{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.7433155080213902,
  "eval_steps": 300,
  "global_step": 3500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.21390374331550802,
      "grad_norm": 4.510887622833252,
      "learning_rate": 9.95e-05,
      "loss": 0.7339,
      "mean_token_accuracy": 0.8075576078891754,
      "num_tokens": 292762.0,
      "step": 200
    },
    {
      "epoch": 0.32085561497326204,
      "eval_loss": 0.9760668277740479,
      "eval_mean_token_accuracy": 0.7563846576213836,
      "eval_num_tokens": 441378.0,
      "eval_runtime": 3.5124,
      "eval_samples_per_second": 112.459,
      "eval_steps_per_second": 14.235,
      "step": 300
    },
    {
      "epoch": 0.42780748663101603,
      "grad_norm": 5.136067867279053,
      "learning_rate": 0.00019950000000000002,
      "loss": 0.9733,
      "mean_token_accuracy": 0.7564339616894722,
      "num_tokens": 589431.0,
      "step": 400
    },
    {
      "epoch": 0.6417112299465241,
      "grad_norm": 4.312187671661377,
      "learning_rate": 0.00019825331493065272,
      "loss": 1.0798,
      "step": 600
    },
    {
      "epoch": 0.6417112299465241,
      "eval_loss": 1.0830684900283813,
      "eval_mean_token_accuracy": 0.73986159324646,
      "eval_num_tokens": 883910.0,
      "eval_runtime": 3.5108,
      "eval_samples_per_second": 112.511,
      "eval_steps_per_second": 14.242,
      "step": 600
    },
    {
      "epoch": 0.8556149732620321,
      "grad_norm": 3.2650740146636963,
      "learning_rate": 0.00019303984150556786,
      "loss": 1.0648,
      "mean_token_accuracy": 0.7415659475326538,
      "num_tokens": 1177487.0,
      "step": 800
    },
    {
      "epoch": 0.9625668449197861,
      "eval_loss": 1.010650873184204,
      "eval_mean_token_accuracy": 0.7560834574699402,
      "eval_num_tokens": 1323462.0,
      "eval_runtime": 3.4996,
      "eval_samples_per_second": 112.87,
      "eval_steps_per_second": 14.287,
      "step": 900
    },
    {
      "epoch": 1.0695187165775402,
      "grad_norm": 3.137941360473633,
      "learning_rate": 0.00018454349100811027,
      "loss": 0.9486,
      "mean_token_accuracy": 0.7646993261575699,
      "num_tokens": 1470191.0,
      "step": 1000
    },
    {
      "epoch": 1.2834224598930482,
      "grad_norm": 2.7449452877044678,
      "learning_rate": 0.000173064054084906,
      "loss": 0.7726,
      "step": 1200
    },
    {
      "epoch": 1.2834224598930482,
      "eval_loss": 1.0005552768707275,
      "eval_mean_token_accuracy": 0.7632422924041748,
      "eval_num_tokens": 1761714.0,
      "eval_runtime": 3.8498,
      "eval_samples_per_second": 102.603,
      "eval_steps_per_second": 12.988,
      "step": 1200
    },
    {
      "epoch": 1.4973262032085561,
      "grad_norm": 2.728714942932129,
      "learning_rate": 0.00015900657850548172,
      "loss": 0.7477,
      "mean_token_accuracy": 0.7985022179782391,
      "num_tokens": 2057318.0,
      "step": 1400
    },
    {
      "epoch": 1.6042780748663101,
      "eval_loss": 0.9276342988014221,
      "eval_mean_token_accuracy": 0.7790945518016815,
      "eval_num_tokens": 2203814.0,
      "eval_runtime": 3.4558,
      "eval_samples_per_second": 114.3,
      "eval_steps_per_second": 14.468,
      "step": 1500
    },
    {
      "epoch": 1.7112299465240641,
      "grad_norm": 2.216341972351074,
      "learning_rate": 0.00014286707719858222,
      "loss": 0.7596,
      "mean_token_accuracy": 0.7985696256160736,
      "num_tokens": 2350582.0,
      "step": 1600
    },
    {
      "epoch": 1.9251336898395723,
      "grad_norm": 2.4397406578063965,
      "learning_rate": 0.00012521502661582825,
      "loss": 0.7178,
      "step": 1800
    },
    {
      "epoch": 1.9251336898395723,
      "eval_loss": 0.8735862970352173,
      "eval_mean_token_accuracy": 0.7864179539680481,
      "eval_num_tokens": 2645772.0,
      "eval_runtime": 3.5561,
      "eval_samples_per_second": 111.075,
      "eval_steps_per_second": 14.06,
      "step": 1800
    },
    {
      "epoch": 2.1390374331550803,
      "grad_norm": 2.1956679821014404,
      "learning_rate": 0.00010667327296160663,
      "loss": 0.5158,
      "mean_token_accuracy": 0.831416589319706,
      "num_tokens": 2939582.0,
      "step": 2000
    },
    {
      "epoch": 2.2459893048128343,
      "eval_loss": 0.9084789156913757,
      "eval_mean_token_accuracy": 0.7915868234634399,
      "eval_num_tokens": 3088145.0,
      "eval_runtime": 3.725,
      "eval_samples_per_second": 106.041,
      "eval_steps_per_second": 13.423,
      "step": 2100
    },
    {
      "epoch": 2.3529411764705883,
      "grad_norm": 2.2886502742767334,
      "learning_rate": 8.78960552905849e-05,
      "loss": 0.4192,
      "mean_token_accuracy": 0.8743109792470932,
      "num_tokens": 3234273.0,
      "step": 2200
    },
    {
      "epoch": 2.5668449197860963,
      "grad_norm": 2.467564105987549,
      "learning_rate": 6.954592091988702e-05,
      "loss": 0.389,
      "step": 2400
    },
    {
      "epoch": 2.5668449197860963,
      "eval_loss": 0.8907138705253601,
      "eval_mean_token_accuracy": 0.7962368977069855,
      "eval_num_tokens": 3530686.0,
      "eval_runtime": 3.8055,
      "eval_samples_per_second": 103.797,
      "eval_steps_per_second": 13.139,
      "step": 2400
    },
    {
      "epoch": 2.7807486631016043,
      "grad_norm": 1.9784467220306396,
      "learning_rate": 5.227034768724037e-05,
      "loss": 0.3688,
      "mean_token_accuracy": 0.8851943111419678,
      "num_tokens": 3821409.0,
      "step": 2600
    },
    {
      "epoch": 2.8877005347593583,
      "eval_loss": 0.8513710498809814,
      "eval_mean_token_accuracy": 0.8049397683143615,
      "eval_num_tokens": 3967119.0,
      "eval_runtime": 3.9058,
      "eval_samples_per_second": 101.132,
      "eval_steps_per_second": 12.802,
      "step": 2700
    },
    {
      "epoch": 2.9946524064171123,
      "grad_norm": 1.8156709671020508,
      "learning_rate": 3.667889793023945e-05,
      "loss": 0.3505,
      "mean_token_accuracy": 0.8934481579065323,
      "num_tokens": 4117627.0,
      "step": 2800
    },
    {
      "epoch": 3.2085561497326203,
      "grad_norm": 2.1861166954040527,
      "learning_rate": 2.3321710300283894e-05,
      "loss": 0.118,
      "step": 3000
    },
    {
      "epoch": 3.2085561497326203,
      "eval_loss": 1.0870622396469116,
      "eval_mean_token_accuracy": 0.802296108007431,
      "eval_num_tokens": 4410515.0,
      "eval_runtime": 3.5012,
      "eval_samples_per_second": 112.82,
      "eval_steps_per_second": 14.281,
      "step": 3000
    },
    {
      "epoch": 3.4224598930481283,
      "grad_norm": 1.6265203952789307,
      "learning_rate": 1.2670088319737627e-05,
      "loss": 0.1034,
      "mean_token_accuracy": 0.9655308005213737,
      "num_tokens": 4703490.0,
      "step": 3200
    },
    {
      "epoch": 3.5294117647058822,
      "eval_loss": 1.1269769668579102,
      "eval_mean_token_accuracy": 0.8054174113273621,
      "eval_num_tokens": 4851389.0,
      "eval_runtime": 3.755,
      "eval_samples_per_second": 105.193,
      "eval_steps_per_second": 13.316,
      "step": 3300
    },
    {
      "epoch": 3.6363636363636362,
      "grad_norm": 1.256324052810669,
      "learning_rate": 5.099870608031565e-06,
      "loss": 0.0959,
      "mean_token_accuracy": 0.969733267724514,
      "num_tokens": 5001670.0,
      "step": 3400
    }
  ],
  "logging_steps": 200,
  "max_steps": 3740,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 4,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 2.0281930094739456e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}