{
  "best_metric": 0.17145082354545593,
  "best_model_checkpoint": "/kaggle/working/ckpts/checkpoint-3500",
  "epoch": 14.969696969696969,
  "eval_steps": 500,
  "global_step": 3705,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.4,
      "grad_norm": 2.2636642456054688,
      "learning_rate": 9.730094466936572e-06,
      "loss": 1.5236,
      "step": 100
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.942427396774292,
      "learning_rate": 9.460188933873145e-06,
      "loss": 1.2036,
      "step": 200
    },
    {
      "epoch": 1.21,
      "grad_norm": 4.616508960723877,
      "learning_rate": 9.192982456140351e-06,
      "loss": 0.8129,
      "step": 300
    },
    {
      "epoch": 1.62,
      "grad_norm": 6.923274517059326,
      "learning_rate": 8.923076923076925e-06,
      "loss": 0.603,
      "step": 400
    },
    {
      "epoch": 2.02,
      "grad_norm": 14.605295181274414,
      "learning_rate": 8.65587044534413e-06,
      "loss": 0.554,
      "step": 500
    },
    {
      "epoch": 2.02,
      "eval_accuracy": 0.9151515364646912,
      "eval_loss": 0.29042723774909973,
      "eval_runtime": 22.943,
      "eval_samples_per_second": 21.575,
      "eval_steps_per_second": 5.405,
      "step": 500
    },
    {
      "epoch": 2.42,
      "grad_norm": 8.531092643737793,
      "learning_rate": 8.385964912280704e-06,
      "loss": 0.4044,
      "step": 600
    },
    {
      "epoch": 2.83,
      "grad_norm": 17.11962890625,
      "learning_rate": 8.118758434547908e-06,
      "loss": 0.3846,
      "step": 700
    },
    {
      "epoch": 3.23,
      "grad_norm": 40.550960540771484,
      "learning_rate": 7.848852901484481e-06,
      "loss": 0.3334,
      "step": 800
    },
    {
      "epoch": 3.64,
      "grad_norm": 0.589801549911499,
      "learning_rate": 7.578947368421054e-06,
      "loss": 0.2088,
      "step": 900
    },
    {
      "epoch": 4.04,
      "grad_norm": 0.7292308211326599,
      "learning_rate": 7.309041835357625e-06,
      "loss": 0.2241,
      "step": 1000
    },
    {
      "epoch": 4.04,
      "eval_accuracy": 0.931313157081604,
      "eval_loss": 0.2854190170764923,
      "eval_runtime": 23.1163,
      "eval_samples_per_second": 21.413,
      "eval_steps_per_second": 5.364,
      "step": 1000
    },
    {
      "epoch": 4.44,
      "grad_norm": 4.302298069000244,
      "learning_rate": 7.039136302294197e-06,
      "loss": 0.209,
      "step": 1100
    },
    {
      "epoch": 4.85,
      "grad_norm": 8.639167785644531,
      "learning_rate": 6.76923076923077e-06,
      "loss": 0.2459,
      "step": 1200
    },
    {
      "epoch": 5.25,
      "grad_norm": 14.07341194152832,
      "learning_rate": 6.502024291497976e-06,
      "loss": 0.2448,
      "step": 1300
    },
    {
      "epoch": 5.66,
      "grad_norm": 25.380226135253906,
      "learning_rate": 6.232118758434549e-06,
      "loss": 0.2045,
      "step": 1400
    },
    {
      "epoch": 6.06,
      "grad_norm": 20.343219757080078,
      "learning_rate": 5.962213225371121e-06,
      "loss": 0.1422,
      "step": 1500
    },
    {
      "epoch": 6.06,
      "eval_accuracy": 0.9333333373069763,
      "eval_loss": 0.31245771050453186,
      "eval_runtime": 23.007,
      "eval_samples_per_second": 21.515,
      "eval_steps_per_second": 5.39,
      "step": 1500
    },
    {
      "epoch": 6.46,
      "grad_norm": 48.67534255981445,
      "learning_rate": 5.692307692307692e-06,
      "loss": 0.1553,
      "step": 1600
    },
    {
      "epoch": 6.87,
      "grad_norm": 0.05280930921435356,
      "learning_rate": 5.4224021592442655e-06,
      "loss": 0.1939,
      "step": 1700
    },
    {
      "epoch": 7.27,
      "grad_norm": 2.396838426589966,
      "learning_rate": 5.152496626180837e-06,
      "loss": 0.1421,
      "step": 1800
    },
    {
      "epoch": 7.68,
      "grad_norm": 0.47542768716812134,
      "learning_rate": 4.882591093117409e-06,
      "loss": 0.1711,
      "step": 1900
    },
    {
      "epoch": 8.08,
      "grad_norm": 24.91346549987793,
      "learning_rate": 4.6126855600539814e-06,
      "loss": 0.1245,
      "step": 2000
    },
    {
      "epoch": 8.08,
      "eval_accuracy": 0.9636363387107849,
      "eval_loss": 0.18670035898685455,
      "eval_runtime": 23.2005,
      "eval_samples_per_second": 21.336,
      "eval_steps_per_second": 5.345,
      "step": 2000
    },
    {
      "epoch": 8.48,
      "grad_norm": 103.44571685791016,
      "learning_rate": 4.342780026990554e-06,
      "loss": 0.1178,
      "step": 2100
    },
    {
      "epoch": 8.89,
      "grad_norm": 3.599931478500366,
      "learning_rate": 4.072874493927126e-06,
      "loss": 0.1012,
      "step": 2200
    },
    {
      "epoch": 9.29,
      "grad_norm": 13.552335739135742,
      "learning_rate": 3.8029689608636982e-06,
      "loss": 0.1241,
      "step": 2300
    },
    {
      "epoch": 9.7,
      "grad_norm": 0.7136185169219971,
      "learning_rate": 3.53306342780027e-06,
      "loss": 0.1054,
      "step": 2400
    },
    {
      "epoch": 10.1,
      "grad_norm": 0.019901130348443985,
      "learning_rate": 3.2631578947368423e-06,
      "loss": 0.0912,
      "step": 2500
    },
    {
      "epoch": 10.1,
      "eval_accuracy": 0.9696969985961914,
      "eval_loss": 0.1746375858783722,
      "eval_runtime": 22.9664,
      "eval_samples_per_second": 21.553,
      "eval_steps_per_second": 5.399,
      "step": 2500
    },
    {
      "epoch": 10.51,
      "grad_norm": 0.025977592915296555,
      "learning_rate": 2.9932523616734146e-06,
      "loss": 0.0672,
      "step": 2600
    },
    {
      "epoch": 10.91,
      "grad_norm": 0.022668780758976936,
      "learning_rate": 2.723346828609987e-06,
      "loss": 0.1198,
      "step": 2700
    },
    {
      "epoch": 11.31,
      "grad_norm": 18.671525955200195,
      "learning_rate": 2.453441295546559e-06,
      "loss": 0.0758,
      "step": 2800
    },
    {
      "epoch": 11.72,
      "grad_norm": 10.470223426818848,
      "learning_rate": 2.183535762483131e-06,
      "loss": 0.1329,
      "step": 2900
    },
    {
      "epoch": 12.12,
      "grad_norm": 0.030979245901107788,
      "learning_rate": 1.913630229419703e-06,
      "loss": 0.0599,
      "step": 3000
    },
    {
      "epoch": 12.12,
      "eval_accuracy": 0.9595959782600403,
      "eval_loss": 0.21210218966007233,
      "eval_runtime": 22.8997,
      "eval_samples_per_second": 21.616,
      "eval_steps_per_second": 5.415,
      "step": 3000
    },
    {
      "epoch": 12.53,
      "grad_norm": 0.3461657166481018,
      "learning_rate": 1.6437246963562757e-06,
      "loss": 0.0821,
      "step": 3100
    },
    {
      "epoch": 12.93,
      "grad_norm": 38.45835876464844,
      "learning_rate": 1.3738191632928477e-06,
      "loss": 0.0919,
      "step": 3200
    },
    {
      "epoch": 13.33,
      "grad_norm": 0.8387115001678467,
      "learning_rate": 1.106612685560054e-06,
      "loss": 0.0814,
      "step": 3300
    },
    {
      "epoch": 13.74,
      "grad_norm": 0.22132369875907898,
      "learning_rate": 8.367071524966262e-07,
      "loss": 0.0954,
      "step": 3400
    },
    {
      "epoch": 14.14,
      "grad_norm": 0.05546307936310768,
      "learning_rate": 5.668016194331984e-07,
      "loss": 0.041,
      "step": 3500
    },
    {
      "epoch": 14.14,
      "eval_accuracy": 0.973737359046936,
      "eval_loss": 0.17145082354545593,
      "eval_runtime": 23.372,
      "eval_samples_per_second": 21.179,
      "eval_steps_per_second": 5.305,
      "step": 3500
    },
    {
      "epoch": 14.55,
      "grad_norm": 17.49441146850586,
      "learning_rate": 2.968960863697706e-07,
      "loss": 0.0978,
      "step": 3600
    },
    {
      "epoch": 14.95,
      "grad_norm": 0.08890076726675034,
      "learning_rate": 2.6990553306342783e-08,
      "loss": 0.0463,
      "step": 3700
    },
    {
      "epoch": 14.97,
      "step": 3705,
      "total_flos": 9.492004863262944e+17,
      "train_loss": 0.2598070231043858,
      "train_runtime": 2003.5076,
      "train_samples_per_second": 14.817,
      "train_steps_per_second": 1.849
    }
  ],
  "logging_steps": 100,
  "max_steps": 3705,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 15,
  "save_steps": 500,
  "total_flos": 9.492004863262944e+17,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}