{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 834,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.17985611510791366,
      "grad_norm": 0.9889341592788696,
      "learning_rate": 0.0001995649347969019,
      "loss": 3.5084,
      "step": 50
    },
    {
      "epoch": 0.3597122302158273,
      "grad_norm": 1.3072373867034912,
      "learning_rate": 0.00019588933215113926,
      "loss": 2.8647,
      "step": 100
    },
    {
      "epoch": 0.539568345323741,
      "grad_norm": 1.416175127029419,
      "learning_rate": 0.00018860113594683148,
      "loss": 2.8011,
      "step": 150
    },
    {
      "epoch": 0.7194244604316546,
      "grad_norm": 1.3511329889297485,
      "learning_rate": 0.00017797492616144256,
      "loss": 2.796,
      "step": 200
    },
    {
      "epoch": 0.8992805755395683,
      "grad_norm": 1.0564734935760498,
      "learning_rate": 0.0001644110411450398,
      "loss": 2.6583,
      "step": 250
    },
    {
      "epoch": 1.0,
      "eval_loss": 2.709125280380249,
      "eval_runtime": 4.5805,
      "eval_samples_per_second": 66.15,
      "eval_steps_per_second": 8.296,
      "step": 278
    },
    {
      "epoch": 1.079136690647482,
      "grad_norm": 1.3963377475738525,
      "learning_rate": 0.0001484204950275565,
      "loss": 2.6333,
      "step": 300
    },
    {
      "epoch": 1.2589928057553956,
      "grad_norm": 0.9653306603431702,
      "learning_rate": 0.00013060572545878875,
      "loss": 2.5746,
      "step": 350
    },
    {
      "epoch": 1.4388489208633093,
      "grad_norm": 1.270931601524353,
      "learning_rate": 0.00011163789700258655,
      "loss": 2.6636,
      "step": 400
    },
    {
      "epoch": 1.6187050359712232,
      "grad_norm": 1.5205957889556885,
      "learning_rate": 9.223161527109937e-05,
      "loss": 2.6855,
      "step": 450
    },
    {
      "epoch": 1.7985611510791366,
      "grad_norm": 1.5966553688049316,
      "learning_rate": 7.311800443430251e-05,
      "loss": 2.4911,
      "step": 500
    },
    {
      "epoch": 1.9784172661870505,
      "grad_norm": 1.0975911617279053,
      "learning_rate": 5.501716239923642e-05,
      "loss": 2.5095,
      "step": 550
    },
    {
      "epoch": 2.0,
      "eval_loss": 2.6189730167388916,
      "eval_runtime": 4.5727,
      "eval_samples_per_second": 66.262,
      "eval_steps_per_second": 8.31,
      "step": 556
    },
    {
      "epoch": 2.158273381294964,
      "grad_norm": 1.6500194072723389,
      "learning_rate": 3.861103139944449e-05,
      "loss": 2.5351,
      "step": 600
    },
    {
      "epoch": 2.338129496402878,
      "grad_norm": 1.2627792358398438,
      "learning_rate": 2.451770608467432e-05,
      "loss": 2.4278,
      "step": 650
    },
    {
      "epoch": 2.5179856115107913,
      "grad_norm": 1.2723170518875122,
      "learning_rate": 1.326814704364262e-05,
      "loss": 2.5119,
      "step": 700
    },
    {
      "epoch": 2.697841726618705,
      "grad_norm": 1.461732268333435,
      "learning_rate": 5.286177068899989e-06,
      "loss": 2.4941,
      "step": 750
    },
    {
      "epoch": 2.8776978417266186,
      "grad_norm": 1.4850504398345947,
      "learning_rate": 8.725137967920738e-07,
      "loss": 2.6372,
      "step": 800
    },
    {
      "epoch": 3.0,
      "eval_loss": 2.6041791439056396,
      "eval_runtime": 4.5702,
      "eval_samples_per_second": 66.299,
      "eval_steps_per_second": 8.315,
      "step": 834
    }
  ],
  "logging_steps": 50,
  "max_steps": 834,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.060585900867584e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}