{
  "best_metric": 0.2568952143192291,
  "best_model_checkpoint": "./results_train/bert-base-cased/qnli/checkpoint-13000",
  "epoch": 10.0,
  "global_step": 65470,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.08,
      "learning_rate": 2.545176889793841e-06,
      "loss": 0.6529,
      "step": 500
    },
    {
      "epoch": 0.08,
      "eval_accuracy": 0.7733845872231375,
      "eval_loss": 0.4945374131202698,
      "eval_runtime": 15.9632,
      "eval_samples_per_second": 342.225,
      "eval_steps_per_second": 42.786,
      "step": 500
    },
    {
      "epoch": 0.15,
      "learning_rate": 5.090353779587682e-06,
      "loss": 0.4731,
      "step": 1000
    },
    {
      "epoch": 0.15,
      "eval_accuracy": 0.8405637927878455,
      "eval_loss": 0.38875827193260193,
      "eval_runtime": 16.0177,
      "eval_samples_per_second": 341.06,
      "eval_steps_per_second": 42.64,
      "step": 1000
    },
    {
      "epoch": 0.23,
      "learning_rate": 7.635530669381522e-06,
      "loss": 0.4113,
      "step": 1500
    },
    {
      "epoch": 0.23,
      "eval_accuracy": 0.8522789676002197,
      "eval_loss": 0.3604711890220642,
      "eval_runtime": 15.9651,
      "eval_samples_per_second": 342.185,
      "eval_steps_per_second": 42.781,
      "step": 1500
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.0180707559175364e-05,
      "loss": 0.4006,
      "step": 2000
    },
    {
      "epoch": 0.31,
      "eval_accuracy": 0.8649093904448105,
      "eval_loss": 0.3175490200519562,
      "eval_runtime": 16.0149,
      "eval_samples_per_second": 341.119,
      "eval_steps_per_second": 42.648,
      "step": 2000
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.2725884448969203e-05,
      "loss": 0.3615,
      "step": 2500
    },
    {
      "epoch": 0.38,
      "eval_accuracy": 0.867838184147904,
      "eval_loss": 0.32201188802719116,
      "eval_runtime": 15.9586,
      "eval_samples_per_second": 342.323,
      "eval_steps_per_second": 42.798,
      "step": 2500
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.5271061338763045e-05,
      "loss": 0.3662,
      "step": 3000
    },
    {
      "epoch": 0.46,
      "eval_accuracy": 0.866739886509244,
      "eval_loss": 0.3197564482688904,
      "eval_runtime": 16.037,
      "eval_samples_per_second": 340.649,
      "eval_steps_per_second": 42.589,
      "step": 3000
    },
    {
      "epoch": 0.53,
      "learning_rate": 1.7816238228556887e-05,
      "loss": 0.3578,
      "step": 3500
    },
    {
      "epoch": 0.53,
      "eval_accuracy": 0.8779059125022881,
      "eval_loss": 0.29696720838546753,
      "eval_runtime": 15.9685,
      "eval_samples_per_second": 342.112,
      "eval_steps_per_second": 42.772,
      "step": 3500
    },
    {
      "epoch": 0.61,
      "learning_rate": 1.997692595180449e-05,
      "loss": 0.3349,
      "step": 4000
    },
    {
      "epoch": 0.61,
      "eval_accuracy": 0.8833974006955885,
      "eval_loss": 0.28639864921569824,
      "eval_runtime": 16.0264,
      "eval_samples_per_second": 340.875,
      "eval_steps_per_second": 42.617,
      "step": 4000
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.9814432654652997e-05,
      "loss": 0.3548,
      "step": 4500
    },
    {
      "epoch": 0.69,
      "eval_accuracy": 0.8945634266886326,
      "eval_loss": 0.2664399743080139,
      "eval_runtime": 15.9996,
      "eval_samples_per_second": 341.447,
      "eval_steps_per_second": 42.689,
      "step": 4500
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.9651939357501505e-05,
      "loss": 0.3254,
      "step": 5000
    },
    {
      "epoch": 0.76,
      "eval_accuracy": 0.8929159802306426,
      "eval_loss": 0.2650640904903412,
      "eval_runtime": 16.02,
      "eval_samples_per_second": 341.011,
      "eval_steps_per_second": 42.634,
      "step": 5000
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.9489446060350013e-05,
      "loss": 0.3212,
      "step": 5500
    },
    {
      "epoch": 0.84,
      "eval_accuracy": 0.8817499542375984,
      "eval_loss": 0.2942565679550171,
      "eval_runtime": 15.947,
      "eval_samples_per_second": 342.573,
      "eval_steps_per_second": 42.829,
      "step": 5500
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.932695276319852e-05,
      "loss": 0.3195,
      "step": 6000
    },
    {
      "epoch": 0.92,
      "eval_accuracy": 0.8960278235401794,
      "eval_loss": 0.26299455761909485,
      "eval_runtime": 15.9626,
      "eval_samples_per_second": 342.237,
      "eval_steps_per_second": 42.787,
      "step": 6000
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.916445946604703e-05,
      "loss": 0.3044,
      "step": 6500
    },
    {
      "epoch": 0.99,
      "eval_accuracy": 0.8881566904631155,
      "eval_loss": 0.27664485573768616,
      "eval_runtime": 15.9599,
      "eval_samples_per_second": 342.295,
      "eval_steps_per_second": 42.795,
      "step": 6500
    },
    {
      "epoch": 1.07,
      "learning_rate": 1.9001966168895533e-05,
      "loss": 0.245,
      "step": 7000
    },
    {
      "epoch": 1.07,
      "eval_accuracy": 0.9013362621270364,
      "eval_loss": 0.3268437087535858,
      "eval_runtime": 16.0333,
      "eval_samples_per_second": 340.728,
      "eval_steps_per_second": 42.599,
      "step": 7000
    },
    {
      "epoch": 1.15,
      "learning_rate": 1.883947287174404e-05,
      "loss": 0.2415,
      "step": 7500
    },
    {
      "epoch": 1.15,
      "eval_accuracy": 0.8991396668497162,
      "eval_loss": 0.29871582984924316,
      "eval_runtime": 15.9621,
      "eval_samples_per_second": 342.249,
      "eval_steps_per_second": 42.789,
      "step": 7500
    },
    {
      "epoch": 1.22,
      "learning_rate": 1.867697957459255e-05,
      "loss": 0.248,
      "step": 8000
    },
    {
      "epoch": 1.22,
      "eval_accuracy": 0.8812008054182684,
      "eval_loss": 0.3372972905635834,
      "eval_runtime": 16.0184,
      "eval_samples_per_second": 341.046,
      "eval_steps_per_second": 42.639,
      "step": 8000
    },
    {
      "epoch": 1.3,
      "learning_rate": 1.8514486277441056e-05,
      "loss": 0.2328,
      "step": 8500
    },
    {
      "epoch": 1.3,
      "eval_accuracy": 0.8808347062053816,
      "eval_loss": 0.34742510318756104,
      "eval_runtime": 15.9539,
      "eval_samples_per_second": 342.424,
      "eval_steps_per_second": 42.811,
      "step": 8500
    },
    {
      "epoch": 1.37,
      "learning_rate": 1.8351992980289564e-05,
      "loss": 0.2427,
      "step": 9000
    },
    {
      "epoch": 1.37,
      "eval_accuracy": 0.8912685337726524,
      "eval_loss": 0.2986329197883606,
      "eval_runtime": 15.9879,
      "eval_samples_per_second": 341.695,
      "eval_steps_per_second": 42.72,
      "step": 9000
    },
    {
      "epoch": 1.45,
      "learning_rate": 1.8189499683138072e-05,
      "loss": 0.243,
      "step": 9500
    },
    {
      "epoch": 1.45,
      "eval_accuracy": 0.8982244188174996,
      "eval_loss": 0.2834506034851074,
      "eval_runtime": 15.9775,
      "eval_samples_per_second": 341.919,
      "eval_steps_per_second": 42.748,
      "step": 9500
    },
    {
      "epoch": 1.53,
      "learning_rate": 1.802700638598658e-05,
      "loss": 0.2375,
      "step": 10000
    },
    {
      "epoch": 1.53,
      "eval_accuracy": 0.8909024345597657,
      "eval_loss": 0.2893359661102295,
      "eval_runtime": 15.9492,
      "eval_samples_per_second": 342.525,
      "eval_steps_per_second": 42.823,
      "step": 10000
    },
    {
      "epoch": 1.6,
      "learning_rate": 1.7864513088835088e-05,
      "loss": 0.2253,
      "step": 10500
    },
    {
      "epoch": 1.6,
      "eval_accuracy": 0.8960278235401794,
      "eval_loss": 0.2941761910915375,
      "eval_runtime": 15.9667,
      "eval_samples_per_second": 342.15,
      "eval_steps_per_second": 42.777,
      "step": 10500
    },
    {
      "epoch": 1.68,
      "learning_rate": 1.7702019791683592e-05,
      "loss": 0.2429,
      "step": 11000
    },
    {
      "epoch": 1.68,
      "eval_accuracy": 0.9028006589785832,
      "eval_loss": 0.28518712520599365,
      "eval_runtime": 16.0239,
      "eval_samples_per_second": 340.928,
      "eval_steps_per_second": 42.624,
      "step": 11000
    },
    {
      "epoch": 1.76,
      "learning_rate": 1.75395264945321e-05,
      "loss": 0.238,
      "step": 11500
    },
    {
      "epoch": 1.76,
      "eval_accuracy": 0.9037159070108,
      "eval_loss": 0.2671617865562439,
      "eval_runtime": 15.9753,
      "eval_samples_per_second": 341.966,
      "eval_steps_per_second": 42.754,
      "step": 11500
    },
    {
      "epoch": 1.83,
      "learning_rate": 1.737703319738061e-05,
      "loss": 0.2344,
      "step": 12000
    },
    {
      "epoch": 1.83,
      "eval_accuracy": 0.9020684605528098,
      "eval_loss": 0.267158567905426,
      "eval_runtime": 16.0345,
      "eval_samples_per_second": 340.703,
      "eval_steps_per_second": 42.596,
      "step": 12000
    },
    {
      "epoch": 1.91,
      "learning_rate": 1.7214539900229116e-05,
      "loss": 0.2368,
      "step": 12500
    },
    {
      "epoch": 1.91,
      "eval_accuracy": 0.9018854109463664,
      "eval_loss": 0.2753552496433258,
      "eval_runtime": 15.9627,
      "eval_samples_per_second": 342.234,
      "eval_steps_per_second": 42.787,
      "step": 12500
    },
    {
      "epoch": 1.99,
      "learning_rate": 1.7052046603077624e-05,
      "loss": 0.2424,
      "step": 13000
    },
    {
      "epoch": 1.99,
      "eval_accuracy": 0.9086582463847702,
      "eval_loss": 0.2568952143192291,
      "eval_runtime": 15.9721,
      "eval_samples_per_second": 342.034,
      "eval_steps_per_second": 42.762,
      "step": 13000
    },
    {
      "epoch": 2.06,
      "learning_rate": 1.6889553305926132e-05,
      "loss": 0.1542,
      "step": 13500
    },
    {
      "epoch": 2.06,
      "eval_accuracy": 0.9026176093721399,
      "eval_loss": 0.38070985674858093,
      "eval_runtime": 15.9896,
      "eval_samples_per_second": 341.659,
      "eval_steps_per_second": 42.715,
      "step": 13500
    },
    {
      "epoch": 2.14,
      "learning_rate": 1.672706000877464e-05,
      "loss": 0.1416,
      "step": 14000
    },
    {
      "epoch": 2.14,
      "eval_accuracy": 0.8980413692110562,
      "eval_loss": 0.3915307819843292,
      "eval_runtime": 15.9993,
      "eval_samples_per_second": 341.452,
      "eval_steps_per_second": 42.689,
      "step": 14000
    },
    {
      "epoch": 2.21,
      "learning_rate": 1.6564566711623148e-05,
      "loss": 0.1445,
      "step": 14500
    },
    {
      "epoch": 2.21,
      "eval_accuracy": 0.8984074684239429,
      "eval_loss": 0.42921990156173706,
      "eval_runtime": 15.964,
      "eval_samples_per_second": 342.208,
      "eval_steps_per_second": 42.784,
      "step": 14500
    },
    {
      "epoch": 2.29,
      "learning_rate": 1.6402073414471655e-05,
      "loss": 0.1631,
      "step": 15000
    },
    {
      "epoch": 2.29,
      "eval_accuracy": 0.8971261211788395,
      "eval_loss": 0.4096887409687042,
      "eval_runtime": 16.0404,
      "eval_samples_per_second": 340.578,
      "eval_steps_per_second": 42.58,
      "step": 15000
    },
    {
      "epoch": 2.37,
      "learning_rate": 1.6239580117320163e-05,
      "loss": 0.1512,
      "step": 15500
    },
    {
      "epoch": 2.37,
      "eval_accuracy": 0.9011532125205931,
      "eval_loss": 0.3880288898944855,
      "eval_runtime": 15.9791,
      "eval_samples_per_second": 341.885,
      "eval_steps_per_second": 42.743,
      "step": 15500
    },
    {
      "epoch": 2.44,
      "learning_rate": 1.607708682016867e-05,
      "loss": 0.1624,
      "step": 16000
    },
    {
      "epoch": 2.44,
      "eval_accuracy": 0.8954786747208493,
      "eval_loss": 0.40831083059310913,
      "eval_runtime": 16.0179,
      "eval_samples_per_second": 341.056,
      "eval_steps_per_second": 42.64,
      "step": 16000
    },
    {
      "epoch": 2.52,
      "learning_rate": 1.5914593523017176e-05,
      "loss": 0.1616,
      "step": 16500
    },
    {
      "epoch": 2.52,
      "eval_accuracy": 0.9038989566172433,
      "eval_loss": 0.3950469195842743,
      "eval_runtime": 15.9181,
      "eval_samples_per_second": 343.195,
      "eval_steps_per_second": 42.907,
      "step": 16500
    },
    {
      "epoch": 2.6,
      "learning_rate": 1.5752100225865684e-05,
      "loss": 0.1587,
      "step": 17000
    },
    {
      "epoch": 2.6,
      "eval_accuracy": 0.9103056928427604,
      "eval_loss": 0.35793736577033997,
      "eval_runtime": 15.9998,
      "eval_samples_per_second": 341.441,
      "eval_steps_per_second": 42.688,
      "step": 17000
    },
    {
      "epoch": 2.67,
      "learning_rate": 1.558960692871419e-05,
      "loss": 0.1615,
      "step": 17500
    },
    {
      "epoch": 2.67,
      "eval_accuracy": 0.9011532125205931,
      "eval_loss": 0.39305415749549866,
      "eval_runtime": 15.9412,
      "eval_samples_per_second": 342.697,
      "eval_steps_per_second": 42.845,
      "step": 17500
    },
    {
      "epoch": 2.75,
      "learning_rate": 1.54271136315627e-05,
      "loss": 0.1623,
      "step": 18000
    },
    {
      "epoch": 2.75,
      "eval_accuracy": 0.90591250228812,
      "eval_loss": 0.3697090446949005,
      "eval_runtime": 15.9757,
      "eval_samples_per_second": 341.957,
      "eval_steps_per_second": 42.752,
      "step": 18000
    },
    {
      "epoch": 2.83,
      "learning_rate": 1.5264620334411207e-05,
      "loss": 0.1687,
      "step": 18500
    },
    {
      "epoch": 2.83,
      "eval_accuracy": 0.9037159070108,
      "eval_loss": 0.34730929136276245,
      "eval_runtime": 15.964,
      "eval_samples_per_second": 342.208,
      "eval_steps_per_second": 42.784,
      "step": 18500
    },
    {
      "epoch": 2.9,
      "learning_rate": 1.5102127037259715e-05,
      "loss": 0.1627,
      "step": 19000
    },
    {
      "epoch": 2.9,
      "eval_accuracy": 0.8982244188174996,
      "eval_loss": 0.3851012885570526,
      "eval_runtime": 15.9956,
      "eval_samples_per_second": 341.531,
      "eval_steps_per_second": 42.699,
      "step": 19000
    },
    {
      "epoch": 2.98,
      "learning_rate": 1.4939633740108221e-05,
      "loss": 0.1593,
      "step": 19500
    },
    {
      "epoch": 2.98,
      "eval_accuracy": 0.9018854109463664,
      "eval_loss": 0.4038921296596527,
      "eval_runtime": 15.9569,
      "eval_samples_per_second": 342.36,
      "eval_steps_per_second": 42.803,
      "step": 19500
    },
    {
      "epoch": 3.05,
      "learning_rate": 1.477714044295673e-05,
      "loss": 0.1136,
      "step": 20000
    },
    {
      "epoch": 3.05,
      "eval_accuracy": 0.9024345597656965,
      "eval_loss": 0.48348602652549744,
      "eval_runtime": 15.9762,
      "eval_samples_per_second": 341.946,
      "eval_steps_per_second": 42.751,
      "step": 20000
    },
    {
      "epoch": 3.13,
      "learning_rate": 1.4614647145805237e-05,
      "loss": 0.0965,
      "step": 20500
    },
    {
      "epoch": 3.13,
      "eval_accuracy": 0.9011532125205931,
      "eval_loss": 0.5060518383979797,
      "eval_runtime": 15.9664,
      "eval_samples_per_second": 342.155,
      "eval_steps_per_second": 42.777,
      "step": 20500
    },
    {
      "epoch": 3.21,
      "learning_rate": 1.4452153848653745e-05,
      "loss": 0.0931,
      "step": 21000
    },
    {
      "epoch": 3.21,
      "eval_accuracy": 0.8991396668497162,
      "eval_loss": 0.5278568267822266,
      "eval_runtime": 16.0118,
      "eval_samples_per_second": 341.186,
      "eval_steps_per_second": 42.656,
      "step": 21000
    },
    {
      "epoch": 3.28,
      "learning_rate": 1.4289660551502251e-05,
      "loss": 0.0993,
      "step": 21500
    },
    {
      "epoch": 3.28,
      "eval_accuracy": 0.9018854109463664,
      "eval_loss": 0.48559123277664185,
      "eval_runtime": 15.9958,
      "eval_samples_per_second": 341.526,
      "eval_steps_per_second": 42.699,
      "step": 21500
    },
    {
      "epoch": 3.36,
      "learning_rate": 1.412716725435076e-05,
      "loss": 0.1187,
      "step": 22000
    },
    {
      "epoch": 3.36,
      "eval_accuracy": 0.9031667581914699,
      "eval_loss": 0.48830872774124146,
      "eval_runtime": 16.0824,
      "eval_samples_per_second": 339.689,
      "eval_steps_per_second": 42.469,
      "step": 22000
    },
    {
      "epoch": 3.44,
      "learning_rate": 1.3964673957199267e-05,
      "loss": 0.1008,
      "step": 22500
    },
    {
      "epoch": 3.44,
      "eval_accuracy": 0.9015193117334798,
      "eval_loss": 0.5054051280021667,
      "eval_runtime": 15.9998,
      "eval_samples_per_second": 341.442,
      "eval_steps_per_second": 42.688,
      "step": 22500
    },
    {
      "epoch": 3.51,
      "learning_rate": 1.3802180660047775e-05,
      "loss": 0.1013,
      "step": 23000
    },
    {
      "epoch": 3.51,
      "eval_accuracy": 0.9022515101592532,
      "eval_loss": 0.5024679899215698,
      "eval_runtime": 16.0433,
      "eval_samples_per_second": 340.516,
      "eval_steps_per_second": 42.572,
      "step": 23000
    },
    {
      "epoch": 3.59,
      "learning_rate": 1.3639687362896281e-05,
      "loss": 0.1092,
      "step": 23500
    },
    {
      "epoch": 3.59,
      "eval_accuracy": 0.8985905180303863,
      "eval_loss": 0.44847574830055237,
      "eval_runtime": 15.9971,
      "eval_samples_per_second": 341.499,
      "eval_steps_per_second": 42.695,
      "step": 23500
    },
    {
      "epoch": 3.67,
      "learning_rate": 1.347719406574479e-05,
      "loss": 0.1135,
      "step": 24000
    },
    {
      "epoch": 3.67,
      "eval_accuracy": 0.8976752699981695,
      "eval_loss": 0.512342095375061,
      "eval_runtime": 16.034,
      "eval_samples_per_second": 340.714,
      "eval_steps_per_second": 42.597,
      "step": 24000
    },
    {
      "epoch": 3.74,
      "learning_rate": 1.3314700768593297e-05,
      "loss": 0.1042,
      "step": 24500
    },
    {
      "epoch": 3.74,
      "eval_accuracy": 0.9009701629141498,
      "eval_loss": 0.4883783161640167,
      "eval_runtime": 16.5969,
      "eval_samples_per_second": 329.158,
      "eval_steps_per_second": 41.152,
      "step": 24500
    },
    {
      "epoch": 3.82,
      "learning_rate": 1.3152207471441804e-05,
      "loss": 0.1178,
      "step": 25000
    },
    {
      "epoch": 3.82,
      "eval_accuracy": 0.900604063701263,
      "eval_loss": 0.4129727780818939,
      "eval_runtime": 16.0114,
      "eval_samples_per_second": 341.195,
      "eval_steps_per_second": 42.657,
      "step": 25000
    },
    {
      "epoch": 3.89,
      "learning_rate": 1.298971417429031e-05,
      "loss": 0.1041,
      "step": 25500
    },
    {
      "epoch": 3.89,
      "eval_accuracy": 0.8951125755079626,
      "eval_loss": 0.4847397208213806,
      "eval_runtime": 15.9759,
      "eval_samples_per_second": 341.952,
      "eval_steps_per_second": 42.752,
      "step": 25500
    },
    {
      "epoch": 3.97,
      "learning_rate": 1.282722087713882e-05,
      "loss": 0.1025,
      "step": 26000
    },
    {
      "epoch": 3.97,
      "eval_accuracy": 0.8985905180303863,
      "eval_loss": 0.4706151783466339,
      "eval_runtime": 15.9981,
      "eval_samples_per_second": 341.478,
      "eval_steps_per_second": 42.693,
      "step": 26000
    },
    {
      "epoch": 4.05,
      "learning_rate": 1.2664727579987326e-05,
      "loss": 0.0833,
      "step": 26500
    },
    {
      "epoch": 4.05,
      "eval_accuracy": 0.90536335346879,
      "eval_loss": 0.5018308162689209,
      "eval_runtime": 15.9984,
      "eval_samples_per_second": 341.472,
      "eval_steps_per_second": 42.692,
      "step": 26500
    },
    {
      "epoch": 4.12,
      "learning_rate": 1.2502234282835834e-05,
      "loss": 0.0632,
      "step": 27000
    },
    {
      "epoch": 4.12,
      "eval_accuracy": 0.9055464030752334,
      "eval_loss": 0.5425559282302856,
      "eval_runtime": 15.9902,
      "eval_samples_per_second": 341.646,
      "eval_steps_per_second": 42.714,
      "step": 27000
    },
    {
      "epoch": 4.2,
      "learning_rate": 1.233974098568434e-05,
      "loss": 0.0613,
      "step": 27500
    },
    {
      "epoch": 4.2,
      "eval_accuracy": 0.9020684605528098,
      "eval_loss": 0.5629337430000305,
      "eval_runtime": 16.0205,
      "eval_samples_per_second": 341.001,
      "eval_steps_per_second": 42.633,
      "step": 27500
    },
    {
      "epoch": 4.28,
      "learning_rate": 1.217724768853285e-05,
      "loss": 0.0703,
      "step": 28000
    },
    {
      "epoch": 4.28,
      "eval_accuracy": 0.9033498077979132,
      "eval_loss": 0.5763201713562012,
      "eval_runtime": 16.043,
      "eval_samples_per_second": 340.522,
      "eval_steps_per_second": 42.573,
      "step": 28000
    },
    {
      "epoch": 4.35,
      "learning_rate": 1.2014754391381356e-05,
      "loss": 0.0687,
      "step": 28500
    },
    {
      "epoch": 4.35,
      "eval_accuracy": 0.9038989566172433,
      "eval_loss": 0.5274228453636169,
      "eval_runtime": 16.0055,
      "eval_samples_per_second": 341.319,
      "eval_steps_per_second": 42.673,
      "step": 28500
    },
    {
      "epoch": 4.43,
      "learning_rate": 1.1852261094229864e-05,
      "loss": 0.0707,
      "step": 29000
    },
    {
      "epoch": 4.43,
      "eval_accuracy": 0.9018854109463664,
      "eval_loss": 0.5681867599487305,
      "eval_runtime": 16.0832,
      "eval_samples_per_second": 339.671,
      "eval_steps_per_second": 42.467,
      "step": 29000
    },
    {
      "epoch": 4.51,
      "learning_rate": 1.168976779707837e-05,
      "loss": 0.0717,
      "step": 29500
    },
    {
      "epoch": 4.51,
      "eval_accuracy": 0.9004210140948197,
      "eval_loss": 0.5382347702980042,
      "eval_runtime": 16.0114,
      "eval_samples_per_second": 341.195,
      "eval_steps_per_second": 42.657,
      "step": 29500
    },
    {
      "epoch": 4.58,
      "learning_rate": 1.152727449992688e-05,
      "loss": 0.0692,
      "step": 30000
    },
    {
      "epoch": 4.58,
      "eval_accuracy": 0.9009701629141498,
      "eval_loss": 0.5901069045066833,
      "eval_runtime": 15.9825,
      "eval_samples_per_second": 341.812,
      "eval_steps_per_second": 42.734,
      "step": 30000
    },
    {
      "epoch": 4.66,
      "learning_rate": 1.1364781202775386e-05,
      "loss": 0.0701,
      "step": 30500
    },
    {
      "epoch": 4.66,
      "eval_accuracy": 0.8989566172432729,
      "eval_loss": 0.5817105770111084,
      "eval_runtime": 16.0192,
      "eval_samples_per_second": 341.027,
      "eval_steps_per_second": 42.636,
      "step": 30500
    },
    {
      "epoch": 4.73,
      "learning_rate": 1.1202287905623894e-05,
      "loss": 0.0708,
      "step": 31000
    },
    {
      "epoch": 4.73,
      "eval_accuracy": 0.9018854109463664,
      "eval_loss": 0.5580092668533325,
      "eval_runtime": 16.0391,
      "eval_samples_per_second": 340.605,
      "eval_steps_per_second": 42.583,
      "step": 31000
    },
    {
      "epoch": 4.81,
      "learning_rate": 1.10397946084724e-05,
      "loss": 0.07,
      "step": 31500
    },
    {
      "epoch": 4.81,
      "eval_accuracy": 0.8989566172432729,
      "eval_loss": 0.5640453100204468,
      "eval_runtime": 15.9857,
      "eval_samples_per_second": 341.744,
      "eval_steps_per_second": 42.726,
      "step": 31500
    },
    {
      "epoch": 4.89,
      "learning_rate": 1.087730131132091e-05,
      "loss": 0.0725,
      "step": 32000
    },
    {
      "epoch": 4.89,
      "eval_accuracy": 0.8993227164561596,
      "eval_loss": 0.5767560005187988,
      "eval_runtime": 16.012,
      "eval_samples_per_second": 341.182,
      "eval_steps_per_second": 42.656,
      "step": 32000
    },
    {
      "epoch": 4.96,
      "learning_rate": 1.0714808014169416e-05,
      "loss": 0.0701,
      "step": 32500
    },
    {
      "epoch": 4.96,
      "eval_accuracy": 0.9035328574043566,
      "eval_loss": 0.5289304852485657,
      "eval_runtime": 16.0251,
      "eval_samples_per_second": 340.903,
      "eval_steps_per_second": 42.621,
      "step": 32500
    },
    {
      "epoch": 5.04,
      "learning_rate": 1.0552314717017924e-05,
      "loss": 0.0441,
      "step": 33000
    },
    {
      "epoch": 5.04,
      "eval_accuracy": 0.9009701629141498,
      "eval_loss": 0.6401296854019165,
      "eval_runtime": 16.0047,
      "eval_samples_per_second": 341.338,
      "eval_steps_per_second": 42.675,
      "step": 33000
    },
    {
      "epoch": 5.12,
      "learning_rate": 1.038982141986643e-05,
      "loss": 0.0388,
      "step": 33500
    },
    {
      "epoch": 5.12,
      "eval_accuracy": 0.8991396668497162,
      "eval_loss": 0.644571840763092,
      "eval_runtime": 15.9856,
      "eval_samples_per_second": 341.745,
      "eval_steps_per_second": 42.726,
      "step": 33500
    },
    {
      "epoch": 5.19,
      "learning_rate": 1.022732812271494e-05,
      "loss": 0.0417,
      "step": 34000
    },
    {
      "epoch": 5.19,
      "eval_accuracy": 0.9038989566172433,
      "eval_loss": 0.6326994299888611,
      "eval_runtime": 15.9965,
      "eval_samples_per_second": 341.512,
      "eval_steps_per_second": 42.697,
      "step": 34000
    },
    {
      "epoch": 5.27,
      "learning_rate": 1.0064834825563446e-05,
      "loss": 0.039,
      "step": 34500
    },
    {
      "epoch": 5.27,
      "eval_accuracy": 0.90481420464946,
      "eval_loss": 0.6385270357131958,
      "eval_runtime": 15.9731,
      "eval_samples_per_second": 342.013,
      "eval_steps_per_second": 42.759,
      "step": 34500
    },
    {
      "epoch": 5.35,
      "learning_rate": 9.902341528411954e-06,
      "loss": 0.0407,
      "step": 35000
    },
    {
      "epoch": 5.35,
      "eval_accuracy": 0.9029837085850265,
      "eval_loss": 0.6509894728660583,
      "eval_runtime": 16.0054,
      "eval_samples_per_second": 341.322,
      "eval_steps_per_second": 42.673,
      "step": 35000
    },
    {
      "epoch": 5.42,
      "learning_rate": 9.739848231260461e-06,
      "loss": 0.0446,
      "step": 35500
    },
    {
      "epoch": 5.42,
      "eval_accuracy": 0.9029837085850265,
      "eval_loss": 0.5787823796272278,
      "eval_runtime": 16.0265,
      "eval_samples_per_second": 340.872,
      "eval_steps_per_second": 42.617,
      "step": 35500
    },
    {
      "epoch": 5.5,
      "learning_rate": 9.57735493410897e-06,
      "loss": 0.0422,
      "step": 36000
    },
    {
      "epoch": 5.5,
      "eval_accuracy": 0.9024345597656965,
      "eval_loss": 0.6722646951675415,
      "eval_runtime": 16.1928,
      "eval_samples_per_second": 337.372,
      "eval_steps_per_second": 42.179,
      "step": 36000
    },
    {
      "epoch": 5.58,
      "learning_rate": 9.414861636957477e-06,
      "loss": 0.04,
      "step": 36500
    },
    {
      "epoch": 5.58,
      "eval_accuracy": 0.9037159070108,
      "eval_loss": 0.6601631045341492,
      "eval_runtime": 15.9359,
      "eval_samples_per_second": 342.812,
      "eval_steps_per_second": 42.859,
      "step": 36500
    },
    {
      "epoch": 5.65,
      "learning_rate": 9.252368339805983e-06,
      "loss": 0.0514,
      "step": 37000
    },
    {
      "epoch": 5.65,
      "eval_accuracy": 0.9024345597656965,
      "eval_loss": 0.6407366394996643,
      "eval_runtime": 16.0327,
      "eval_samples_per_second": 340.742,
      "eval_steps_per_second": 42.601,
      "step": 37000
    },
    {
      "epoch": 5.73,
      "learning_rate": 9.089875042654491e-06,
      "loss": 0.0462,
      "step": 37500
    },
    {
      "epoch": 5.73,
      "eval_accuracy": 0.90481420464946,
      "eval_loss": 0.6144635677337646,
      "eval_runtime": 16.0227,
      "eval_samples_per_second": 340.955,
      "eval_steps_per_second": 42.627,
      "step": 37500
    },
    {
      "epoch": 5.8,
      "learning_rate": 8.927381745502999e-06,
      "loss": 0.0479,
      "step": 38000
    },
    {
      "epoch": 5.8,
      "eval_accuracy": 0.9007871133077063,
      "eval_loss": 0.5881094336509705,
      "eval_runtime": 16.0441,
      "eval_samples_per_second": 340.5,
      "eval_steps_per_second": 42.57,
      "step": 38000
    },
    {
      "epoch": 5.88,
      "learning_rate": 8.764888448351507e-06,
      "loss": 0.0503,
      "step": 38500
    },
    {
      "epoch": 5.88,
      "eval_accuracy": 0.9007871133077063,
      "eval_loss": 0.6001153588294983,
      "eval_runtime": 15.9844,
      "eval_samples_per_second": 341.77,
      "eval_steps_per_second": 42.729,
      "step": 38500
    },
    {
      "epoch": 5.96,
      "learning_rate": 8.602395151200013e-06,
      "loss": 0.0385,
      "step": 39000
    },
    {
      "epoch": 5.96,
      "eval_accuracy": 0.9051803038623467,
      "eval_loss": 0.6463525295257568,
      "eval_runtime": 16.0778,
      "eval_samples_per_second": 339.785,
      "eval_steps_per_second": 42.481,
      "step": 39000
    },
    {
      "epoch": 6.03,
      "learning_rate": 8.439901854048521e-06,
      "loss": 0.0436,
      "step": 39500
    },
    {
      "epoch": 6.03,
      "eval_accuracy": 0.9038989566172433,
      "eval_loss": 0.6683228611946106,
      "eval_runtime": 15.9874,
      "eval_samples_per_second": 341.707,
      "eval_steps_per_second": 42.721,
      "step": 39500
    },
    {
      "epoch": 6.11,
      "learning_rate": 8.277408556897029e-06,
      "loss": 0.0296,
      "step": 40000
    },
    {
      "epoch": 6.11,
      "eval_accuracy": 0.9037159070108,
      "eval_loss": 0.696530282497406,
      "eval_runtime": 15.9785,
      "eval_samples_per_second": 341.896,
      "eval_steps_per_second": 42.745,
      "step": 40000
    },
    {
      "epoch": 6.19,
      "learning_rate": 8.114915259745537e-06,
      "loss": 0.0275,
      "step": 40500
    },
    {
      "epoch": 6.19,
      "eval_accuracy": 0.9038989566172433,
      "eval_loss": 0.7192733883857727,
      "eval_runtime": 15.9602,
      "eval_samples_per_second": 342.288,
      "eval_steps_per_second": 42.794,
      "step": 40500
    },
    {
      "epoch": 6.26,
      "learning_rate": 7.952421962594043e-06,
      "loss": 0.0258,
      "step": 41000
    },
    {
      "epoch": 6.26,
      "eval_accuracy": 0.9038989566172433,
      "eval_loss": 0.7229453921318054,
      "eval_runtime": 15.9822,
      "eval_samples_per_second": 341.817,
      "eval_steps_per_second": 42.735,
      "step": 41000
    },
    {
      "epoch": 6.34,
      "learning_rate": 7.789928665442551e-06,
      "loss": 0.0223,
      "step": 41500
    },
    {
      "epoch": 6.34,
      "eval_accuracy": 0.9073768991396669,
      "eval_loss": 0.697367250919342,
      "eval_runtime": 15.9944,
      "eval_samples_per_second": 341.557,
      "eval_steps_per_second": 42.702,
      "step": 41500
    },
    {
      "epoch": 6.42,
      "learning_rate": 7.627435368291059e-06,
      "loss": 0.0268,
      "step": 42000
    },
    {
      "epoch": 6.42,
      "eval_accuracy": 0.9028006589785832,
      "eval_loss": 0.704451858997345,
      "eval_runtime": 16.0075,
      "eval_samples_per_second": 341.277,
      "eval_steps_per_second": 42.667,
      "step": 42000
    },
    {
      "epoch": 6.49,
      "learning_rate": 7.464942071139566e-06,
      "loss": 0.0297,
      "step": 42500
    },
    {
      "epoch": 6.49,
      "eval_accuracy": 0.9055464030752334,
      "eval_loss": 0.72894686460495,
      "eval_runtime": 15.9896,
      "eval_samples_per_second": 341.659,
      "eval_steps_per_second": 42.715,
      "step": 42500
    },
    {
      "epoch": 6.57,
      "learning_rate": 7.302448773988074e-06,
      "loss": 0.0295,
      "step": 43000
    },
    {
      "epoch": 6.57,
      "eval_accuracy": 0.9049972542559034,
      "eval_loss": 0.6809965372085571,
      "eval_runtime": 16.0,
      "eval_samples_per_second": 341.438,
      "eval_steps_per_second": 42.688,
      "step": 43000
    },
    {
      "epoch": 6.64,
      "learning_rate": 7.139955476836581e-06,
      "loss": 0.0265,
      "step": 43500
    },
    {
      "epoch": 6.64,
      "eval_accuracy": 0.9064616511074501,
      "eval_loss": 0.6833410263061523,
      "eval_runtime": 15.9737,
      "eval_samples_per_second": 341.999,
      "eval_steps_per_second": 42.758,
      "step": 43500
    },
    {
      "epoch": 6.72,
      "learning_rate": 6.9774621796850885e-06,
      "loss": 0.0268,
      "step": 44000
    },
    {
      "epoch": 6.72,
      "eval_accuracy": 0.9035328574043566,
      "eval_loss": 0.7155495882034302,
      "eval_runtime": 16.0634,
      "eval_samples_per_second": 340.089,
      "eval_steps_per_second": 42.519,
      "step": 44000
    },
    {
      "epoch": 6.8,
      "learning_rate": 6.8149688825335956e-06,
      "loss": 0.0293,
      "step": 44500
    },
    {
      "epoch": 6.8,
      "eval_accuracy": 0.9015193117334798,
      "eval_loss": 0.7632263898849487,
      "eval_runtime": 15.9819,
      "eval_samples_per_second": 341.824,
      "eval_steps_per_second": 42.736,
      "step": 44500
    },
    {
      "epoch": 6.87,
      "learning_rate": 6.6524755853821034e-06,
      "loss": 0.0289,
      "step": 45000
    },
    {
      "epoch": 6.87,
      "eval_accuracy": 0.9033498077979132,
      "eval_loss": 0.7229273319244385,
      "eval_runtime": 16.0622,
      "eval_samples_per_second": 340.116,
      "eval_steps_per_second": 42.522,
      "step": 45000
    },
    {
      "epoch": 6.95,
      "learning_rate": 6.4899822882306105e-06,
      "loss": 0.0286,
      "step": 45500
    },
    {
      "epoch": 6.95,
      "eval_accuracy": 0.90536335346879,
      "eval_loss": 0.6670674681663513,
      "eval_runtime": 15.9524,
      "eval_samples_per_second": 342.456,
      "eval_steps_per_second": 42.815,
      "step": 45500
    },
    {
      "epoch": 7.03,
      "learning_rate": 6.327488991079118e-06,
      "loss": 0.024,
      "step": 46000
    },
    {
      "epoch": 7.03,
      "eval_accuracy": 0.9033498077979132,
      "eval_loss": 0.7658922672271729,
      "eval_runtime": 16.0192,
      "eval_samples_per_second": 341.028,
      "eval_steps_per_second": 42.636,
      "step": 46000
    },
    {
      "epoch": 7.1,
      "learning_rate": 6.164995693927625e-06,
      "loss": 0.0141,
      "step": 46500
    },
    {
      "epoch": 7.1,
      "eval_accuracy": 0.9022515101592532,
      "eval_loss": 0.7980563044548035,
      "eval_runtime": 16.0395,
      "eval_samples_per_second": 340.597,
      "eval_steps_per_second": 42.582,
      "step": 46500
    },
    {
      "epoch": 7.18,
      "learning_rate": 6.002502396776133e-06,
      "loss": 0.0212,
      "step": 47000
    },
    {
      "epoch": 7.18,
      "eval_accuracy": 0.9088412959912137,
      "eval_loss": 0.7021394968032837,
      "eval_runtime": 15.9988,
      "eval_samples_per_second": 341.463,
      "eval_steps_per_second": 42.691,
      "step": 47000
    },
    {
      "epoch": 7.26,
      "learning_rate": 5.84000909962464e-06,
      "loss": 0.0183,
      "step": 47500
    },
    {
      "epoch": 7.26,
      "eval_accuracy": 0.909573494416987,
      "eval_loss": 0.7122032046318054,
      "eval_runtime": 16.0037,
      "eval_samples_per_second": 341.358,
      "eval_steps_per_second": 42.678,
      "step": 47500
    },
    {
      "epoch": 7.33,
      "learning_rate": 5.677515802473148e-06,
      "loss": 0.0221,
      "step": 48000
    },
    {
      "epoch": 7.33,
      "eval_accuracy": 0.9064616511074501,
      "eval_loss": 0.7080280780792236,
      "eval_runtime": 16.0516,
      "eval_samples_per_second": 340.339,
      "eval_steps_per_second": 42.55,
      "step": 48000
    },
    {
      "epoch": 7.41,
      "learning_rate": 5.515022505321655e-06,
      "loss": 0.0146,
      "step": 48500
    },
    {
      "epoch": 7.41,
      "eval_accuracy": 0.9073768991396669,
      "eval_loss": 0.7343564033508301,
      "eval_runtime": 16.0182,
      "eval_samples_per_second": 341.049,
      "eval_steps_per_second": 42.639,
      "step": 48500
    },
    {
      "epoch": 7.48,
      "learning_rate": 5.352529208170163e-06,
      "loss": 0.0181,
      "step": 49000
    },
    {
      "epoch": 7.48,
      "eval_accuracy": 0.9104887424492037,
      "eval_loss": 0.7273329496383667,
      "eval_runtime": 15.9794,
      "eval_samples_per_second": 341.877,
      "eval_steps_per_second": 42.742,
      "step": 49000
    },
    {
      "epoch": 7.56,
      "learning_rate": 5.19003591101867e-06,
      "loss": 0.0161,
      "step": 49500
    },
    {
      "epoch": 7.56,
      "eval_accuracy": 0.9082921471718836,
      "eval_loss": 0.73320472240448,
      "eval_runtime": 15.9747,
      "eval_samples_per_second": 341.978,
      "eval_steps_per_second": 42.755,
      "step": 49500
    },
    {
      "epoch": 7.64,
      "learning_rate": 5.027542613867178e-06,
      "loss": 0.0193,
      "step": 50000
    },
    {
      "epoch": 7.64,
      "eval_accuracy": 0.9093904448105437,
      "eval_loss": 0.7117404937744141,
      "eval_runtime": 16.028,
      "eval_samples_per_second": 340.841,
      "eval_steps_per_second": 42.613,
      "step": 50000
    },
    {
      "epoch": 7.71,
      "learning_rate": 4.865049316715686e-06,
      "loss": 0.0165,
      "step": 50500
    },
    {
      "epoch": 7.71,
      "eval_accuracy": 0.9070107999267801,
      "eval_loss": 0.7796688675880432,
      "eval_runtime": 16.0196,
      "eval_samples_per_second": 341.02,
      "eval_steps_per_second": 42.635,
      "step": 50500
    },
    {
      "epoch": 7.79,
      "learning_rate": 4.702556019564194e-06,
      "loss": 0.0173,
      "step": 51000
    },
    {
      "epoch": 7.79,
      "eval_accuracy": 0.9106717920556471,
      "eval_loss": 0.7127623558044434,
      "eval_runtime": 16.0489,
      "eval_samples_per_second": 340.398,
      "eval_steps_per_second": 42.558,
      "step": 51000
    },
    {
      "epoch": 7.87,
      "learning_rate": 4.540062722412701e-06,
      "loss": 0.0158,
      "step": 51500
    },
    {
      "epoch": 7.87,
      "eval_accuracy": 0.9110378912685337,
      "eval_loss": 0.7419800758361816,
      "eval_runtime": 15.9669,
      "eval_samples_per_second": 342.144,
      "eval_steps_per_second": 42.776,
      "step": 51500
    },
    {
      "epoch": 7.94,
      "learning_rate": 4.377569425261209e-06,
      "loss": 0.0157,
      "step": 52000
    },
    {
      "epoch": 7.94,
      "eval_accuracy": 0.9128683873329673,
      "eval_loss": 0.7197983860969543,
      "eval_runtime": 16.0117,
      "eval_samples_per_second": 341.189,
      "eval_steps_per_second": 42.656,
      "step": 52000
    },
    {
      "epoch": 8.02,
      "learning_rate": 4.215076128109716e-06,
      "loss": 0.0142,
      "step": 52500
    },
    {
      "epoch": 8.02,
      "eval_accuracy": 0.9114039904814205,
      "eval_loss": 0.7032836675643921,
      "eval_runtime": 15.9817,
      "eval_samples_per_second": 341.829,
      "eval_steps_per_second": 42.736,
      "step": 52500
    },
    {
      "epoch": 8.1,
      "learning_rate": 4.0525828309582235e-06,
      "loss": 0.0107,
      "step": 53000
    },
    {
      "epoch": 8.1,
      "eval_accuracy": 0.9108548416620904,
      "eval_loss": 0.776072084903717,
      "eval_runtime": 16.015,
      "eval_samples_per_second": 341.118,
      "eval_steps_per_second": 42.648,
      "step": 53000
    },
    {
      "epoch": 8.17,
      "learning_rate": 3.8900895338067306e-06,
      "loss": 0.0122,
      "step": 53500
    },
    {
      "epoch": 8.17,
      "eval_accuracy": 0.9110378912685337,
      "eval_loss": 0.7777400612831116,
      "eval_runtime": 15.947,
      "eval_samples_per_second": 342.572,
      "eval_steps_per_second": 42.829,
      "step": 53500
    },
    {
      "epoch": 8.25,
      "learning_rate": 3.727596236655238e-06,
      "loss": 0.013,
      "step": 54000
    },
    {
      "epoch": 8.25,
      "eval_accuracy": 0.9075599487461102,
      "eval_loss": 0.7880035042762756,
      "eval_runtime": 15.9944,
      "eval_samples_per_second": 341.558,
      "eval_steps_per_second": 42.703,
      "step": 54000
    },
    {
      "epoch": 8.32,
      "learning_rate": 3.5651029395037455e-06,
      "loss": 0.0092,
      "step": 54500
    },
    {
      "epoch": 8.32,
      "eval_accuracy": 0.9073768991396669,
      "eval_loss": 0.8070389032363892,
      "eval_runtime": 15.9764,
      "eval_samples_per_second": 341.942,
      "eval_steps_per_second": 42.751,
      "step": 54500
    },
    {
      "epoch": 8.4,
      "learning_rate": 3.402609642352253e-06,
      "loss": 0.0102,
      "step": 55000
    },
    {
      "epoch": 8.4,
      "eval_accuracy": 0.9055464030752334,
      "eval_loss": 0.8112868666648865,
      "eval_runtime": 16.2072,
      "eval_samples_per_second": 337.073,
      "eval_steps_per_second": 42.142,
      "step": 55000
    },
    {
      "epoch": 8.48,
      "learning_rate": 3.2401163452007604e-06,
      "loss": 0.0099,
      "step": 55500
    },
    {
      "epoch": 8.48,
      "eval_accuracy": 0.9079260479589969,
      "eval_loss": 0.8152795433998108,
      "eval_runtime": 16.1355,
      "eval_samples_per_second": 338.57,
      "eval_steps_per_second": 42.329,
      "step": 55500
    },
    {
      "epoch": 8.55,
      "learning_rate": 3.0776230480492682e-06,
      "loss": 0.0087,
      "step": 56000
    },
    {
      "epoch": 8.55,
      "eval_accuracy": 0.9112209408749771,
      "eval_loss": 0.8045311570167542,
      "eval_runtime": 16.0534,
      "eval_samples_per_second": 340.302,
      "eval_steps_per_second": 42.546,
      "step": 56000
    },
    {
      "epoch": 8.63,
      "learning_rate": 2.9151297508977757e-06,
      "loss": 0.0133,
      "step": 56500
    },
    {
      "epoch": 8.63,
      "eval_accuracy": 0.9081090975654402,
      "eval_loss": 0.8173390030860901,
      "eval_runtime": 15.9908,
      "eval_samples_per_second": 341.634,
      "eval_steps_per_second": 42.712,
      "step": 56500
    },
    {
      "epoch": 8.71,
      "learning_rate": 2.752636453746283e-06,
      "loss": 0.017,
      "step": 57000
    },
    {
      "epoch": 8.71,
      "eval_accuracy": 0.9101226432363171,
      "eval_loss": 0.7646049857139587,
      "eval_runtime": 16.0101,
      "eval_samples_per_second": 341.222,
      "eval_steps_per_second": 42.661,
      "step": 57000
    },
    {
      "epoch": 8.78,
      "learning_rate": 2.5901431565947906e-06,
      "loss": 0.0093,
      "step": 57500
    },
    {
      "epoch": 8.78,
      "eval_accuracy": 0.9119531393007505,
      "eval_loss": 0.7681165337562561,
      "eval_runtime": 15.9413,
      "eval_samples_per_second": 342.694,
      "eval_steps_per_second": 42.845,
      "step": 57500
    },
    {
      "epoch": 8.86,
      "learning_rate": 2.427649859443298e-06,
      "loss": 0.0085,
      "step": 58000
    },
    {
      "epoch": 8.86,
      "eval_accuracy": 0.9082921471718836,
      "eval_loss": 0.8067189455032349,
      "eval_runtime": 16.0314,
      "eval_samples_per_second": 340.768,
      "eval_steps_per_second": 42.604,
      "step": 58000
    },
    {
      "epoch": 8.94,
      "learning_rate": 2.2651565622918055e-06,
      "loss": 0.0137,
      "step": 58500
    },
    {
      "epoch": 8.94,
      "eval_accuracy": 0.9106717920556471,
      "eval_loss": 0.7645466327667236,
      "eval_runtime": 15.9969,
      "eval_samples_per_second": 341.504,
      "eval_steps_per_second": 42.696,
      "step": 58500
    },
    {
      "epoch": 9.01,
      "learning_rate": 2.102663265140313e-06,
      "loss": 0.0099,
      "step": 59000
    },
    {
      "epoch": 9.01,
      "eval_accuracy": 0.9123192385136372,
      "eval_loss": 0.7742260098457336,
      "eval_runtime": 16.0529,
      "eval_samples_per_second": 340.313,
      "eval_steps_per_second": 42.547,
      "step": 59000
    },
    {
      "epoch": 9.09,
      "learning_rate": 1.9401699679888204e-06,
      "loss": 0.0074,
      "step": 59500
    },
    {
      "epoch": 9.09,
      "eval_accuracy": 0.90591250228812,
      "eval_loss": 0.8052135705947876,
      "eval_runtime": 15.9958,
      "eval_samples_per_second": 341.527,
      "eval_steps_per_second": 42.699,
      "step": 59500
    },
    {
      "epoch": 9.16,
      "learning_rate": 1.777676670837328e-06,
      "loss": 0.0057,
      "step": 60000
    },
    {
      "epoch": 9.16,
      "eval_accuracy": 0.9132344865458539,
      "eval_loss": 0.790639340877533,
      "eval_runtime": 16.0441,
      "eval_samples_per_second": 340.499,
      "eval_steps_per_second": 42.57,
      "step": 60000
    },
    {
      "epoch": 9.24,
      "learning_rate": 1.6151833736858355e-06,
      "loss": 0.0046,
      "step": 60500
    },
    {
      "epoch": 9.24,
      "eval_accuracy": 0.9136005857587406,
      "eval_loss": 0.7914955019950867,
      "eval_runtime": 15.9828,
      "eval_samples_per_second": 341.806,
      "eval_steps_per_second": 42.734,
      "step": 60500
    },
    {
      "epoch": 9.32,
      "learning_rate": 1.452690076534343e-06,
      "loss": 0.0036,
      "step": 61000
    },
    {
      "epoch": 9.32,
      "eval_accuracy": 0.9101226432363171,
      "eval_loss": 0.8114162087440491,
      "eval_runtime": 16.0918,
      "eval_samples_per_second": 339.489,
      "eval_steps_per_second": 42.444,
      "step": 61000
    },
    {
      "epoch": 9.39,
      "learning_rate": 1.2901967793828504e-06,
      "loss": 0.0036,
      "step": 61500
    },
    {
      "epoch": 9.39,
      "eval_accuracy": 0.9086582463847702,
      "eval_loss": 0.8381993174552917,
      "eval_runtime": 16.0882,
      "eval_samples_per_second": 339.565,
      "eval_steps_per_second": 42.453,
      "step": 61500
    },
    {
      "epoch": 9.47,
      "learning_rate": 1.127703482231358e-06,
      "loss": 0.0079,
      "step": 62000
    },
    {
      "epoch": 9.47,
      "eval_accuracy": 0.9110378912685337,
      "eval_loss": 0.8012856841087341,
      "eval_runtime": 16.0031,
      "eval_samples_per_second": 341.371,
      "eval_steps_per_second": 42.679,
      "step": 62000
    },
    {
      "epoch": 9.55,
      "learning_rate": 9.652101850798656e-07,
      "loss": 0.0079,
      "step": 62500
    },
    {
      "epoch": 9.55,
      "eval_accuracy": 0.9099395936298736,
      "eval_loss": 0.8116302490234375,
      "eval_runtime": 16.0181,
      "eval_samples_per_second": 341.052,
      "eval_steps_per_second": 42.639,
      "step": 62500
    },
    {
      "epoch": 9.62,
      "learning_rate": 8.02716887928373e-07,
      "loss": 0.0054,
      "step": 63000
    },
    {
      "epoch": 9.62,
      "eval_accuracy": 0.9110378912685337,
      "eval_loss": 0.8142715096473694,
      "eval_runtime": 16.0228,
      "eval_samples_per_second": 340.951,
      "eval_steps_per_second": 42.627,
      "step": 63000
    },
    {
      "epoch": 9.7,
      "learning_rate": 6.402235907768806e-07,
      "loss": 0.0036,
      "step": 63500
    },
    {
      "epoch": 9.7,
      "eval_accuracy": 0.9114039904814205,
      "eval_loss": 0.8137949109077454,
      "eval_runtime": 16.0052,
      "eval_samples_per_second": 341.327,
      "eval_steps_per_second": 42.674,
      "step": 63500
    },
    {
      "epoch": 9.78,
      "learning_rate": 4.777302936253879e-07,
      "loss": 0.0053,
      "step": 64000
    },
    {
      "epoch": 9.78,
      "eval_accuracy": 0.9106717920556471,
      "eval_loss": 0.8157973289489746,
      "eval_runtime": 16.048,
      "eval_samples_per_second": 340.415,
      "eval_steps_per_second": 42.56,
      "step": 64000
    },
    {
      "epoch": 9.85,
      "learning_rate": 3.1523699647389547e-07,
      "loss": 0.0035,
      "step": 64500
    },
    {
      "epoch": 9.85,
      "eval_accuracy": 0.9112209408749771,
      "eval_loss": 0.814604640007019,
      "eval_runtime": 16.0042,
      "eval_samples_per_second": 341.348,
      "eval_steps_per_second": 42.676,
      "step": 64500
    },
    {
      "epoch": 9.93,
      "learning_rate": 1.5274369932240295e-07,
      "loss": 0.0051,
      "step": 65000
    },
    {
      "epoch": 9.93,
      "eval_accuracy": 0.9106717920556471,
      "eval_loss": 0.8167830109596252,
      "eval_runtime": 16.0336,
      "eval_samples_per_second": 340.722,
      "eval_steps_per_second": 42.598,
      "step": 65000
    },
    {
      "epoch": 10.0,
      "step": 65470,
      "total_flos": 6.88976031789312e+16,
      "train_loss": 0.10566671685770447,
      "train_runtime": 14393.9986,
      "train_samples_per_second": 72.769,
      "train_steps_per_second": 4.548
    }
  ],
  "max_steps": 65470,
  "num_train_epochs": 10,
  "total_flos": 6.88976031789312e+16,
  "trial_name": null,
  "trial_params": null
}