{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.997105643994211,
  "eval_steps": 1000,
  "global_step": 8284,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0004828585224529213,
      "grad_norm": 4.887877941131592,
      "learning_rate": 0.0,
      "loss": 1.2995,
      "step": 1
    },
    {
      "epoch": 0.48285852245292127,
      "grad_norm": 5.8714141845703125,
      "learning_rate": 9.7719651240778e-06,
      "loss": 1.1192,
      "step": 1000
    },
    {
      "epoch": 0.48285852245292127,
      "eval_cosine_accuracy": 0.9682406187057495,
      "eval_loss": 0.4464344084262848,
      "eval_runtime": 35.6815,
      "eval_samples_per_second": 266.497,
      "eval_steps_per_second": 1.065,
      "step": 1000
    },
    {
      "epoch": 0.9657170449058425,
      "grad_norm": 7.519778251647949,
      "learning_rate": 8.430583501006036e-06,
      "loss": 0.7419,
      "step": 2000
    },
    {
      "epoch": 0.9657170449058425,
      "eval_cosine_accuracy": 0.9675044417381287,
      "eval_loss": 0.4303629696369171,
      "eval_runtime": 34.0778,
      "eval_samples_per_second": 279.038,
      "eval_steps_per_second": 1.115,
      "step": 2000
    },
    {
      "epoch": 1.448142788229619,
      "grad_norm": 5.296412944793701,
      "learning_rate": 7.089201877934273e-06,
      "loss": 0.867,
      "step": 3000
    },
    {
      "epoch": 1.448142788229619,
      "eval_cosine_accuracy": 0.9691870808601379,
      "eval_loss": 0.44264352321624756,
      "eval_runtime": 34.494,
      "eval_samples_per_second": 275.671,
      "eval_steps_per_second": 1.102,
      "step": 3000
    },
    {
      "epoch": 1.930535455861071,
      "grad_norm": 5.242275714874268,
      "learning_rate": 5.7478202548625085e-06,
      "loss": 0.8798,
      "step": 4000
    },
    {
      "epoch": 1.930535455861071,
      "eval_cosine_accuracy": 0.9688715934753418,
      "eval_loss": 0.43794524669647217,
      "eval_runtime": 34.2759,
      "eval_samples_per_second": 277.425,
      "eval_steps_per_second": 1.109,
      "step": 4000
    },
    {
      "epoch": 2.412928123492523,
      "grad_norm": 5.030921459197998,
      "learning_rate": 4.409121395036888e-06,
      "loss": 0.8611,
      "step": 5000
    },
    {
      "epoch": 2.412928123492523,
      "eval_cosine_accuracy": 0.968766450881958,
      "eval_loss": 0.44171351194381714,
      "eval_runtime": 33.931,
      "eval_samples_per_second": 280.245,
      "eval_steps_per_second": 1.12,
      "step": 5000
    },
    {
      "epoch": 2.895320791123975,
      "grad_norm": 4.831454277038574,
      "learning_rate": 3.067739771965124e-06,
      "loss": 0.8396,
      "step": 6000
    },
    {
      "epoch": 2.895320791123975,
      "eval_cosine_accuracy": 0.9703438878059387,
      "eval_loss": 0.43668919801712036,
      "eval_runtime": 36.8551,
      "eval_samples_per_second": 258.01,
      "eval_steps_per_second": 1.031,
      "step": 6000
    },
    {
      "epoch": 3.377713458755427,
      "grad_norm": 5.794971466064453,
      "learning_rate": 1.7263581488933603e-06,
      "loss": 0.8236,
      "step": 7000
    },
    {
      "epoch": 3.377713458755427,
      "eval_cosine_accuracy": 0.9699232578277588,
      "eval_loss": 0.4372423589229584,
      "eval_runtime": 35.1919,
      "eval_samples_per_second": 270.204,
      "eval_steps_per_second": 1.08,
      "step": 7000
    },
    {
      "epoch": 3.8601061263868788,
      "grad_norm": 5.954846382141113,
      "learning_rate": 3.849765258215963e-07,
      "loss": 0.827,
      "step": 8000
    },
    {
      "epoch": 3.8601061263868788,
      "eval_cosine_accuracy": 0.9696077108383179,
      "eval_loss": 0.4386484920978546,
      "eval_runtime": 34.2104,
      "eval_samples_per_second": 277.956,
      "eval_steps_per_second": 1.111,
      "step": 8000
    }
  ],
  "logging_steps": 1000,
  "max_steps": 8284,
  "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
}