
      "grad_norm": 0.8067519068717957,
      "learning_rate": 6.500617828773169e-05,
      "loss": 3.0417,
      "step": 151000
    },
    {
      "epoch": 8.17,
      "grad_norm": 0.7817720770835876,
      "learning_rate": 6.471197411003237e-05,
      "loss": 3.0428,
      "step": 152000
    },
    {
      "epoch": 8.23,
      "grad_norm": 0.785193145275116,
      "learning_rate": 6.441806413651074e-05,
      "loss": 3.0471,
      "step": 153000
    },
    {
      "epoch": 8.28,
      "grad_norm": 0.7578865885734558,
      "learning_rate": 6.412385995881142e-05,
      "loss": 3.049,
      "step": 154000
    },
    {
      "epoch": 8.34,
      "grad_norm": 0.75961834192276,
      "learning_rate": 6.382994998528979e-05,
      "loss": 3.0465,
      "step": 155000
    },
    {
      "epoch": 8.39,
      "grad_norm": 0.7525945901870728,
      "learning_rate": 6.353574580759047e-05,
      "loss": 3.0529,
      "step": 156000
    },
    {
      "epoch": 8.44,
      "grad_norm": 0.7648568153381348,
      "learning_rate": 6.324183583406885e-05,
      "loss": 3.055,
      "step": 157000
    },
    {
      "epoch": 8.5,
      "grad_norm": 0.7817012071609497,
      "learning_rate": 6.294763165636952e-05,
      "loss": 3.059,
      "step": 158000
    },
    {
      "epoch": 8.55,
      "grad_norm": 0.7443410158157349,
      "learning_rate": 6.26534274786702e-05,
      "loss": 3.0597,
      "step": 159000
    },
    {
      "epoch": 8.6,
      "grad_norm": 0.7928184866905212,
      "learning_rate": 6.235951750514857e-05,
      "loss": 3.0572,
      "step": 160000
    },
    {
      "epoch": 8.66,
      "grad_norm": 0.7541539669036865,
      "learning_rate": 6.206531332744924e-05,
      "loss": 3.0582,
      "step": 161000
    },
    {
      "epoch": 8.71,
      "grad_norm": 0.7730678915977478,
      "learning_rate": 6.177110914974992e-05,
      "loss": 3.0563,
      "step": 162000
    },
    {
      "epoch": 8.77,
      "grad_norm": 0.7646264433860779,
      "learning_rate": 6.14769049720506e-05,
      "loss": 3.0607,
      "step": 163000
    },
    {
      "epoch": 8.82,
      "grad_norm": 0.7786790728569031,
      "learning_rate": 6.118270079435127e-05,
      "loss": 3.0632,
      "step": 164000
    },
    {
      "epoch": 8.87,
      "grad_norm": 0.7602713704109192,
      "learning_rate": 6.0888790820829656e-05,
      "loss": 3.0601,
      "step": 165000
    },
    {
      "epoch": 8.93,
      "grad_norm": 0.7597308158874512,
      "learning_rate": 6.059458664313034e-05,
      "loss": 3.0615,
      "step": 166000
    },
    {
      "epoch": 8.98,
      "grad_norm": 0.8023492693901062,
      "learning_rate": 6.030038246543102e-05,
      "loss": 3.0613,
      "step": 167000
    },
    {
      "epoch": 9.0,
      "eval_accuracy": 0.4030179440035626,
      "eval_loss": 3.4169299602508545,
      "eval_runtime": 154.797,
      "eval_samples_per_second": 374.174,
      "eval_steps_per_second": 5.853,
      "step": 167355
    },
    {
      "epoch": 9.03,
      "grad_norm": 0.7707279324531555,
      "learning_rate": 6.000676669608709e-05,
      "loss": 3.0235,
      "step": 168000
    },
    {
      "epoch": 9.09,
      "grad_norm": 0.8100076913833618,
      "learning_rate": 5.971256251838776e-05,
      "loss": 3.0067,
      "step": 169000
    },
    {
      "epoch": 9.14,
      "grad_norm": 0.761734127998352,
      "learning_rate": 5.941835834068844e-05,
      "loss": 3.0091,
      "step": 170000
    },
    {
      "epoch": 9.2,
      "grad_norm": 0.8010926842689514,
      "learning_rate": 5.912415416298912e-05,
      "loss": 3.0168,
      "step": 171000
    },
    {
      "epoch": 9.25,
      "grad_norm": 0.8410947322845459,
      "learning_rate": 5.88302441894675e-05,
      "loss": 3.014,
      "step": 172000
    },
    {
      "epoch": 9.3,
      "grad_norm": 0.7763057351112366,
      "learning_rate": 5.853633421594586e-05,
      "loss": 3.019,
      "step": 173000
    },
    {
      "epoch": 9.36,
      "grad_norm": 0.7941156625747681,
      "learning_rate": 5.8242130038246544e-05,
      "loss": 3.0273,
      "step": 174000
    },
    {
      "epoch": 9.41,
      "grad_norm": 0.821797251701355,
      "learning_rate": 5.7947925860547225e-05,
      "loss": 3.0265,
      "step": 175000
    },
    {
      "epoch": 9.46,
      "grad_norm": 0.7570799589157104,
      "learning_rate": 5.76540158870256e-05,
      "loss": 3.0298,
      "step": 176000
    },
    {
      "epoch": 9.52,
      "grad_norm": 0.7869284152984619,
      "learning_rate": 5.735981170932627e-05,
      "loss": 3.0312,
      "step": 177000
    },
    {
      "epoch": 9.57,
      "grad_norm": 0.8196205496788025,
      "learning_rate": 5.706560753162695e-05,
      "loss": 3.0273,
      "step": 178000
    },
    {
      "epoch": 9.63,
      "grad_norm": 0.7756869792938232,
      "learning_rate": 5.677169755810533e-05,
      "loss": 3.0253,
      "step": 179000
    },
    {
      "epoch": 9.68,
      "grad_norm": 0.8128073811531067,
      "learning_rate": 5.6477493380406e-05,
      "loss": 3.0288,
      "step": 180000
    },
    {
      "epoch": 9.73,
      "grad_norm": 0.7797793745994568,
      "learning_rate": 5.618328920270668e-05,
      "loss": 3.0322,
      "step": 181000
    },
    {
      "epoch": 9.79,
      "grad_norm": 0.789199948310852,
      "learning_rate": 5.588908502500736e-05,
      "loss": 3.036,
      "step": 182000
    },
    {
      "epoch": 9.84,
      "grad_norm": 0.7418550848960876,
      "learning_rate": 5.5595175051485736e-05,
      "loss": 3.0368,
      "step": 183000
    },
    {
      "epoch": 9.9,
      "grad_norm": 0.8195221424102783,
      "learning_rate": 5.5300970873786404e-05,
      "loss": 3.0334,
      "step": 184000
    },
    {
      "epoch": 9.95,
      "grad_norm": 0.7761098742485046,
      "learning_rate": 5.5006766696087085e-05,
      "loss": 3.0322,
      "step": 185000
    },
    {
      "epoch": 10.0,
      "eval_accuracy": 0.4050055566763182,
      "eval_loss": 3.4141781330108643,
      "eval_runtime": 154.9579,
      "eval_samples_per_second": 373.785,
      "eval_steps_per_second": 5.847,
      "step": 185950
    },
    {
      "epoch": 10.0,
      "grad_norm": 0.7680907249450684,
      "learning_rate": 5.471256251838777e-05,
      "loss": 3.0339,
      "step": 186000
    },
    {
      "epoch": 10.06,
      "grad_norm": 0.7804239392280579,
      "learning_rate": 5.441894674904384e-05,
      "loss": 2.9809,
      "step": 187000
    },
    {
      "epoch": 10.11,
      "grad_norm": 0.8002775311470032,
      "learning_rate": 5.4124742571344514e-05,
      "loss": 2.9844,
      "step": 188000
    },
    {
      "epoch": 10.16,
      "grad_norm": 0.8047283291816711,
      "learning_rate": 5.383083259782289e-05,
      "loss": 2.9912,
      "step": 189000
    },
    {
      "epoch": 10.22,
      "grad_norm": 0.8343998193740845,
      "learning_rate": 5.353692262430127e-05,
      "loss": 2.989,
      "step": 190000
    },
    {
      "epoch": 10.27,
      "grad_norm": 0.798310399055481,
      "learning_rate": 5.3242718446601943e-05,
      "loss": 2.9925,
      "step": 191000
    },
    {
      "epoch": 10.33,
      "grad_norm": 0.8054454326629639,
      "learning_rate": 5.294880847308033e-05,
      "loss": 2.9986,
      "step": 192000
    },
    {
      "epoch": 10.38,
      "grad_norm": 0.810845136642456,
      "learning_rate": 5.2654604295380995e-05,
      "loss": 2.9996,
      "step": 193000
    },
    {
      "epoch": 10.43,
      "grad_norm": 0.7934587597846985,
      "learning_rate": 5.2360400117681677e-05,
      "loss": 3.0,
      "step": 194000
    },
    {
      "epoch": 10.49,
      "grad_norm": 0.8145301938056946,
      "learning_rate": 5.206619593998235e-05,
      "loss": 2.9977,
      "step": 195000
    },
    {
      "epoch": 10.54,
      "grad_norm": 0.7856894135475159,
      "learning_rate": 5.177228596646072e-05,
      "loss": 3.0025,
      "step": 196000
    },
    {
      "epoch": 10.59,
      "grad_norm": 0.8008938431739807,
      "learning_rate": 5.14780817887614e-05,
      "loss": 3.0059,
      "step": 197000
    },
    {
      "epoch": 10.65,
      "grad_norm": 0.8000041246414185,
      "learning_rate": 5.1183877611062084e-05,
      "loss": 3.0025,
      "step": 198000
    },
    {
      "epoch": 10.7,
      "grad_norm": 0.8157011866569519,
      "learning_rate": 5.088967343336276e-05,
      "loss": 3.0018,
      "step": 199000
    },
    {
      "epoch": 10.76,
      "grad_norm": 0.7584720253944397,
      "learning_rate": 5.059576345984113e-05,
      "loss": 3.0051,
      "step": 200000
    },
    {
      "epoch": 10.81,
      "grad_norm": 0.7773544192314148,
      "learning_rate": 5.030155928214181e-05,
      "loss": 3.0052,
      "step": 201000
    },
    {
      "epoch": 10.86,
      "grad_norm": 0.7875356078147888,
      "learning_rate": 5.000764930862019e-05,
      "loss": 3.0137,
      "step": 202000
    },
    {
      "epoch": 10.92,
      "grad_norm": 0.7761446833610535,
      "learning_rate": 4.971344513092086e-05,
      "loss": 3.0143,
      "step": 203000
    },
    {
      "epoch": 10.97,
      "grad_norm": 0.8193157315254211,
      "learning_rate": 4.941924095322154e-05,
      "loss": 3.0107,
      "step": 204000
    },
    {
      "epoch": 11.0,
      "eval_accuracy": 0.40485742790545853,
      "eval_loss": 3.4051809310913086,
      "eval_runtime": 154.8665,
      "eval_samples_per_second": 374.006,
      "eval_steps_per_second": 5.85,
      "step": 204545
    },
    {
      "epoch": 11.02,
      "grad_norm": 0.8065283298492432,
      "learning_rate": 4.9125330979699914e-05,
      "loss": 2.9828,
      "step": 205000
    },
    {
      "epoch": 11.08,
      "grad_norm": 0.8200777769088745,
      "learning_rate": 4.8831126802000595e-05,
      "loss": 2.96,
      "step": 206000
    },
    {
      "epoch": 11.13,
      "grad_norm": 0.8049452900886536,
      "learning_rate": 4.8537216828478965e-05,
      "loss": 2.9624,
      "step": 207000
    },
    {
      "epoch": 11.19,
      "grad_norm": 0.8497307300567627,
      "learning_rate": 4.824301265077965e-05,
      "loss": 2.971,
      "step": 208000
    },
    {
      "epoch": 11.24,
      "grad_norm": 0.815825343132019,
      "learning_rate": 4.794880847308032e-05,
      "loss": 2.9686,
      "step": 209000
    },
    {
      "epoch": 11.29,
      "grad_norm": 0.8359244465827942,
      "learning_rate": 4.7654604295380996e-05,
      "loss": 2.972,
      "step": 210000
    },
    {
      "epoch": 11.35,
      "grad_norm": 0.804949164390564,
      "learning_rate": 4.736069432185937e-05,
      "loss": 2.9756,
      "step": 211000
    },
    {
      "epoch": 11.4,
      "grad_norm": 0.8300769925117493,
      "learning_rate": 4.706678434833775e-05,
      "loss": 2.9777,
      "step": 212000
    },
    {
      "epoch": 11.45,
      "grad_norm": 0.8034248352050781,
      "learning_rate": 4.6772580170638425e-05,
      "loss": 2.9736,
      "step": 213000
    },
    {
      "epoch": 11.51,
      "grad_norm": 0.7853664755821228,
      "learning_rate": 4.64783759929391e-05,
      "loss": 2.9769,
      "step": 214000
    },
    {
      "epoch": 11.56,
      "grad_norm": 0.8291251063346863,
      "learning_rate": 4.618417181523978e-05,
      "loss": 2.9748,
      "step": 215000
    },
    {
      "epoch": 11.62,
      "grad_norm": 0.8467937111854553,
      "learning_rate": 4.5890556045895854e-05,
      "loss": 2.9849,
      "step": 216000
    },
    {
      "epoch": 11.67,
      "grad_norm": 0.8191725015640259,
      "learning_rate": 4.559635186819653e-05,
      "loss": 2.983,
      "step": 217000
    },
    {
      "epoch": 11.72,
      "grad_norm": 0.7857011556625366,
      "learning_rate": 4.530214769049721e-05,
      "loss": 2.9797,
      "step": 218000
    },
    {
      "epoch": 11.78,
      "grad_norm": 0.8048397302627563,
      "learning_rate": 4.500823771697558e-05,
      "loss": 2.9844,
      "step": 219000
    },
    {
      "epoch": 11.83,
      "grad_norm": 0.816554069519043,
      "learning_rate": 4.471403353927626e-05,
      "loss": 2.986,
      "step": 220000
    },
    {
      "epoch": 11.88,
      "grad_norm": 0.8050833940505981,
      "learning_rate": 4.442012356575463e-05,
      "loss": 2.9887,
      "step": 221000
    },
    {
      "epoch": 11.94,
      "grad_norm": 0.8382700681686401,
      "learning_rate": 4.412621359223301e-05,
      "loss": 2.9892,
      "step": 222000
    },
    {
      "epoch": 11.99,
      "grad_norm": 0.8233689665794373,
      "learning_rate": 4.3832009414533684e-05,
      "loss": 2.9844,
      "step": 223000
    },
    {
      "epoch": 12.0,
      "eval_accuracy": 0.4053066510758616,
      "eval_loss": 3.4128060340881348,
      "eval_runtime": 155.24,
      "eval_samples_per_second": 373.106,
      "eval_steps_per_second": 5.836,
      "step": 223140
    },
    {
      "epoch": 12.05,
      "grad_norm": 0.8400596976280212,
      "learning_rate": 4.353809944101207e-05,
      "loss": 2.9383,
      "step": 224000
    },
    {
      "epoch": 12.1,
      "grad_norm": 0.8800355792045593,
      "learning_rate": 4.324389526331274e-05,
      "loss": 2.9441,
      "step": 225000
    },
    {
      "epoch": 12.15,
      "grad_norm": 0.8398945927619934,
      "learning_rate": 4.2949691085613417e-05,
      "loss": 2.9484,
      "step": 226000
    },
    {
      "epoch": 12.21,
      "grad_norm": 0.8549145460128784,
      "learning_rate": 4.265548690791409e-05,
      "loss": 2.9443,
      "step": 227000
    },
    {
      "epoch": 12.26,
      "grad_norm": 0.8680046796798706,
      "learning_rate": 4.2361282730214765e-05,
      "loss": 2.9485,
      "step": 228000
    },
    {
      "epoch": 12.32,
      "grad_norm": 0.8534417748451233,
      "learning_rate": 4.206737275669315e-05,
      "loss": 2.9479,
      "step": 229000
    },
    {
      "epoch": 12.37,
      "grad_norm": 0.8083067536354065,
      "learning_rate": 4.1773168578993824e-05,
      "loss": 2.9497,
      "step": 230000
    },
    {
      "epoch": 12.42,
      "grad_norm": 0.810694694519043,
      "learning_rate": 4.14789644012945e-05,
      "loss": 2.959,
      "step": 231000
    },
    {
      "epoch": 12.48,
      "grad_norm": 0.8079540133476257,
      "learning_rate": 4.1185054427772876e-05,
      "loss": 2.9501,
      "step": 232000
    },
    {
      "epoch": 12.53,
      "grad_norm": 0.8427674174308777,
      "learning_rate": 4.089085025007355e-05,
      "loss": 2.9553,
      "step": 233000
    },
    {
      "epoch": 12.58,
      "grad_norm": 0.8193045854568481,
      "learning_rate": 4.0596646072374225e-05,
      "loss": 2.9604,
      "step": 234000
    },
    {
      "epoch": 12.64,
      "grad_norm": 0.8305484652519226,
      "learning_rate": 4.030273609885261e-05,
      "loss": 2.9623,
      "step": 235000
    },
    {
      "epoch": 12.69,
      "grad_norm": 0.8170352578163147,
      "learning_rate": 4.0008826125330986e-05,
      "loss": 2.9616,
      "step": 236000
    },
    {
      "epoch": 12.75,
      "grad_norm": 0.8222061395645142,
      "learning_rate": 3.971462194763166e-05,
      "loss": 2.9659,
      "step": 237000
    },
    {
      "epoch": 12.8,
      "grad_norm": 0.8384312987327576,
      "learning_rate": 3.9420417769932335e-05,
      "loss": 2.9645,
      "step": 238000
    },
    {
      "epoch": 12.85,
      "grad_norm": 0.8427971005439758,
      "learning_rate": 3.9126213592233016e-05,
      "loss": 2.9628,
      "step": 239000
    },
    {
      "epoch": 12.91,
      "grad_norm": 0.8048679232597351,
      "learning_rate": 3.883200941453369e-05,
      "loss": 2.9675,
      "step": 240000
    },
    {
      "epoch": 12.96,
      "grad_norm": 0.8325705528259277,
      "learning_rate": 3.853809944101207e-05,
      "loss": 2.9671,
      "step": 241000
    },
    {
      "epoch": 13.0,
      "eval_accuracy": 0.4061510186590524,
      "eval_loss": 3.414961576461792,
      "eval_runtime": 154.7616,
      "eval_samples_per_second": 374.26,
      "eval_steps_per_second": 5.854,
      "step": 241735
    },
    {
      "epoch": 13.01,
      "grad_norm": 0.856779158115387,
      "learning_rate": 3.824418946749044e-05,
      "loss": 2.9534,
      "step": 242000
    },
    {
      "epoch": 13.07,
      "grad_norm": 0.8315408229827881,
      "learning_rate": 3.794998528979112e-05,
      "loss": 2.9192,
      "step": 243000
    },
    {
      "epoch": 13.12,
      "grad_norm": 0.8712647557258606,
      "learning_rate": 3.7655781112091794e-05,
      "loss": 2.9236,
      "step": 244000
    },
    {
      "epoch": 13.18,
      "grad_norm": 0.8575468063354492,
      "learning_rate": 3.7361576934392476e-05,
      "loss": 2.9209,
      "step": 245000
    },
    {
      "epoch": 13.23,
      "grad_norm": 0.848013699054718,
      "learning_rate": 3.7067666960870846e-05,
      "loss": 2.9224,
      "step": 246000
    },
    {
      "epoch": 13.28,
      "grad_norm": 0.8788365721702576,
      "learning_rate": 3.677346278317153e-05,
      "loss": 2.9313,
      "step": 247000
    },
    {
      "epoch": 13.34,
      "grad_norm": 0.8671311140060425,
      "learning_rate": 3.64792586054722e-05,
      "loss": 2.9354,
      "step": 248000
    },
    {
      "epoch": 13.39,
      "grad_norm": 0.8452686667442322,
      "learning_rate": 3.618534863195058e-05,
      "loss": 2.9335,
      "step": 249000
    },
    {
      "epoch": 13.44,
      "grad_norm": 0.8493407368659973,
      "learning_rate": 3.5891144454251254e-05,
      "loss": 2.9352,
      "step": 250000
    },
    {
      "epoch": 13.5,
      "grad_norm": 0.8630385398864746,
      "learning_rate": 3.559694027655193e-05,
      "loss": 2.9366,
      "step": 251000
    },
    {
      "epoch": 13.55,
      "grad_norm": 0.8659392595291138,
      "learning_rate": 3.5303030303030305e-05,
      "loss": 2.9396,
      "step": 252000
    },
    {
      "epoch": 13.61,
      "grad_norm": 0.8576540350914001,
      "learning_rate": 3.500912032950868e-05,
      "loss": 2.9371,
      "step": 253000
    },
    {
      "epoch": 13.66,
      "grad_norm": 0.839272141456604,
      "learning_rate": 3.471491615180936e-05,
      "loss": 2.9399,
      "step": 254000
    },
    {
      "epoch": 13.71,
      "grad_norm": 0.8816892504692078,
      "learning_rate": 3.4421006178287734e-05,
      "loss": 2.9414,
      "step": 255000
    },
    {
      "epoch": 13.77,
      "grad_norm": 0.8923898339271545,
      "learning_rate": 3.412680200058841e-05,
      "loss": 2.9429,
      "step": 256000
    },
    {
      "epoch": 13.82,
      "grad_norm": 0.84392249584198,
      "learning_rate": 3.383259782288909e-05,
      "loss": 2.9459,
      "step": 257000
    },
    {
      "epoch": 13.87,
      "grad_norm": 0.8665987849235535,
      "learning_rate": 3.3538393645189765e-05,
      "loss": 2.9467,
      "step": 258000
    },
    {
      "epoch": 13.93,
      "grad_norm": 0.8305437564849854,
      "learning_rate": 3.324448367166814e-05,
      "loss": 2.9475,
      "step": 259000
    },
    {
      "epoch": 13.98,
      "grad_norm": 0.8880412578582764,
      "learning_rate": 3.2950279493968816e-05,
      "loss": 2.9477,
      "step": 260000
    },
    {
      "epoch": 14.0,
      "eval_accuracy": 0.4062352605994869,
      "eval_loss": 3.417443037033081,
      "eval_runtime": 154.6446,
      "eval_samples_per_second": 374.543,
      "eval_steps_per_second": 5.859,
      "step": 260330
    },
    {
      "epoch": 14.04,
      "grad_norm": 0.8674964308738708,
      "learning_rate": 3.265607531626949e-05,
      "loss": 2.9136,
      "step": 261000
    },
    {
      "epoch": 14.09,
      "grad_norm": 0.8518526554107666,
      "learning_rate": 3.236216534274787e-05,
      "loss": 2.9045,
      "step": 262000
    },
    {
      "epoch": 14.14,
      "grad_norm": 0.8985535502433777,
      "learning_rate": 3.206796116504854e-05,
      "loss": 2.9094,
      "step": 263000
    },
    {
      "epoch": 14.2,
      "grad_norm": 0.9025124311447144,
      "learning_rate": 3.177405119152692e-05,
      "loss": 2.9072,
      "step": 264000
    },
    {
      "epoch": 14.25,
      "grad_norm": 0.866775631904602,
      "learning_rate": 3.1479847013827594e-05,
      "loss": 2.912,
      "step": 265000
    },
    {
      "epoch": 14.3,
      "grad_norm": 0.875120222568512,
      "learning_rate": 3.1185642836128276e-05,
      "loss": 2.9127,
      "step": 266000
    },
    {
      "epoch": 14.36,
      "grad_norm": 0.8747627139091492,
      "learning_rate": 3.089143865842895e-05,
      "loss": 2.9133,
      "step": 267000
    },
    {
      "epoch": 14.41,
      "grad_norm": 0.8658396005630493,
      "learning_rate": 3.059752868490733e-05,
      "loss": 2.9134,
      "step": 268000
    },
    {
      "epoch": 14.47,
      "grad_norm": 0.8496713638305664,
      "learning_rate": 3.0303324507208002e-05,
      "loss": 2.9196,
      "step": 269000
    },
    {
      "epoch": 14.52,
      "grad_norm": 0.8578941822052002,
      "learning_rate": 3.0009120329508683e-05,
      "loss": 2.92,
      "step": 270000
    },
    {
      "epoch": 14.57,
      "grad_norm": 0.8853529691696167,
      "learning_rate": 2.9715504560164753e-05,
      "loss": 2.9222,
      "step": 271000
    },
    {
      "epoch": 14.63,
      "grad_norm": 0.8526709079742432,
      "learning_rate": 2.942159458664313e-05,
      "loss": 2.9165,
      "step": 272000
    },
    {
      "epoch": 14.68,
      "grad_norm": 0.8756287693977356,
      "learning_rate": 2.912739040894381e-05,
      "loss": 2.926,
      "step": 273000
    },
    {
      "epoch": 14.74,
      "grad_norm": 0.8718807697296143,
      "learning_rate": 2.8833186231244486e-05,
      "loss": 2.924,
      "step": 274000
    },
    {
      "epoch": 14.79,
      "grad_norm": 0.874072790145874,
      "learning_rate": 2.853898205354516e-05,
      "loss": 2.9244,
      "step": 275000
    },
    {
      "epoch": 14.84,
      "grad_norm": 0.8762756586074829,
      "learning_rate": 2.824477787584584e-05,
      "loss": 2.9211,
      "step": 276000
    },
    {
      "epoch": 14.9,
      "grad_norm": 0.9252573251724243,
      "learning_rate": 2.7950867902324212e-05,
      "loss": 2.9225,
      "step": 277000
    },
    {
      "epoch": 14.95,
      "grad_norm": 0.8735837936401367,
      "learning_rate": 2.7656663724624893e-05,
      "loss": 2.9272,
      "step": 278000
    },
    {
      "epoch": 15.0,
      "eval_accuracy": 0.4066799828049704,
      "eval_loss": 3.427541494369507,
      "eval_runtime": 155.041,
      "eval_samples_per_second": 373.585,
      "eval_steps_per_second": 5.844,
      "step": 278925
    },
    {
      "epoch": 15.0,
      "grad_norm": 0.8828846216201782,
      "learning_rate": 2.7362459546925568e-05,
      "loss": 2.9242,
      "step": 279000
    },
    {
      "epoch": 15.06,
      "grad_norm": 0.8483282923698425,
      "learning_rate": 2.7068255369226242e-05,
      "loss": 2.8881,
      "step": 280000
    },
    {
      "epoch": 15.11,
      "grad_norm": 0.8667340278625488,
      "learning_rate": 2.6774051191526924e-05,
      "loss": 2.8852,
      "step": 281000
    },
    {
      "epoch": 15.17,
      "grad_norm": 0.8646563291549683,
      "learning_rate": 2.6480435422182997e-05,
      "loss": 2.8886,
      "step": 282000
    },
    {
      "epoch": 15.22,
      "grad_norm": 0.8643233776092529,
      "learning_rate": 2.618623124448367e-05,
      "loss": 2.8935,
      "step": 283000
    },
    {
      "epoch": 15.27,
      "grad_norm": 0.9026147127151489,
      "learning_rate": 2.5892027066784353e-05,
      "loss": 2.8939,
      "step": 284000
    },
    {
      "epoch": 15.33,
      "grad_norm": 0.8915610313415527,
      "learning_rate": 2.5597822889085027e-05,
      "loss": 2.8964,
      "step": 285000
    },
    {
      "epoch": 15.38,
      "grad_norm": 0.8528517484664917,
      "learning_rate": 2.53036187113857e-05,
      "loss": 2.8976,
      "step": 286000
    },
    {
      "epoch": 15.43,
      "grad_norm": 0.8766651749610901,
      "learning_rate": 2.500970873786408e-05,
      "loss": 2.9009,
      "step": 287000
    },
    {
      "epoch": 15.49,
      "grad_norm": 0.897165060043335,
      "learning_rate": 2.4715504560164757e-05,
      "loss": 2.9008,
      "step": 288000
    },
    {
      "epoch": 15.54,
      "grad_norm": 0.8595399260520935,
      "learning_rate": 2.442159458664313e-05,
      "loss": 2.9063,
      "step": 289000
    },
    {
      "epoch": 15.6,
      "grad_norm": 0.8562242984771729,
      "learning_rate": 2.412739040894381e-05,
      "loss": 2.8988,
      "step": 290000
    },
    {
      "epoch": 15.65,
      "grad_norm": 0.8943966627120972,
      "learning_rate": 2.3833186231244483e-05,
      "loss": 2.905,
      "step": 291000
    },
    {
      "epoch": 15.7,
      "grad_norm": 0.8769930601119995,
      "learning_rate": 2.353927625772286e-05,
      "loss": 2.9032,
      "step": 292000
    },
    {
      "epoch": 15.76,
      "grad_norm": 0.8928766250610352,
      "learning_rate": 2.3245072080023535e-05,
      "loss": 2.907,
      "step": 293000
    },
    {
      "epoch": 15.81,
      "grad_norm": 0.8999136686325073,
      "learning_rate": 2.2950867902324213e-05,
      "loss": 2.9099,
      "step": 294000
    },
    {
      "epoch": 15.86,
      "grad_norm": 0.8893632888793945,
      "learning_rate": 2.265695792880259e-05,
      "loss": 2.9039,
      "step": 295000
    },
    {
      "epoch": 15.92,
      "grad_norm": 0.9111586213111877,
      "learning_rate": 2.2362753751103264e-05,
      "loss": 2.9071,
      "step": 296000
    },
    {
      "epoch": 15.97,
      "grad_norm": 0.8878467082977295,
      "learning_rate": 2.2068549573403942e-05,
      "loss": 2.9088,
      "step": 297000
    },
    {
      "epoch": 16.0,
      "eval_accuracy": 0.4068374493985737,
      "eval_loss": 3.427147626876831,
      "eval_runtime": 154.7655,
      "eval_samples_per_second": 374.25,
      "eval_steps_per_second": 5.854,
      "step": 297520
    },
    {
      "epoch": 16.03,
      "grad_norm": 0.9069207310676575,
      "learning_rate": 2.177434539570462e-05,
      "loss": 2.8909,
      "step": 298000
    },
    {
      "epoch": 16.08,
      "grad_norm": 0.8874634504318237,
      "learning_rate": 2.1480435422182994e-05,
      "loss": 2.8723,
      "step": 299000
    },
    {
      "epoch": 16.13,
      "grad_norm": 0.8833279609680176,
      "learning_rate": 2.1186231244483672e-05,
      "loss": 2.876,
      "step": 300000
    },
    {
      "epoch": 16.19,
      "grad_norm": 0.9021320939064026,
      "learning_rate": 2.089202706678435e-05,
      "loss": 2.8793,
      "step": 301000
    },
    {
      "epoch": 16.24,
      "grad_norm": 0.8948860764503479,
      "learning_rate": 2.0598117093262727e-05,
      "loss": 2.8789,
      "step": 302000
    },
    {
      "epoch": 16.29,
      "grad_norm": 0.8972275257110596,
      "learning_rate": 2.03039129155634e-05,
      "loss": 2.8778,
      "step": 303000
    },
    {
      "epoch": 16.35,
      "grad_norm": 0.8987729549407959,
      "learning_rate": 2.0009708737864076e-05,
      "loss": 2.8827,
      "step": 304000
    },
    {
      "epoch": 16.4,
      "grad_norm": 0.8773179650306702,
      "learning_rate": 1.9715798764342457e-05,
      "loss": 2.8866,
      "step": 305000
    },
    {
      "epoch": 16.46,
      "grad_norm": 0.9324170351028442,
      "learning_rate": 1.942159458664313e-05,
      "loss": 2.8842,
      "step": 306000
    },
    {
      "epoch": 16.51,
      "grad_norm": 0.9289742708206177,
      "learning_rate": 1.9127390408943806e-05,
      "loss": 2.8813,
      "step": 307000
    },
    {
      "epoch": 16.56,
      "grad_norm": 0.9093849062919617,
      "learning_rate": 1.8833480435422186e-05,
      "loss": 2.8824,
      "step": 308000
    },
    {
      "epoch": 16.62,
      "grad_norm": 0.9236767888069153,
      "learning_rate": 1.853927625772286e-05,
      "loss": 2.8845,
      "step": 309000
    },
    {
      "epoch": 16.67,
      "grad_norm": 0.8720325827598572,
      "learning_rate": 1.8245366284201238e-05,
      "loss": 2.8859,
      "step": 310000
    },
    {
      "epoch": 16.72,
      "grad_norm": 0.9025481939315796,
      "learning_rate": 1.7951456310679612e-05,
      "loss": 2.8921,
      "step": 311000
    },
    {
      "epoch": 16.78,
      "grad_norm": 0.9536554217338562,
      "learning_rate": 1.765725213298029e-05,
      "loss": 2.8898,
      "step": 312000
    },
    {
      "epoch": 16.83,
      "grad_norm": 0.909812867641449,
      "learning_rate": 1.7363047955280968e-05,
      "loss": 2.8901,
      "step": 313000
    },
    {
      "epoch": 16.89,
      "grad_norm": 0.9195666313171387,
      "learning_rate": 1.7068843777581642e-05,
      "loss": 2.8897,
      "step": 314000
    },
    {
      "epoch": 16.94,
      "grad_norm": 0.8690800666809082,
      "learning_rate": 1.677463959988232e-05,
      "loss": 2.888,
      "step": 315000
    },
    {
      "epoch": 16.99,
      "grad_norm": 0.8886986374855042,
      "learning_rate": 1.6480435422182998e-05,
      "loss": 2.8915,
      "step": 316000
    },
    {
      "epoch": 17.0,
      "eval_accuracy": 0.4071312885113811,
      "eval_loss": 3.424515962600708,
      "eval_runtime": 155.1079,
      "eval_samples_per_second": 373.424,
      "eval_steps_per_second": 5.841,
      "step": 316115
    },
    {
      "epoch": 17.05,
      "grad_norm": 0.9326255321502686,
      "learning_rate": 1.6186525448661372e-05,
      "loss": 2.8626,
      "step": 317000
    },
    {
      "epoch": 17.1,
      "grad_norm": 0.9530662298202515,
      "learning_rate": 1.5892615475139746e-05,
      "loss": 2.8605,
      "step": 318000
    },
    {
      "epoch": 17.16,
      "grad_norm": 0.9299532771110535,
      "learning_rate": 1.5598411297440423e-05,
      "loss": 2.8598,
      "step": 319000
    },
    {
      "epoch": 17.21,
      "grad_norm": 0.9452057480812073,
      "learning_rate": 1.53042071197411e-05,
      "loss": 2.8637,
      "step": 320000
    },
    {
      "epoch": 17.26,
      "grad_norm": 0.9302668571472168,
      "learning_rate": 1.5010591350397176e-05,
      "loss": 2.8636,
      "step": 321000
    },
    {
      "epoch": 17.32,
      "grad_norm": 0.9411625266075134,
      "learning_rate": 1.4716387172697854e-05,
      "loss": 2.8669,
      "step": 322000
    },
    {
      "epoch": 17.37,
      "grad_norm": 0.9255250096321106,
      "learning_rate": 1.4422182994998529e-05,
      "loss": 2.8643,
      "step": 323000
    },
    {
      "epoch": 17.42,
      "grad_norm": 0.9399036765098572,
      "learning_rate": 1.4127978817299207e-05,
      "loss": 2.8686,
      "step": 324000
    },
    {
      "epoch": 17.48,
      "grad_norm": 0.9294721484184265,
      "learning_rate": 1.3833774639599883e-05,
      "loss": 2.8693,
      "step": 325000
    },
    {
      "epoch": 17.53,
      "grad_norm": 0.9378616213798523,
      "learning_rate": 1.3539864666078258e-05,
      "loss": 2.8689,
      "step": 326000
    },
    {
      "epoch": 17.59,
      "grad_norm": 0.9260066747665405,
      "learning_rate": 1.3245954692556634e-05,
      "loss": 2.8697,
      "step": 327000
    },
    {
      "epoch": 17.64,
      "grad_norm": 0.9042669534683228,
      "learning_rate": 1.2951750514857312e-05,
      "loss": 2.8717,
      "step": 328000
    },
    {
      "epoch": 17.69,
      "grad_norm": 0.9497202634811401,
      "learning_rate": 1.2657546337157988e-05,
      "loss": 2.8715,
      "step": 329000
    },
    {
      "epoch": 17.75,
      "grad_norm": 0.9359462857246399,
      "learning_rate": 1.2363342159458664e-05,
      "loss": 2.8724,
      "step": 330000
    },
    {
      "epoch": 17.8,
      "grad_norm": 0.9586021900177002,
      "learning_rate": 1.2069432185937041e-05,
      "loss": 2.8724,
      "step": 331000
    },
    {
      "epoch": 17.85,
      "grad_norm": 0.92441725730896,
      "learning_rate": 1.1775228008237718e-05,
      "loss": 2.8739,
      "step": 332000
    },
    {
      "epoch": 17.91,
      "grad_norm": 0.9491459727287292,
      "learning_rate": 1.1481023830538394e-05,
      "loss": 2.8785,
      "step": 333000
    },
    {
      "epoch": 17.96,
      "grad_norm": 0.9118024110794067,
      "learning_rate": 1.1187113857016771e-05,
      "loss": 2.872,
      "step": 334000
    },
    {
      "epoch": 18.0,
      "eval_accuracy": 0.4070309908901142,
      "eval_loss": 3.426220417022705,
      "eval_runtime": 155.4694,
      "eval_samples_per_second": 372.556,
      "eval_steps_per_second": 5.828,
      "step": 334710
    },
    {
      "epoch": 18.02,
      "grad_norm": 0.9377329349517822,
      "learning_rate": 1.0892909679317447e-05,
      "loss": 2.8661,
      "step": 335000
    },
    {
      "epoch": 18.07,
      "grad_norm": 0.9395567178726196,
      "learning_rate": 1.0598705501618123e-05,
      "loss": 2.8488,
      "step": 336000
    },
    {
      "epoch": 18.12,
      "grad_norm": 0.9508826732635498,
      "learning_rate": 1.0304795528096499e-05,
      "loss": 2.8486,
      "step": 337000
    },
    {
      "epoch": 18.18,
      "grad_norm": 0.9117801785469055,
      "learning_rate": 1.0010591350397177e-05,
      "loss": 2.8487,
      "step": 338000
    },
    {
      "epoch": 18.23,
      "grad_norm": 0.932178795337677,
      "learning_rate": 9.716681376875552e-06,
      "loss": 2.8528,
      "step": 339000
    },
    {
      "epoch": 18.28,
      "grad_norm": 0.9204062223434448,
      "learning_rate": 9.422477199176229e-06,
      "loss": 2.8579,
      "step": 340000
    },
    {
      "epoch": 18.34,
      "grad_norm": 0.9516026377677917,
      "learning_rate": 9.128273021476906e-06,
      "loss": 2.8507,
      "step": 341000
    },
    {
      "epoch": 18.39,
      "grad_norm": 0.9432997703552246,
      "learning_rate": 8.834068843777583e-06,
      "loss": 2.8571,
      "step": 342000
    },
    {
      "epoch": 18.45,
      "grad_norm": 0.9799352288246155,
      "learning_rate": 8.539864666078259e-06,
      "loss": 2.8517,
      "step": 343000
    },
    {
      "epoch": 18.5,
      "grad_norm": 0.9541887044906616,
      "learning_rate": 8.245954692556634e-06,
      "loss": 2.8557,
      "step": 344000
    },
    {
      "epoch": 18.55,
      "grad_norm": 0.945749044418335,
      "learning_rate": 7.951750514857312e-06,
      "loss": 2.86,
      "step": 345000
    },
    {
      "epoch": 18.61,
      "grad_norm": 0.9562386274337769,
      "learning_rate": 7.657546337157988e-06,
      "loss": 2.8559,
      "step": 346000
    },
    {
      "epoch": 18.66,
      "grad_norm": 0.9685288071632385,
      "learning_rate": 7.363930567814063e-06,
      "loss": 2.8546,
      "step": 347000
    },
    {
      "epoch": 18.71,
      "grad_norm": 0.9271679520606995,
      "learning_rate": 7.06972639011474e-06,
      "loss": 2.8533,
      "step": 348000
    },
    {
      "epoch": 18.77,
      "grad_norm": 0.9129212498664856,
      "learning_rate": 6.775522212415417e-06,
      "loss": 2.8612,
      "step": 349000
    },
    {
      "epoch": 18.82,
      "grad_norm": 0.9596994519233704,
      "learning_rate": 6.481318034716093e-06,
      "loss": 2.8562,
      "step": 350000
    },
    {
      "epoch": 18.88,
      "grad_norm": 0.9252175092697144,
      "learning_rate": 6.18711385701677e-06,
      "loss": 2.8577,
      "step": 351000
    },
    {
      "epoch": 18.93,
      "grad_norm": 0.9257832169532776,
      "learning_rate": 5.893203883495145e-06,
      "loss": 2.8585,
      "step": 352000
    },
    {
      "epoch": 18.98,
      "grad_norm": 0.9196575880050659,
      "learning_rate": 5.598999705795822e-06,
      "loss": 2.8514,
      "step": 353000
    },
    {
      "epoch": 19.0,
      "eval_accuracy": 0.4072752522102257,
      "eval_loss": 3.4321706295013428,
      "eval_runtime": 155.3795,
      "eval_samples_per_second": 372.771,
      "eval_steps_per_second": 5.831,
      "step": 353305
    },
    {
      "epoch": 19.04,
      "grad_norm": 0.9688698649406433,
      "learning_rate": 5.305089732274199e-06,
      "loss": 2.8452,
      "step": 354000
    },
    {
      "epoch": 19.09,
      "grad_norm": 0.9426762461662292,
      "learning_rate": 5.010885554574876e-06,
      "loss": 2.8401,
      "step": 355000
    },
    {
      "epoch": 19.14,
      "grad_norm": 0.9447112679481506,
      "learning_rate": 4.716975581053251e-06,
      "loss": 2.839,
      "step": 356000
    },
    {
      "epoch": 19.2,
      "grad_norm": 0.9574055671691895,
      "learning_rate": 4.423065607531627e-06,
      "loss": 2.8447,
      "step": 357000
    },
    {
      "epoch": 19.25,
      "grad_norm": 0.9574621319770813,
      "learning_rate": 4.128861429832304e-06,
      "loss": 2.8391,
      "step": 358000
    },
    {
      "epoch": 19.31,
      "grad_norm": 0.9345229268074036,
      "learning_rate": 3.83465725213298e-06,
      "loss": 2.8444,
      "step": 359000
    },
    {
      "epoch": 19.36,
      "grad_norm": 0.9420320391654968,
      "learning_rate": 3.5404530744336568e-06,
      "loss": 2.8459,
      "step": 360000
    },
    {
      "epoch": 19.41,
      "grad_norm": 0.9578850865364075,
      "learning_rate": 3.2465431009120332e-06,
      "loss": 2.8409,
      "step": 361000
    },
    {
      "epoch": 19.47,
      "grad_norm": 0.963036060333252,
      "learning_rate": 2.95233892321271e-06,
      "loss": 2.8392,
      "step": 362000
    },
    {
      "epoch": 19.52,
      "grad_norm": 0.9696721434593201,
      "learning_rate": 2.6581347455133864e-06,
      "loss": 2.8451,
      "step": 363000
    },
    {
      "epoch": 19.58,
      "grad_norm": 0.9655833840370178,
      "learning_rate": 2.3642247719917624e-06,
      "loss": 2.8446,
      "step": 364000
    },
    {
      "epoch": 19.63,
      "grad_norm": 0.9735895395278931,
      "learning_rate": 2.070314798470138e-06,
      "loss": 2.8436,
      "step": 365000
    },
    {
      "epoch": 19.68,
      "grad_norm": 0.9257269501686096,
      "learning_rate": 1.7761106207708148e-06,
      "loss": 2.8413,
      "step": 366000
    },
    {
      "epoch": 19.74,
      "grad_norm": 0.9528219103813171,
      "learning_rate": 1.482200647249191e-06,
      "loss": 2.8403,
      "step": 367000
    },
    {
      "epoch": 19.79,
      "grad_norm": 0.9575457572937012,
      "learning_rate": 1.1879964695498676e-06,
      "loss": 2.8421,
      "step": 368000
    },
    {
      "epoch": 19.84,
      "grad_norm": 0.9463661909103394,
      "learning_rate": 8.937922918505443e-07,
      "loss": 2.8405,
      "step": 369000
    },
    {
      "epoch": 19.9,
      "grad_norm": 0.9508777856826782,
      "learning_rate": 5.99588114151221e-07,
      "loss": 2.842,
      "step": 370000
    },
    {
      "epoch": 19.95,
      "grad_norm": 0.93899005651474,
      "learning_rate": 3.0567814062959694e-07,
      "loss": 2.8467,
      "step": 371000
    },
    {
      "epoch": 20.0,
      "eval_accuracy": 0.40723568402608223,
      "eval_loss": 3.4373319149017334,
      "eval_runtime": 154.9871,
      "eval_samples_per_second": 373.715,
      "eval_steps_per_second": 5.846,
      "step": 371900
    },
    {
      "epoch": 20.0,
      "step": 371900,
      "total_flos": 1.5669257538816e+18,
      "train_loss": 3.1537178330448374,
      "train_runtime": 81513.9803,
      "train_samples_per_second": 145.996,
      "train_steps_per_second": 4.562
    }
  ],
  "logging_steps": 1000,
  "max_steps": 371900,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 20,
  "save_steps": 5000,
  "total_flos": 1.5669257538816e+18,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}