{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.997826086956522,
  "eval_steps": 1000,
  "global_step": 11032,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00036258158085569254,
      "grad_norm": 3.518895387649536,
      "learning_rate": 0.0,
      "loss": 1.2042,
      "step": 1
    },
    {
      "epoch": 0.36258158085569253,
      "grad_norm": 4.220677852630615,
      "learning_rate": 9.053013140009063e-06,
      "loss": 1.1885,
      "step": 1000
    },
    {
      "epoch": 0.36258158085569253,
      "eval_cosine_accuracy": 0.9711852073669434,
      "eval_loss": 0.3902958035469055,
      "eval_runtime": 34.6603,
      "eval_samples_per_second": 274.348,
      "eval_steps_per_second": 1.096,
      "step": 1000
    },
    {
      "epoch": 0.7251631617113851,
      "grad_norm": 4.6132588386535645,
      "learning_rate": 1.811508835523335e-05,
      "loss": 0.8207,
      "step": 2000
    },
    {
      "epoch": 0.7251631617113851,
      "eval_cosine_accuracy": 0.9699232578277588,
      "eval_loss": 0.37875595688819885,
      "eval_runtime": 33.7082,
      "eval_samples_per_second": 282.098,
      "eval_steps_per_second": 1.127,
      "step": 2000
    },
    {
      "epoch": 1.0876811594202898,
      "grad_norm": 4.405287742614746,
      "learning_rate": 1.8205099150141645e-05,
      "loss": 0.7257,
      "step": 3000
    },
    {
      "epoch": 1.0876811594202898,
      "eval_cosine_accuracy": 0.9716058373451233,
      "eval_loss": 0.3854234516620636,
      "eval_runtime": 34.2421,
      "eval_samples_per_second": 277.699,
      "eval_steps_per_second": 1.11,
      "step": 3000
    },
    {
      "epoch": 1.45,
      "grad_norm": 4.425130367279053,
      "learning_rate": 1.5941076487252125e-05,
      "loss": 0.9126,
      "step": 4000
    },
    {
      "epoch": 1.45,
      "eval_cosine_accuracy": 0.9708697199821472,
      "eval_loss": 0.3814031183719635,
      "eval_runtime": 33.9647,
      "eval_samples_per_second": 279.967,
      "eval_steps_per_second": 1.119,
      "step": 4000
    },
    {
      "epoch": 1.8123188405797102,
      "grad_norm": 4.010558128356934,
      "learning_rate": 1.3677053824362606e-05,
      "loss": 0.8911,
      "step": 5000
    },
    {
      "epoch": 1.8123188405797102,
      "eval_cosine_accuracy": 0.9710800051689148,
      "eval_loss": 0.38102903962135315,
      "eval_runtime": 34.858,
      "eval_samples_per_second": 272.792,
      "eval_steps_per_second": 1.09,
      "step": 5000
    },
    {
      "epoch": 2.17463768115942,
      "grad_norm": 3.913022041320801,
      "learning_rate": 1.1410764872521247e-05,
      "loss": 0.8678,
      "step": 6000
    },
    {
      "epoch": 2.17463768115942,
      "eval_cosine_accuracy": 0.9716058373451233,
      "eval_loss": 0.38304567337036133,
      "eval_runtime": 34.2523,
      "eval_samples_per_second": 277.616,
      "eval_steps_per_second": 1.109,
      "step": 6000
    },
    {
      "epoch": 2.5369565217391306,
      "grad_norm": 3.9786877632141113,
      "learning_rate": 9.144475920679887e-06,
      "loss": 0.8551,
      "step": 7000
    },
    {
      "epoch": 2.5369565217391306,
      "eval_cosine_accuracy": 0.9721316695213318,
      "eval_loss": 0.3812798261642456,
      "eval_runtime": 33.9373,
      "eval_samples_per_second": 280.193,
      "eval_steps_per_second": 1.12,
      "step": 7000
    },
    {
      "epoch": 2.8992753623188405,
      "grad_norm": 3.7564237117767334,
      "learning_rate": 6.878186968838527e-06,
      "loss": 0.8576,
      "step": 8000
    },
    {
      "epoch": 2.8992753623188405,
      "eval_cosine_accuracy": 0.9715006947517395,
      "eval_loss": 0.38419124484062195,
      "eval_runtime": 34.0546,
      "eval_samples_per_second": 279.228,
      "eval_steps_per_second": 1.116,
      "step": 8000
    },
    {
      "epoch": 3.261594202898551,
      "grad_norm": 3.8467016220092773,
      "learning_rate": 4.611898016997168e-06,
      "loss": 0.8325,
      "step": 9000
    },
    {
      "epoch": 3.261594202898551,
      "eval_cosine_accuracy": 0.9712903499603271,
      "eval_loss": 0.3847886919975281,
      "eval_runtime": 34.006,
      "eval_samples_per_second": 279.627,
      "eval_steps_per_second": 1.117,
      "step": 9000
    },
    {
      "epoch": 3.623913043478261,
      "grad_norm": 5.066013813018799,
      "learning_rate": 2.3456090651558077e-06,
      "loss": 0.8453,
      "step": 10000
    },
    {
      "epoch": 3.623913043478261,
      "eval_cosine_accuracy": 0.970974862575531,
      "eval_loss": 0.3841721713542938,
      "eval_runtime": 33.9641,
      "eval_samples_per_second": 279.972,
      "eval_steps_per_second": 1.119,
      "step": 10000
    },
    {
      "epoch": 3.986231884057971,
      "grad_norm": 4.202202796936035,
      "learning_rate": 8.158640226628895e-08,
      "loss": 0.817,
      "step": 11000
    },
    {
      "epoch": 3.986231884057971,
      "eval_cosine_accuracy": 0.9705542325973511,
      "eval_loss": 0.38476327061653137,
      "eval_runtime": 34.6061,
      "eval_samples_per_second": 274.778,
      "eval_steps_per_second": 1.098,
      "step": 11000
    }
  ],
  "logging_steps": 1000,
  "max_steps": 11032,
  "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": true
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 256,
  "trial_name": null,
  "trial_params": null
}