{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 1000,
  "global_step": 17853,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00016803898504453034,
      "grad_norm": 6.990426540374756,
      "learning_rate": 0.0,
      "loss": 2.5326,
      "step": 1
    },
    {
      "epoch": 0.16803898504453033,
      "grad_norm": 4.953593730926514,
      "learning_rate": 4.196597353497165e-06,
      "loss": 2.6779,
      "step": 1000
    },
    {
      "epoch": 0.16803898504453033,
      "eval_cosine_accuracy": 0.9457356333732605,
      "eval_loss": 1.3122717142105103,
      "eval_runtime": 21.2029,
      "eval_samples_per_second": 448.476,
      "eval_steps_per_second": 3.537,
      "step": 1000
    },
    {
      "epoch": 0.33607797008906065,
      "grad_norm": 5.172318458557129,
      "learning_rate": 8.397395505145979e-06,
      "loss": 2.1925,
      "step": 2000
    },
    {
      "epoch": 0.33607797008906065,
      "eval_cosine_accuracy": 0.951729953289032,
      "eval_loss": 1.2547080516815186,
      "eval_runtime": 21.0477,
      "eval_samples_per_second": 451.783,
      "eval_steps_per_second": 3.563,
      "step": 2000
    },
    {
      "epoch": 0.504116955133591,
      "grad_norm": 5.013792514801025,
      "learning_rate": 1.2598193656794792e-05,
      "loss": 1.9607,
      "step": 3000
    },
    {
      "epoch": 0.504116955133591,
      "eval_cosine_accuracy": 0.9543590545654297,
      "eval_loss": 1.2117713689804077,
      "eval_runtime": 21.1746,
      "eval_samples_per_second": 449.076,
      "eval_steps_per_second": 3.542,
      "step": 3000
    },
    {
      "epoch": 0.6721559401781213,
      "grad_norm": 5.219577312469482,
      "learning_rate": 1.6798991808443605e-05,
      "loss": 1.7977,
      "step": 4000
    },
    {
      "epoch": 0.6721559401781213,
      "eval_cosine_accuracy": 0.9588810801506042,
      "eval_loss": 1.1573827266693115,
      "eval_runtime": 20.9565,
      "eval_samples_per_second": 453.749,
      "eval_steps_per_second": 3.579,
      "step": 4000
    },
    {
      "epoch": 0.8401949252226516,
      "grad_norm": 5.110533714294434,
      "learning_rate": 1.9751089639237518e-05,
      "loss": 1.6686,
      "step": 5000
    },
    {
      "epoch": 0.8401949252226516,
      "eval_cosine_accuracy": 0.9588810801506042,
      "eval_loss": 1.1438082456588745,
      "eval_runtime": 20.8496,
      "eval_samples_per_second": 456.075,
      "eval_steps_per_second": 3.597,
      "step": 5000
    },
    {
      "epoch": 1.008233910267182,
      "grad_norm": 4.6514716148376465,
      "learning_rate": 1.870398571653626e-05,
      "loss": 1.5091,
      "step": 6000
    },
    {
      "epoch": 1.008233910267182,
      "eval_cosine_accuracy": 0.9621411561965942,
      "eval_loss": 1.1130626201629639,
      "eval_runtime": 22.292,
      "eval_samples_per_second": 426.566,
      "eval_steps_per_second": 3.364,
      "step": 6000
    },
    {
      "epoch": 1.1762728953117123,
      "grad_norm": 5.095766544342041,
      "learning_rate": 1.765373102977472e-05,
      "loss": 1.5015,
      "step": 7000
    },
    {
      "epoch": 1.1762728953117123,
      "eval_cosine_accuracy": 0.9629824161529541,
      "eval_loss": 1.0990869998931885,
      "eval_runtime": 20.8764,
      "eval_samples_per_second": 455.49,
      "eval_steps_per_second": 3.593,
      "step": 7000
    },
    {
      "epoch": 1.3443118803562426,
      "grad_norm": 4.731724262237549,
      "learning_rate": 1.6603476343013184e-05,
      "loss": 1.4365,
      "step": 8000
    },
    {
      "epoch": 1.3443118803562426,
      "eval_cosine_accuracy": 0.9641392230987549,
      "eval_loss": 1.0847463607788086,
      "eval_runtime": 20.8961,
      "eval_samples_per_second": 455.06,
      "eval_steps_per_second": 3.589,
      "step": 8000
    },
    {
      "epoch": 1.512350865400773,
      "grad_norm": 4.810811519622803,
      "learning_rate": 1.5553221656251644e-05,
      "loss": 1.4074,
      "step": 9000
    },
    {
      "epoch": 1.512350865400773,
      "eval_cosine_accuracy": 0.9646650552749634,
      "eval_loss": 1.072762131690979,
      "eval_runtime": 20.8161,
      "eval_samples_per_second": 456.809,
      "eval_steps_per_second": 3.603,
      "step": 9000
    },
    {
      "epoch": 1.6803898504453034,
      "grad_norm": 4.884791374206543,
      "learning_rate": 1.4502966969490102e-05,
      "loss": 1.3672,
      "step": 10000
    },
    {
      "epoch": 1.6803898504453034,
      "eval_cosine_accuracy": 0.9636133909225464,
      "eval_loss": 1.0661137104034424,
      "eval_runtime": 21.082,
      "eval_samples_per_second": 451.049,
      "eval_steps_per_second": 3.558,
      "step": 10000
    },
    {
      "epoch": 1.8484288354898335,
      "grad_norm": 5.121285915374756,
      "learning_rate": 1.3452712282728561e-05,
      "loss": 1.3327,
      "step": 11000
    },
    {
      "epoch": 1.8484288354898335,
      "eval_cosine_accuracy": 0.9652960300445557,
      "eval_loss": 1.0594764947891235,
      "eval_runtime": 21.5369,
      "eval_samples_per_second": 441.522,
      "eval_steps_per_second": 3.482,
      "step": 11000
    },
    {
      "epoch": 2.016467820534364,
      "grad_norm": 5.241132736206055,
      "learning_rate": 1.2403507850653784e-05,
      "loss": 1.2517,
      "step": 12000
    },
    {
      "epoch": 2.016467820534364,
      "eval_cosine_accuracy": 0.9668734669685364,
      "eval_loss": 1.0559396743774414,
      "eval_runtime": 21.0313,
      "eval_samples_per_second": 452.135,
      "eval_steps_per_second": 3.566,
      "step": 12000
    },
    {
      "epoch": 2.184506805578894,
      "grad_norm": 4.983489513397217,
      "learning_rate": 1.1353253163892244e-05,
      "loss": 1.267,
      "step": 13000
    },
    {
      "epoch": 2.184506805578894,
      "eval_cosine_accuracy": 0.965927004814148,
      "eval_loss": 1.0504363775253296,
      "eval_runtime": 21.0307,
      "eval_samples_per_second": 452.149,
      "eval_steps_per_second": 3.566,
      "step": 13000
    },
    {
      "epoch": 2.3525457906234246,
      "grad_norm": 5.162084102630615,
      "learning_rate": 1.0302998477130705e-05,
      "loss": 1.2527,
      "step": 14000
    },
    {
      "epoch": 2.3525457906234246,
      "eval_cosine_accuracy": 0.9664528369903564,
      "eval_loss": 1.0555484294891357,
      "eval_runtime": 21.0321,
      "eval_samples_per_second": 452.118,
      "eval_steps_per_second": 3.566,
      "step": 14000
    },
    {
      "epoch": 2.520584775667955,
      "grad_norm": 5.536242485046387,
      "learning_rate": 9.252743790369165e-06,
      "loss": 1.2288,
      "step": 15000
    },
    {
      "epoch": 2.520584775667955,
      "eval_cosine_accuracy": 0.9668734669685364,
      "eval_loss": 1.0455682277679443,
      "eval_runtime": 20.9211,
      "eval_samples_per_second": 454.517,
      "eval_steps_per_second": 3.585,
      "step": 15000
    },
    {
      "epoch": 2.688623760712485,
      "grad_norm": 4.500242233276367,
      "learning_rate": 8.202489103607626e-06,
      "loss": 1.2237,
      "step": 16000
    },
    {
      "epoch": 2.688623760712485,
      "eval_cosine_accuracy": 0.9684509634971619,
      "eval_loss": 1.0359514951705933,
      "eval_runtime": 21.2644,
      "eval_samples_per_second": 447.18,
      "eval_steps_per_second": 3.527,
      "step": 16000
    },
    {
      "epoch": 2.8566627457570157,
      "grad_norm": 5.441108703613281,
      "learning_rate": 7.153284671532848e-06,
      "loss": 1.2137,
      "step": 17000
    },
    {
      "epoch": 2.8566627457570157,
      "eval_cosine_accuracy": 0.9691870808601379,
      "eval_loss": 1.0284314155578613,
      "eval_runtime": 22.2121,
      "eval_samples_per_second": 428.099,
      "eval_steps_per_second": 3.377,
      "step": 17000
    }
  ],
  "logging_steps": 1000,
  "max_steps": 23804,
  "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": 128,
  "trial_name": null,
  "trial_params": null
}