{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 70.0,
  "eval_steps": 500,
  "global_step": 91910,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.38,
      "learning_rate": 0.004999955274106163,
      "loss": 9.6382,
      "step": 500
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.004999821098024974,
      "loss": 9.5189,
      "step": 1000
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.00022625107915810782,
      "eval_loss": 9.40969181060791,
      "eval_runtime": 6168.4952,
      "eval_samples_per_second": 27.228,
      "eval_steps_per_second": 1.702,
      "step": 1313
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.0049995974765573515,
      "loss": 9.363,
      "step": 1500
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.00499928441770463,
      "loss": 9.2724,
      "step": 2000
    },
    {
      "epoch": 1.9,
      "learning_rate": 0.00499888193266828,
      "loss": 9.1635,
      "step": 2500
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.0006430293828704117,
      "eval_loss": 9.041797637939453,
      "eval_runtime": 4719.6922,
      "eval_samples_per_second": 35.586,
      "eval_steps_per_second": 2.224,
      "step": 2626
    },
    {
      "epoch": 2.28,
      "learning_rate": 0.004998390035849502,
      "loss": 8.8459,
      "step": 3000
    },
    {
      "epoch": 2.67,
      "learning_rate": 0.0049978087448487194,
      "loss": 8.3432,
      "step": 3500
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.006466017683308029,
      "eval_loss": 7.845377445220947,
      "eval_runtime": 2624.5595,
      "eval_samples_per_second": 63.994,
      "eval_steps_per_second": 4.0,
      "step": 3939
    },
    {
      "epoch": 3.05,
      "learning_rate": 0.004997138080464938,
      "loss": 7.8952,
      "step": 4000
    },
    {
      "epoch": 3.43,
      "learning_rate": 0.004996378066695007,
      "loss": 7.4467,
      "step": 4500
    },
    {
      "epoch": 3.81,
      "learning_rate": 0.004995528730732766,
      "loss": 6.9913,
      "step": 5000
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.04887023309815129,
      "eval_loss": 6.311844825744629,
      "eval_runtime": 2630.4579,
      "eval_samples_per_second": 63.85,
      "eval_steps_per_second": 3.991,
      "step": 5252
    },
    {
      "epoch": 4.19,
      "learning_rate": 0.0049945901029680604,
      "loss": 6.4357,
      "step": 5500
    },
    {
      "epoch": 4.57,
      "learning_rate": 0.004993562216985665,
      "loss": 5.9619,
      "step": 6000
    },
    {
      "epoch": 4.95,
      "learning_rate": 0.004992445109564073,
      "loss": 5.5048,
      "step": 6500
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.1423000208389152,
      "eval_loss": 4.949326515197754,
      "eval_runtime": 2650.8854,
      "eval_samples_per_second": 63.358,
      "eval_steps_per_second": 3.96,
      "step": 6565
    },
    {
      "epoch": 5.33,
      "learning_rate": 0.004991238820674189,
      "loss": 4.982,
      "step": 7000
    },
    {
      "epoch": 5.71,
      "learning_rate": 0.00498994339347789,
      "loss": 4.6895,
      "step": 7500
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.24502396475246346,
      "eval_loss": 3.960129499435425,
      "eval_runtime": 2617.6761,
      "eval_samples_per_second": 64.162,
      "eval_steps_per_second": 4.01,
      "step": 7878
    },
    {
      "epoch": 6.09,
      "learning_rate": 0.004988558874326489,
      "loss": 4.3479,
      "step": 8000
    },
    {
      "epoch": 6.47,
      "learning_rate": 0.004987085312759071,
      "loss": 3.9972,
      "step": 8500
    },
    {
      "epoch": 6.85,
      "learning_rate": 0.004985522761500722,
      "loss": 3.8881,
      "step": 9000
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.31358994968890475,
      "eval_loss": 3.4186415672302246,
      "eval_runtime": 2616.4405,
      "eval_samples_per_second": 64.192,
      "eval_steps_per_second": 4.012,
      "step": 9191
    },
    {
      "epoch": 7.24,
      "learning_rate": 0.004983871276460643,
      "loss": 3.6025,
      "step": 9500
    },
    {
      "epoch": 7.62,
      "learning_rate": 0.0049821309167301505,
      "loss": 3.4698,
      "step": 10000
    },
    {
      "epoch": 8.0,
      "learning_rate": 0.004980301744580561,
      "loss": 3.391,
      "step": 10500
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.37655919740406657,
      "eval_loss": 2.9798192977905273,
      "eval_runtime": 2644.3802,
      "eval_samples_per_second": 63.514,
      "eval_steps_per_second": 3.97,
      "step": 10504
    },
    {
      "epoch": 8.38,
      "learning_rate": 0.004978383825460959,
      "loss": 3.1113,
      "step": 11000
    },
    {
      "epoch": 8.76,
      "learning_rate": 0.004976377227995865,
      "loss": 3.0887,
      "step": 11500
    },
    {
      "epoch": 9.0,
      "eval_accuracy": 0.422124973951356,
      "eval_loss": 2.705430030822754,
      "eval_runtime": 2622.7588,
      "eval_samples_per_second": 64.038,
      "eval_steps_per_second": 4.003,
      "step": 11817
    },
    {
      "epoch": 9.14,
      "learning_rate": 0.00497428202398277,
      "loss": 2.925,
      "step": 12000
    },
    {
      "epoch": 9.52,
      "learning_rate": 0.004972098288389572,
      "loss": 2.7977,
      "step": 12500
    },
    {
      "epoch": 9.9,
      "learning_rate": 0.004969826099351891,
      "loss": 2.7935,
      "step": 13000
    },
    {
      "epoch": 10.0,
      "eval_accuracy": 0.45515167753267244,
      "eval_loss": 2.501314401626587,
      "eval_runtime": 2679.1466,
      "eval_samples_per_second": 62.69,
      "eval_steps_per_second": 3.918,
      "step": 13130
    },
    {
      "epoch": 10.28,
      "learning_rate": 0.004967465538170278,
      "loss": 2.5817,
      "step": 13500
    },
    {
      "epoch": 10.66,
      "learning_rate": 0.004965016689307298,
      "loss": 2.5629,
      "step": 14000
    },
    {
      "epoch": 11.0,
      "eval_accuracy": 0.4803846268345688,
      "eval_loss": 2.3581297397613525,
      "eval_runtime": 2615.9499,
      "eval_samples_per_second": 64.204,
      "eval_steps_per_second": 4.013,
      "step": 14443
    },
    {
      "epoch": 11.04,
      "learning_rate": 0.004962479640384515,
      "loss": 2.5549,
      "step": 14500
    },
    {
      "epoch": 11.42,
      "learning_rate": 0.004959854482179353,
      "loss": 2.3342,
      "step": 15000
    },
    {
      "epoch": 11.81,
      "learning_rate": 0.004957141308621851,
      "loss": 2.3777,
      "step": 15500
    },
    {
      "epoch": 12.0,
      "eval_accuracy": 0.48086689887172157,
      "eval_loss": 2.3543388843536377,
      "eval_runtime": 2613.6118,
      "eval_samples_per_second": 64.262,
      "eval_steps_per_second": 4.017,
      "step": 15756
    },
    {
      "epoch": 12.19,
      "learning_rate": 0.004954340216791298,
      "loss": 2.1546,
      "step": 16000
    },
    {
      "epoch": 12.57,
      "learning_rate": 0.004951451306912763,
      "loss": 2.2162,
      "step": 16500
    },
    {
      "epoch": 12.95,
      "learning_rate": 0.0049484746823535075,
      "loss": 2.2264,
      "step": 17000
    },
    {
      "epoch": 13.0,
      "eval_accuracy": 0.5178887201929088,
      "eval_loss": 2.163175344467163,
      "eval_runtime": 5970.7626,
      "eval_samples_per_second": 28.13,
      "eval_steps_per_second": 1.758,
      "step": 17069
    },
    {
      "epoch": 13.33,
      "learning_rate": 0.004945410449619286,
      "loss": 2.035,
      "step": 17500
    },
    {
      "epoch": 13.71,
      "learning_rate": 0.0049422587183505385,
      "loss": 2.0932,
      "step": 18000
    },
    {
      "epoch": 14.0,
      "eval_accuracy": 0.5219076538358489,
      "eval_loss": 2.136173725128174,
      "eval_runtime": 2759.3303,
      "eval_samples_per_second": 60.868,
      "eval_steps_per_second": 3.805,
      "step": 18382
    },
    {
      "epoch": 14.09,
      "learning_rate": 0.004939019601318461,
      "loss": 2.0545,
      "step": 18500
    },
    {
      "epoch": 14.47,
      "learning_rate": 0.004935693214420979,
      "loss": 1.9224,
      "step": 19000
    },
    {
      "epoch": 14.85,
      "learning_rate": 0.004932279676678594,
      "loss": 1.9667,
      "step": 19500
    },
    {
      "epoch": 15.0,
      "eval_accuracy": 0.5591200023815903,
      "eval_loss": 1.956689715385437,
      "eval_runtime": 2781.1993,
      "eval_samples_per_second": 60.389,
      "eval_steps_per_second": 3.775,
      "step": 19695
    },
    {
      "epoch": 15.23,
      "learning_rate": 0.004928779110230126,
      "loss": 1.8635,
      "step": 20000
    },
    {
      "epoch": 15.61,
      "learning_rate": 0.004925191640328348,
      "loss": 1.8396,
      "step": 20500
    },
    {
      "epoch": 15.99,
      "learning_rate": 0.004921517395335496,
      "loss": 1.8788,
      "step": 21000
    },
    {
      "epoch": 16.0,
      "eval_accuracy": 0.5609538269179244,
      "eval_loss": 1.9347097873687744,
      "eval_runtime": 2696.6196,
      "eval_samples_per_second": 62.284,
      "eval_steps_per_second": 3.893,
      "step": 21008
    },
    {
      "epoch": 16.37,
      "learning_rate": 0.004917756506718685,
      "loss": 1.7225,
      "step": 21500
    },
    {
      "epoch": 16.76,
      "learning_rate": 0.004913909109045198,
      "loss": 1.7705,
      "step": 22000
    },
    {
      "epoch": 17.0,
      "eval_accuracy": 0.5684439284332112,
      "eval_loss": 1.948270559310913,
      "eval_runtime": 2619.7155,
      "eval_samples_per_second": 64.112,
      "eval_steps_per_second": 4.007,
      "step": 22321
    },
    {
      "epoch": 17.14,
      "learning_rate": 0.004909975339977676,
      "loss": 1.7215,
      "step": 22500
    },
    {
      "epoch": 17.52,
      "learning_rate": 0.004905955340269188,
      "loss": 1.6354,
      "step": 23000
    },
    {
      "epoch": 17.9,
      "learning_rate": 0.004901849253758198,
      "loss": 1.7089,
      "step": 23500
    },
    {
      "epoch": 18.0,
      "eval_accuracy": 0.5790598672263404,
      "eval_loss": 1.8927983045578003,
      "eval_runtime": 2625.6089,
      "eval_samples_per_second": 63.968,
      "eval_steps_per_second": 3.998,
      "step": 23634
    },
    {
      "epoch": 18.28,
      "learning_rate": 0.004897657227363419,
      "loss": 1.579,
      "step": 24000
    },
    {
      "epoch": 18.66,
      "learning_rate": 0.004893379411078551,
      "loss": 1.6068,
      "step": 24500
    },
    {
      "epoch": 19.0,
      "eval_accuracy": 0.5855497008127177,
      "eval_loss": 1.843549370765686,
      "eval_runtime": 2690.4067,
      "eval_samples_per_second": 62.427,
      "eval_steps_per_second": 3.902,
      "step": 24947
    },
    {
      "epoch": 19.04,
      "learning_rate": 0.004889015957966921,
      "loss": 1.6159,
      "step": 25000
    },
    {
      "epoch": 19.42,
      "learning_rate": 0.004884567024156001,
      "loss": 1.4949,
      "step": 25500
    },
    {
      "epoch": 19.8,
      "learning_rate": 0.004880032768831825,
      "loss": 1.5572,
      "step": 26000
    },
    {
      "epoch": 20.0,
      "eval_accuracy": 0.5879670149742491,
      "eval_loss": 1.840835690498352,
      "eval_runtime": 2628.5762,
      "eval_samples_per_second": 63.896,
      "eval_steps_per_second": 3.994,
      "step": 26260
    },
    {
      "epoch": 20.18,
      "learning_rate": 0.004875413354233289,
      "loss": 1.4928,
      "step": 26500
    },
    {
      "epoch": 20.56,
      "learning_rate": 0.0048707089456463525,
      "loss": 1.4466,
      "step": 27000
    },
    {
      "epoch": 20.94,
      "learning_rate": 0.004865919711398117,
      "loss": 1.4938,
      "step": 27500
    },
    {
      "epoch": 21.0,
      "eval_accuracy": 0.6109672233634009,
      "eval_loss": 1.7412927150726318,
      "eval_runtime": 2614.3175,
      "eval_samples_per_second": 64.244,
      "eval_steps_per_second": 4.016,
      "step": 27573
    },
    {
      "epoch": 21.33,
      "learning_rate": 0.0048610458228508105,
      "loss": 1.3796,
      "step": 28000
    },
    {
      "epoch": 21.71,
      "learning_rate": 0.004856087454395648,
      "loss": 1.4182,
      "step": 28500
    },
    {
      "epoch": 22.0,
      "eval_accuracy": 0.6154565210919591,
      "eval_loss": 1.7196179628372192,
      "eval_runtime": 2640.6742,
      "eval_samples_per_second": 63.603,
      "eval_steps_per_second": 3.975,
      "step": 28886
    },
    {
      "epoch": 22.09,
      "learning_rate": 0.0048510447834466,
      "loss": 1.4096,
      "step": 29000
    },
    {
      "epoch": 22.47,
      "learning_rate": 0.0048459179904340385,
      "loss": 1.3308,
      "step": 29500
    },
    {
      "epoch": 22.85,
      "learning_rate": 0.004840707258798284,
      "loss": 1.3784,
      "step": 30000
    },
    {
      "epoch": 23.0,
      "eval_accuracy": 0.6238218570450418,
      "eval_loss": 1.7105211019515991,
      "eval_runtime": 2885.4428,
      "eval_samples_per_second": 58.208,
      "eval_steps_per_second": 3.638,
      "step": 30199
    },
    {
      "epoch": 23.23,
      "learning_rate": 0.004835412774983039,
      "loss": 1.3104,
      "step": 30500
    },
    {
      "epoch": 23.61,
      "learning_rate": 0.004830034728428723,
      "loss": 1.301,
      "step": 31000
    },
    {
      "epoch": 23.99,
      "learning_rate": 0.004824573311565684,
      "loss": 1.3578,
      "step": 31500
    },
    {
      "epoch": 24.0,
      "eval_accuracy": 0.6176059063439612,
      "eval_loss": 1.7758694887161255,
      "eval_runtime": 2617.3025,
      "eval_samples_per_second": 64.171,
      "eval_steps_per_second": 4.011,
      "step": 31512
    },
    {
      "epoch": 24.37,
      "learning_rate": 0.004819028719807326,
      "loss": 1.2176,
      "step": 32000
    },
    {
      "epoch": 24.75,
      "learning_rate": 0.004813401151543105,
      "loss": 1.2763,
      "step": 32500
    },
    {
      "epoch": 25.0,
      "eval_accuracy": 0.6218749069691286,
      "eval_loss": 1.7364755868911743,
      "eval_runtime": 5925.4842,
      "eval_samples_per_second": 28.345,
      "eval_steps_per_second": 1.772,
      "step": 32825
    },
    {
      "epoch": 25.13,
      "learning_rate": 0.004807690808131437,
      "loss": 1.2551,
      "step": 33000
    },
    {
      "epoch": 25.51,
      "learning_rate": 0.004801897893892493,
      "loss": 1.1861,
      "step": 33500
    },
    {
      "epoch": 25.89,
      "learning_rate": 0.004796022616100888,
      "loss": 1.2484,
      "step": 34000
    },
    {
      "epoch": 26.0,
      "eval_accuracy": 0.6199220029174481,
      "eval_loss": 1.7483400106430054,
      "eval_runtime": 2684.7282,
      "eval_samples_per_second": 62.559,
      "eval_steps_per_second": 3.91,
      "step": 34138
    },
    {
      "epoch": 26.28,
      "learning_rate": 0.004790065184978262,
      "loss": 1.1726,
      "step": 34500
    },
    {
      "epoch": 26.66,
      "learning_rate": 0.004784025813685761,
      "loss": 1.1936,
      "step": 35000
    },
    {
      "epoch": 27.0,
      "eval_accuracy": 0.6313953142210711,
      "eval_loss": 1.700264573097229,
      "eval_runtime": 2640.722,
      "eval_samples_per_second": 63.602,
      "eval_steps_per_second": 3.975,
      "step": 35451
    },
    {
      "epoch": 27.04,
      "learning_rate": 0.004777904718316407,
      "loss": 1.2231,
      "step": 35500
    },
    {
      "epoch": 27.42,
      "learning_rate": 0.00477170211788737,
      "loss": 1.1121,
      "step": 36000
    },
    {
      "epoch": 27.8,
      "learning_rate": 0.00476541823433213,
      "loss": 1.1499,
      "step": 36500
    },
    {
      "epoch": 28.0,
      "eval_accuracy": 0.6247209073859069,
      "eval_loss": 1.7398767471313477,
      "eval_runtime": 2691.471,
      "eval_samples_per_second": 62.403,
      "eval_steps_per_second": 3.9,
      "step": 36764
    },
    {
      "epoch": 28.18,
      "learning_rate": 0.004759053292492532,
      "loss": 1.1329,
      "step": 37000
    },
    {
      "epoch": 28.56,
      "learning_rate": 0.004752607520110748,
      "loss": 1.1121,
      "step": 37500
    },
    {
      "epoch": 28.94,
      "learning_rate": 0.004746081147821121,
      "loss": 1.1418,
      "step": 38000
    },
    {
      "epoch": 29.0,
      "eval_accuracy": 0.6316751510821351,
      "eval_loss": 1.7091107368469238,
      "eval_runtime": 2632.2426,
      "eval_samples_per_second": 63.807,
      "eval_steps_per_second": 3.988,
      "step": 38077
    },
    {
      "epoch": 29.32,
      "learning_rate": 0.0047394744091419205,
      "loss": 1.0493,
      "step": 38500
    },
    {
      "epoch": 29.7,
      "learning_rate": 0.004732787540466979,
      "loss": 1.0895,
      "step": 39000
    },
    {
      "epoch": 30.0,
      "eval_accuracy": 0.6383078800869281,
      "eval_loss": 1.7165930271148682,
      "eval_runtime": 2645.0605,
      "eval_samples_per_second": 63.498,
      "eval_steps_per_second": 3.969,
      "step": 39390
    },
    {
      "epoch": 30.08,
      "learning_rate": 0.004726020781057241,
      "loss": 1.0905,
      "step": 39500
    },
    {
      "epoch": 30.46,
      "learning_rate": 0.004719174373032195,
      "loss": 1.0191,
      "step": 40000
    },
    {
      "epoch": 30.85,
      "learning_rate": 0.0047122485613612175,
      "loss": 1.0706,
      "step": 40500
    },
    {
      "epoch": 31.0,
      "eval_accuracy": 0.6373611979399244,
      "eval_loss": 1.7384090423583984,
      "eval_runtime": 2658.4828,
      "eval_samples_per_second": 63.177,
      "eval_steps_per_second": 3.949,
      "step": 40703
    },
    {
      "epoch": 31.23,
      "learning_rate": 0.004705243593854802,
      "loss": 1.0205,
      "step": 41000
    },
    {
      "epoch": 31.61,
      "learning_rate": 0.004698159721155694,
      "loss": 1.0072,
      "step": 41500
    },
    {
      "epoch": 31.99,
      "learning_rate": 0.004690997196729925,
      "loss": 1.0541,
      "step": 42000
    },
    {
      "epoch": 32.0,
      "eval_accuracy": 0.640868089666875,
      "eval_loss": 1.7335774898529053,
      "eval_runtime": 2595.2091,
      "eval_samples_per_second": 64.717,
      "eval_steps_per_second": 4.045,
      "step": 42016
    },
    {
      "epoch": 32.37,
      "learning_rate": 0.004683756276857741,
      "loss": 0.9544,
      "step": 42500
    },
    {
      "epoch": 32.75,
      "learning_rate": 0.004676437220624432,
      "loss": 1.0013,
      "step": 43000
    },
    {
      "epoch": 33.0,
      "eval_accuracy": 0.6451489982435772,
      "eval_loss": 1.7185142040252686,
      "eval_runtime": 2605.1367,
      "eval_samples_per_second": 64.471,
      "eval_steps_per_second": 4.03,
      "step": 43329
    },
    {
      "epoch": 33.13,
      "learning_rate": 0.004669040289911064,
      "loss": 0.9986,
      "step": 43500
    },
    {
      "epoch": 33.51,
      "learning_rate": 0.004661565749385107,
      "loss": 0.9399,
      "step": 44000
    },
    {
      "epoch": 33.89,
      "learning_rate": 0.004654013866490966,
      "loss": 0.9811,
      "step": 44500
    },
    {
      "epoch": 34.0,
      "eval_accuracy": 0.6478640111934745,
      "eval_loss": 1.7246044874191284,
      "eval_runtime": 2595.1453,
      "eval_samples_per_second": 64.719,
      "eval_steps_per_second": 4.045,
      "step": 44642
    },
    {
      "epoch": 34.27,
      "learning_rate": 0.004646384911440411,
      "loss": 0.914,
      "step": 45000
    },
    {
      "epoch": 34.65,
      "learning_rate": 0.004638679157202909,
      "loss": 0.9447,
      "step": 45500
    },
    {
      "epoch": 35.0,
      "eval_accuracy": 0.6540263761126492,
      "eval_loss": 1.7245060205459595,
      "eval_runtime": 2605.4725,
      "eval_samples_per_second": 64.462,
      "eval_steps_per_second": 4.029,
      "step": 45955
    },
    {
      "epoch": 35.03,
      "learning_rate": 0.0005810494104962897,
      "loss": 0.8424,
      "step": 46000
    },
    {
      "epoch": 35.42,
      "learning_rate": 0.0005391326120998649,
      "loss": 0.692,
      "step": 46500
    },
    {
      "epoch": 35.8,
      "learning_rate": 0.0004986016371937035,
      "loss": 0.6587,
      "step": 47000
    },
    {
      "epoch": 36.0,
      "eval_accuracy": 0.7018546634515197,
      "eval_loss": 1.5848796367645264,
      "eval_runtime": 4946.5461,
      "eval_samples_per_second": 33.954,
      "eval_steps_per_second": 2.122,
      "step": 47268
    },
    {
      "epoch": 36.18,
      "learning_rate": 0.0004594851306405101,
      "loss": 0.6234,
      "step": 47500
    },
    {
      "epoch": 36.56,
      "learning_rate": 0.0004218107376415739,
      "loss": 0.6053,
      "step": 48000
    },
    {
      "epoch": 36.94,
      "learning_rate": 0.00038560508419879514,
      "loss": 0.6044,
      "step": 48500
    },
    {
      "epoch": 37.0,
      "eval_accuracy": 0.706195111785895,
      "eval_loss": 1.6146132946014404,
      "eval_runtime": 2680.1847,
      "eval_samples_per_second": 62.665,
      "eval_steps_per_second": 3.917,
      "step": 48581
    },
    {
      "epoch": 37.32,
      "learning_rate": 0.0003508937582970281,
      "loss": 0.5763,
      "step": 49000
    },
    {
      "epoch": 37.7,
      "learning_rate": 0.0003177012918200253,
      "loss": 0.572,
      "step": 49500
    },
    {
      "epoch": 38.0,
      "eval_accuracy": 0.7081241999345063,
      "eval_loss": 1.6583054065704346,
      "eval_runtime": 2714.09,
      "eval_samples_per_second": 61.883,
      "eval_steps_per_second": 3.868,
      "step": 49894
    },
    {
      "epoch": 38.08,
      "learning_rate": 0.0002860511432127716,
      "loss": 0.5693,
      "step": 50000
    },
    {
      "epoch": 38.46,
      "learning_rate": 0.00025596568090246547,
      "loss": 0.5492,
      "step": 50500
    },
    {
      "epoch": 38.84,
      "learning_rate": 0.00022746616748985177,
      "loss": 0.545,
      "step": 51000
    },
    {
      "epoch": 39.0,
      "eval_accuracy": 0.7086898276324015,
      "eval_loss": 1.6993166208267212,
      "eval_runtime": 8190.9805,
      "eval_samples_per_second": 20.505,
      "eval_steps_per_second": 1.282,
      "step": 51207
    },
    {
      "epoch": 39.22,
      "learning_rate": 0.0002005727447220948,
      "loss": 0.5409,
      "step": 51500
    },
    {
      "epoch": 39.6,
      "learning_rate": 0.00017530441925779654,
      "loss": 0.5351,
      "step": 52000
    },
    {
      "epoch": 39.98,
      "learning_rate": 0.00015167904923422986,
      "loss": 0.5341,
      "step": 52500
    },
    {
      "epoch": 40.0,
      "eval_accuracy": 0.7105593760233396,
      "eval_loss": 1.7077844142913818,
      "eval_runtime": 33225.1715,
      "eval_samples_per_second": 5.055,
      "eval_steps_per_second": 0.316,
      "step": 52520
    },
    {
      "epoch": 40.37,
      "learning_rate": 0.0001297133316462798,
      "loss": 0.5184,
      "step": 53000
    },
    {
      "epoch": 40.75,
      "learning_rate": 0.00010942279054600657,
      "loss": 0.5284,
      "step": 53500
    },
    {
      "epoch": 41.0,
      "eval_accuracy": 0.7104641124110624,
      "eval_loss": 1.7240744829177856,
      "eval_runtime": 3401.4075,
      "eval_samples_per_second": 49.378,
      "eval_steps_per_second": 3.086,
      "step": 53833
    },
    {
      "epoch": 41.13,
      "learning_rate": 9.082176607117799e-05,
      "loss": 0.5239,
      "step": 54000
    },
    {
      "epoch": 41.51,
      "learning_rate": 7.392340431051969e-05,
      "loss": 0.5152,
      "step": 54500
    },
    {
      "epoch": 41.89,
      "learning_rate": 5.87396480128477e-05,
      "loss": 0.5186,
      "step": 55000
    },
    {
      "epoch": 42.0,
      "eval_accuracy": 0.7111666815516061,
      "eval_loss": 1.7408066987991333,
      "eval_runtime": 4964.1178,
      "eval_samples_per_second": 33.834,
      "eval_steps_per_second": 2.115,
      "step": 55146
    },
    {
      "epoch": 42.27,
      "learning_rate": 4.528122814664942e-05,
      "loss": 0.5113,
      "step": 55500
    },
    {
      "epoch": 42.65,
      "learning_rate": 3.35576563160786e-05,
      "loss": 0.506,
      "step": 56000
    },
    {
      "epoch": 43.0,
      "eval_accuracy": 0.7105772379506415,
      "eval_loss": 1.7486644983291626,
      "eval_runtime": 6355.0593,
      "eval_samples_per_second": 26.429,
      "eval_steps_per_second": 1.652,
      "step": 56459
    },
    {
      "epoch": 43.03,
      "learning_rate": 2.357721803872476e-05,
      "loss": 0.5137,
      "step": 56500
    },
    {
      "epoch": 43.41,
      "learning_rate": 1.5346966889905334e-05,
      "loss": 0.5062,
      "step": 57000
    },
    {
      "epoch": 43.79,
      "learning_rate": 8.872719517621642e-06,
      "loss": 0.5043,
      "step": 57500
    },
    {
      "epoch": 44.0,
      "eval_accuracy": 0.7108689827632402,
      "eval_loss": 1.7547060251235962,
      "eval_runtime": 4684.6858,
      "eval_samples_per_second": 35.852,
      "eval_steps_per_second": 2.241,
      "step": 57772
    },
    {
      "epoch": 44.17,
      "learning_rate": 4.159051531700819e-06,
      "loss": 0.5084,
      "step": 58000
    },
    {
      "epoch": 44.55,
      "learning_rate": 1.2092942700298037e-06,
      "loss": 0.5043,
      "step": 58500
    },
    {
      "epoch": 44.94,
      "learning_rate": 2.5532444165032865e-08,
      "loss": 0.5094,
      "step": 59000
    },
    {
      "epoch": 45.0,
      "eval_accuracy": 0.711053556012027,
      "eval_loss": 1.7535769939422607,
      "eval_runtime": 2625.3482,
      "eval_samples_per_second": 63.974,
      "eval_steps_per_second": 3.999,
      "step": 59085
    },
    {
      "epoch": 45.32,
      "learning_rate": 0.0007032209920746526,
      "loss": 0.5386,
      "step": 59500
    },
    {
      "epoch": 45.7,
      "learning_rate": 0.0006689210221610515,
      "loss": 0.5547,
      "step": 60000
    },
    {
      "epoch": 46.0,
      "eval_accuracy": 0.7068619570718347,
      "eval_loss": 1.7073622941970825,
      "eval_runtime": 6607.5308,
      "eval_samples_per_second": 25.419,
      "eval_steps_per_second": 1.589,
      "step": 60398
    },
    {
      "epoch": 46.08,
      "learning_rate": 0.0006353490004690119,
      "loss": 0.553,
      "step": 60500
    },
    {
      "epoch": 46.46,
      "learning_rate": 0.0006025182736066154,
      "loss": 0.5329,
      "step": 61000
    },
    {
      "epoch": 46.84,
      "learning_rate": 0.000570441893479062,
      "loss": 0.5391,
      "step": 61500
    },
    {
      "epoch": 47.0,
      "eval_accuracy": 0.7089637105176982,
      "eval_loss": 1.740086317062378,
      "eval_runtime": 5830.6066,
      "eval_samples_per_second": 28.806,
      "eval_steps_per_second": 1.8,
      "step": 61711
    },
    {
      "epoch": 47.22,
      "learning_rate": 0.0005391326120998649,
      "loss": 0.5284,
      "step": 62000
    },
    {
      "epoch": 47.6,
      "learning_rate": 0.0005086028765212663,
      "loss": 0.5223,
      "step": 62500
    },
    {
      "epoch": 47.98,
      "learning_rate": 0.00047886482388589615,
      "loss": 0.5253,
      "step": 63000
    },
    {
      "epoch": 48.0,
      "eval_accuracy": 0.7092852252091334,
      "eval_loss": 1.7769759893417358,
      "eval_runtime": 2617.8614,
      "eval_samples_per_second": 64.157,
      "eval_steps_per_second": 4.01,
      "step": 63024
    },
    {
      "epoch": 48.36,
      "learning_rate": 0.0004499302766016297,
      "loss": 0.5041,
      "step": 63500
    },
    {
      "epoch": 48.74,
      "learning_rate": 0.0004218107376415739,
      "loss": 0.5066,
      "step": 64000
    },
    {
      "epoch": 49.0,
      "eval_accuracy": 0.7102319073561371,
      "eval_loss": 1.813459873199463,
      "eval_runtime": 2608.5743,
      "eval_samples_per_second": 64.386,
      "eval_steps_per_second": 4.024,
      "step": 64337
    },
    {
      "epoch": 49.12,
      "learning_rate": 0.00039451738597103935,
      "loss": 0.5057,
      "step": 64500
    },
    {
      "epoch": 49.5,
      "learning_rate": 0.00036806107210333034,
      "loss": 0.4897,
      "step": 65000
    },
    {
      "epoch": 49.89,
      "learning_rate": 0.00034245231378610494,
      "loss": 0.495,
      "step": 65500
    },
    {
      "epoch": 50.0,
      "eval_accuracy": 0.7110476020362597,
      "eval_loss": 1.8451627492904663,
      "eval_runtime": 2646.4398,
      "eval_samples_per_second": 63.465,
      "eval_steps_per_second": 3.967,
      "step": 65650
    },
    {
      "epoch": 50.27,
      "learning_rate": 0.0003177012918200253,
      "loss": 0.4867,
      "step": 66000
    },
    {
      "epoch": 50.65,
      "learning_rate": 0.0002938178460113711,
      "loss": 0.4813,
      "step": 66500
    },
    {
      "epoch": 51.0,
      "eval_accuracy": 0.7107320413205919,
      "eval_loss": 1.8846126794815063,
      "eval_runtime": 5243.8791,
      "eval_samples_per_second": 32.029,
      "eval_steps_per_second": 2.002,
      "step": 66963
    },
    {
      "epoch": 51.03,
      "learning_rate": 0.00027081147126020445,
      "loss": 0.4829,
      "step": 67000
    },
    {
      "epoch": 51.41,
      "learning_rate": 0.00024869131378566,
      "loss": 0.466,
      "step": 67500
    },
    {
      "epoch": 51.79,
      "learning_rate": 0.00022746616748985177,
      "loss": 0.4704,
      "step": 68000
    },
    {
      "epoch": 52.0,
      "eval_accuracy": 0.7123693846566045,
      "eval_loss": 1.8989039659500122,
      "eval_runtime": 2607.2421,
      "eval_samples_per_second": 64.419,
      "eval_steps_per_second": 4.026,
      "step": 68276
    },
    {
      "epoch": 52.17,
      "learning_rate": 0.00020714447046184416,
      "loss": 0.47,
      "step": 68500
    },
    {
      "epoch": 52.55,
      "learning_rate": 0.0001877343016230762,
      "loss": 0.465,
      "step": 69000
    },
    {
      "epoch": 52.93,
      "learning_rate": 0.0001692433775155744,
      "loss": 0.4689,
      "step": 69500
    },
    {
      "epoch": 53.0,
      "eval_accuracy": 0.7131850793367271,
      "eval_loss": 1.931063175201416,
      "eval_runtime": 2614.8439,
      "eval_samples_per_second": 64.231,
      "eval_steps_per_second": 4.015,
      "step": 69589
    },
    {
      "epoch": 53.31,
      "learning_rate": 0.00015167904923422986,
      "loss": 0.4572,
      "step": 70000
    },
    {
      "epoch": 53.69,
      "learning_rate": 0.0001350482995043595,
      "loss": 0.4611,
      "step": 70500
    },
    {
      "epoch": 54.0,
      "eval_accuracy": 0.7130659998213807,
      "eval_loss": 1.93539559841156,
      "eval_runtime": 2604.5499,
      "eval_samples_per_second": 64.485,
      "eval_steps_per_second": 4.031,
      "step": 70902
    },
    {
      "epoch": 54.07,
      "learning_rate": 0.00011935773990571164,
      "loss": 0.4579,
      "step": 71000
    },
    {
      "epoch": 54.46,
      "learning_rate": 0.00010461360824402521,
      "loss": 0.4554,
      "step": 71500
    },
    {
      "epoch": 54.84,
      "learning_rate": 9.082176607117799e-05,
      "loss": 0.4547,
      "step": 72000
    },
    {
      "epoch": 55.0,
      "eval_accuracy": 0.7133160668036082,
      "eval_loss": 1.974073052406311,
      "eval_runtime": 2574.37,
      "eval_samples_per_second": 65.241,
      "eval_steps_per_second": 4.078,
      "step": 72215
    },
    {
      "epoch": 55.22,
      "learning_rate": 7.798769635491976e-05,
      "loss": 0.4497,
      "step": 72500
    },
    {
      "epoch": 55.6,
      "learning_rate": 6.611650129910801e-05,
      "loss": 0.4458,
      "step": 73000
    },
    {
      "epoch": 55.98,
      "learning_rate": 5.521290031532178e-05,
      "loss": 0.4481,
      "step": 73500
    },
    {
      "epoch": 56.0,
      "eval_accuracy": 0.713071953797148,
      "eval_loss": 1.9898993968963623,
      "eval_runtime": 2622.4841,
      "eval_samples_per_second": 64.044,
      "eval_steps_per_second": 4.003,
      "step": 73528
    },
    {
      "epoch": 56.36,
      "learning_rate": 0.00045401414969093167,
      "loss": 0.4607,
      "step": 74000
    },
    {
      "epoch": 56.74,
      "learning_rate": 0.0004297610695623985,
      "loss": 0.4709,
      "step": 74500
    },
    {
      "epoch": 57.0,
      "eval_accuracy": 0.7104283885564586,
      "eval_loss": 1.9411875009536743,
      "eval_runtime": 3269.1549,
      "eval_samples_per_second": 51.376,
      "eval_steps_per_second": 3.211,
      "step": 74841
    },
    {
      "epoch": 57.12,
      "learning_rate": 0.0004061126673698029,
      "loss": 0.4736,
      "step": 75000
    },
    {
      "epoch": 57.5,
      "learning_rate": 0.0003830758503675777,
      "loss": 0.4601,
      "step": 75500
    },
    {
      "epoch": 57.88,
      "learning_rate": 0.00036065734717759815,
      "loss": 0.4647,
      "step": 76000
    },
    {
      "epoch": 58.0,
      "eval_accuracy": 0.7097913131493555,
      "eval_loss": 1.9707293510437012,
      "eval_runtime": 2595.9104,
      "eval_samples_per_second": 64.7,
      "eval_steps_per_second": 4.044,
      "step": 76154
    },
    {
      "epoch": 58.26,
      "learning_rate": 0.00033886370582388,
      "loss": 0.4562,
      "step": 76500
    },
    {
      "epoch": 58.64,
      "learning_rate": 0.0003177012918200253,
      "loss": 0.4566,
      "step": 77000
    },
    {
      "epoch": 59.0,
      "eval_accuracy": 0.7116251376856896,
      "eval_loss": 2.0150792598724365,
      "eval_runtime": 2589.1949,
      "eval_samples_per_second": 64.868,
      "eval_steps_per_second": 4.055,
      "step": 77467
    },
    {
      "epoch": 59.03,
      "learning_rate": 0.0002971762863099739,
      "loss": 0.4535,
      "step": 77500
    },
    {
      "epoch": 59.41,
      "learning_rate": 0.0002772946842626087,
      "loss": 0.448,
      "step": 78000
    },
    {
      "epoch": 59.79,
      "learning_rate": 0.000258062292720736,
      "loss": 0.4511,
      "step": 78500
    },
    {
      "epoch": 60.0,
      "eval_accuracy": 0.7113929326307642,
      "eval_loss": 2.0363075733184814,
      "eval_runtime": 2640.6575,
      "eval_samples_per_second": 63.603,
      "eval_steps_per_second": 3.976,
      "step": 78780
    },
    {
      "epoch": 60.17,
      "learning_rate": 0.00023948472910495817,
      "loss": 0.4446,
      "step": 79000
    },
    {
      "epoch": 60.55,
      "learning_rate": 0.0002215674195729317,
      "loss": 0.4403,
      "step": 79500
    },
    {
      "epoch": 60.93,
      "learning_rate": 0.0002043155974344868,
      "loss": 0.4423,
      "step": 80000
    },
    {
      "epoch": 61.0,
      "eval_accuracy": 0.7112143133577447,
      "eval_loss": 2.0710268020629883,
      "eval_runtime": 2593.0457,
      "eval_samples_per_second": 64.771,
      "eval_steps_per_second": 4.049,
      "step": 80093
    },
    {
      "epoch": 61.31,
      "learning_rate": 0.0001877343016230762,
      "loss": 0.4304,
      "step": 80500
    },
    {
      "epoch": 61.69,
      "learning_rate": 0.00017182837522399742,
      "loss": 0.4356,
      "step": 81000
    },
    {
      "epoch": 62.0,
      "eval_accuracy": 0.7116251376856896,
      "eval_loss": 2.061058521270752,
      "eval_runtime": 2582.6346,
      "eval_samples_per_second": 65.032,
      "eval_steps_per_second": 4.065,
      "step": 81406
    },
    {
      "epoch": 62.07,
      "learning_rate": 0.00015660246405981793,
      "loss": 0.4332,
      "step": 81500
    },
    {
      "epoch": 62.45,
      "learning_rate": 0.00014206101533341926,
      "loss": 0.43,
      "step": 82000
    },
    {
      "epoch": 62.83,
      "learning_rate": 0.0001282082763290507,
      "loss": 0.4272,
      "step": 82500
    },
    {
      "epoch": 63.0,
      "eval_accuracy": 0.7117918490071745,
      "eval_loss": 2.089148998260498,
      "eval_runtime": 2586.4124,
      "eval_samples_per_second": 64.937,
      "eval_steps_per_second": 4.059,
      "step": 82719
    },
    {
      "epoch": 63.21,
      "learning_rate": 0.00011504829317177723,
      "loss": 0.4259,
      "step": 83000
    },
    {
      "epoch": 63.59,
      "learning_rate": 0.00010258490964568406,
      "loss": 0.4271,
      "step": 83500
    },
    {
      "epoch": 63.98,
      "learning_rate": 9.082176607117799e-05,
      "loss": 0.4254,
      "step": 84000
    },
    {
      "epoch": 64.0,
      "eval_accuracy": 0.7123812926081391,
      "eval_loss": 2.0879101753234863,
      "eval_runtime": 2610.5128,
      "eval_samples_per_second": 64.338,
      "eval_steps_per_second": 4.021,
      "step": 84032
    },
    {
      "epoch": 64.36,
      "learning_rate": 7.976229824171872e-05,
      "loss": 0.4249,
      "step": 84500
    },
    {
      "epoch": 64.74,
      "learning_rate": 6.940973642028724e-05,
      "loss": 0.4221,
      "step": 85000
    },
    {
      "epoch": 65.0,
      "eval_accuracy": 0.7130957697002174,
      "eval_loss": 2.1167192459106445,
      "eval_runtime": 2625.6926,
      "eval_samples_per_second": 63.966,
      "eval_steps_per_second": 3.998,
      "step": 85345
    },
    {
      "epoch": 65.12,
      "learning_rate": 5.976710439588779e-05,
      "loss": 0.4227,
      "step": 85500
    },
    {
      "epoch": 65.5,
      "learning_rate": 5.083721860035529e-05,
      "loss": 0.421,
      "step": 86000
    },
    {
      "epoch": 65.88,
      "learning_rate": 4.262268728572882e-05,
      "loss": 0.4189,
      "step": 86500
    },
    {
      "epoch": 66.0,
      "eval_accuracy": 0.7129111964514304,
      "eval_loss": 2.1362690925598145,
      "eval_runtime": 2775.8163,
      "eval_samples_per_second": 60.507,
      "eval_steps_per_second": 3.782,
      "step": 86658
    },
    {
      "epoch": 66.26,
      "learning_rate": 3.512590976242902e-05,
      "loss": 0.4192,
      "step": 87000
    },
    {
      "epoch": 66.64,
      "learning_rate": 2.834907569846612e-05,
      "loss": 0.4219,
      "step": 87500
    },
    {
      "epoch": 67.0,
      "eval_accuracy": 0.712958828257569,
      "eval_loss": 2.1355273723602295,
      "eval_runtime": 2581.0691,
      "eval_samples_per_second": 65.072,
      "eval_steps_per_second": 4.067,
      "step": 87971
    },
    {
      "epoch": 67.02,
      "learning_rate": 2.2294164479878255e-05,
      "loss": 0.4197,
      "step": 88000
    },
    {
      "epoch": 67.4,
      "learning_rate": 1.6962944632589183e-05,
      "loss": 0.4157,
      "step": 88500
    },
    {
      "epoch": 67.78,
      "learning_rate": 1.2356973305856779e-05,
      "loss": 0.4149,
      "step": 89000
    },
    {
      "epoch": 68.0,
      "eval_accuracy": 0.7132267571670984,
      "eval_loss": 2.1466283798217773,
      "eval_runtime": 2769.7004,
      "eval_samples_per_second": 60.64,
      "eval_steps_per_second": 3.79,
      "step": 89284
    },
    {
      "epoch": 68.16,
      "learning_rate": 8.477595817457106e-06,
      "loss": 0.4138,
      "step": 89500
    },
    {
      "epoch": 68.55,
      "learning_rate": 5.325945260744314e-06,
      "loss": 0.4147,
      "step": 90000
    },
    {
      "epoch": 68.93,
      "learning_rate": 2.90294217369369e-06,
      "loss": 0.4125,
      "step": 90500
    },
    {
      "epoch": 69.0,
      "eval_accuracy": 0.7131434015063559,
      "eval_loss": 2.1478145122528076,
      "eval_runtime": 2585.5905,
      "eval_samples_per_second": 64.958,
      "eval_steps_per_second": 4.06,
      "step": 90597
    },
    {
      "epoch": 69.31,
      "learning_rate": 1.2092942700298037e-06,
      "loss": 0.4146,
      "step": 91000
    },
    {
      "epoch": 69.69,
      "learning_rate": 2.4549623251851615e-07,
      "loss": 0.4162,
      "step": 91500
    },
    {
      "epoch": 70.0,
      "eval_accuracy": 0.7132267571670984,
      "eval_loss": 2.1484131813049316,
      "eval_runtime": 2581.5568,
      "eval_samples_per_second": 65.06,
      "eval_steps_per_second": 4.067,
      "step": 91910
    },
    {
      "epoch": 70.0,
      "step": 91910,
      "total_flos": 3.2073215097814647e+21,
      "train_loss": 0.08673005339085323,
      "train_runtime": 125919.3774,
      "train_samples_per_second": 373.471,
      "train_steps_per_second": 0.73
    }
  ],
  "logging_steps": 500,
  "max_steps": 91910,
  "num_train_epochs": 70,
  "save_steps": 500,
  "total_flos": 3.2073215097814647e+21,
  "trial_name": null,
  "trial_params": null
}