{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.0,
  "eval_steps": 2000,
  "global_step": 22112,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 9.048136083966703e-05,
      "grad_norm": 6.883802890777588,
      "learning_rate": 0.0,
      "loss": 3.6871,
      "step": 1
    },
    {
      "epoch": 0.18096272167933405,
      "grad_norm": 4.156403064727783,
      "learning_rate": 5.426658220975478e-06,
      "loss": 3.6171,
      "step": 2000
    },
    {
      "epoch": 0.18096272167933405,
      "eval_cosine_accuracy": 0.9458659291267395,
      "eval_loss": 1.19174063205719,
      "eval_runtime": 24.1437,
      "eval_samples_per_second": 391.738,
      "eval_steps_per_second": 1.532,
      "step": 2000
    },
    {
      "epoch": 0.3619254433586681,
      "grad_norm": 4.531202793121338,
      "learning_rate": 1.085603112840467e-05,
      "loss": 2.9866,
      "step": 4000
    },
    {
      "epoch": 0.3619254433586681,
      "eval_cosine_accuracy": 0.9442799687385559,
      "eval_loss": 1.1735926866531372,
      "eval_runtime": 25.7204,
      "eval_samples_per_second": 367.723,
      "eval_steps_per_second": 1.439,
      "step": 4000
    },
    {
      "epoch": 0.5428881650380022,
      "grad_norm": 4.661669731140137,
      "learning_rate": 1.6282689349380148e-05,
      "loss": 2.6071,
      "step": 6000
    },
    {
      "epoch": 0.5428881650380022,
      "eval_cosine_accuracy": 0.9467117786407471,
      "eval_loss": 1.1516979932785034,
      "eval_runtime": 25.8461,
      "eval_samples_per_second": 365.935,
      "eval_steps_per_second": 1.432,
      "step": 6000
    },
    {
      "epoch": 0.7238508867173362,
      "grad_norm": 4.64531135559082,
      "learning_rate": 2.1709347570355623e-05,
      "loss": 2.3272,
      "step": 8000
    },
    {
      "epoch": 0.7238508867173362,
      "eval_cosine_accuracy": 0.9510467052459717,
      "eval_loss": 1.1301954984664917,
      "eval_runtime": 26.5489,
      "eval_samples_per_second": 356.248,
      "eval_steps_per_second": 1.394,
      "step": 8000
    },
    {
      "epoch": 0.9048136083966702,
      "grad_norm": 4.573121070861816,
      "learning_rate": 2.71360057913311e-05,
      "loss": 2.1236,
      "step": 10000
    },
    {
      "epoch": 0.9048136083966702,
      "eval_cosine_accuracy": 0.9563332796096802,
      "eval_loss": 1.0893216133117676,
      "eval_runtime": 25.6992,
      "eval_samples_per_second": 368.027,
      "eval_steps_per_second": 1.44,
      "step": 10000
    },
    {
      "epoch": 1.0857142857142856,
      "grad_norm": 4.530196666717529,
      "learning_rate": 2.9940039535626347e-05,
      "loss": 2.0434,
      "step": 12000
    },
    {
      "epoch": 1.0857142857142856,
      "eval_cosine_accuracy": 0.954641580581665,
      "eval_loss": 1.0764466524124146,
      "eval_runtime": 25.4391,
      "eval_samples_per_second": 371.79,
      "eval_steps_per_second": 1.454,
      "step": 12000
    },
    {
      "epoch": 1.2665461121157324,
      "grad_norm": 4.325132369995117,
      "learning_rate": 2.9420604401266522e-05,
      "loss": 1.8618,
      "step": 14000
    },
    {
      "epoch": 1.2665461121157324,
      "eval_cosine_accuracy": 0.9577077627182007,
      "eval_loss": 1.0421515703201294,
      "eval_runtime": 25.53,
      "eval_samples_per_second": 370.466,
      "eval_steps_per_second": 1.449,
      "step": 14000
    },
    {
      "epoch": 1.447377938517179,
      "grad_norm": 4.479296684265137,
      "learning_rate": 2.8385395034797537e-05,
      "loss": 1.7865,
      "step": 16000
    },
    {
      "epoch": 1.447377938517179,
      "eval_cosine_accuracy": 0.9604567289352417,
      "eval_loss": 1.029998779296875,
      "eval_runtime": 26.1797,
      "eval_samples_per_second": 361.273,
      "eval_steps_per_second": 1.413,
      "step": 16000
    },
    {
      "epoch": 1.6282097649186258,
      "grad_norm": 4.179690361022949,
      "learning_rate": 2.6872305921528123e-05,
      "loss": 1.7094,
      "step": 18000
    },
    {
      "epoch": 1.6282097649186258,
      "eval_cosine_accuracy": 0.9583421349525452,
      "eval_loss": 1.0228686332702637,
      "eval_runtime": 26.1715,
      "eval_samples_per_second": 361.385,
      "eval_steps_per_second": 1.414,
      "step": 18000
    },
    {
      "epoch": 1.8090415913200724,
      "grad_norm": 4.186921119689941,
      "learning_rate": 2.4932879923820447e-05,
      "loss": 1.6576,
      "step": 20000
    },
    {
      "epoch": 1.8090415913200724,
      "eval_cosine_accuracy": 0.9603510499000549,
      "eval_loss": 0.9981576800346375,
      "eval_runtime": 26.1781,
      "eval_samples_per_second": 361.294,
      "eval_steps_per_second": 1.413,
      "step": 20000
    },
    {
      "epoch": 1.989873417721519,
      "grad_norm": 4.027351379394531,
      "learning_rate": 2.2637835327220155e-05,
      "loss": 1.6159,
      "step": 22000
    },
    {
      "epoch": 1.989873417721519,
      "eval_cosine_accuracy": 0.9623599052429199,
      "eval_loss": 0.9824916124343872,
      "eval_runtime": 25.2661,
      "eval_samples_per_second": 374.336,
      "eval_steps_per_second": 1.464,
      "step": 22000
    }
  ],
  "logging_steps": 2000,
  "max_steps": 44208,
  "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": 0.0,
  "train_batch_size": 256,
  "trial_name": null,
  "trial_params": null
}