{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.998221431747443,
  "eval_steps": 2000,
  "global_step": 26964,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00011125945705384957,
      "grad_norm": 6.3739542961120605,
      "learning_rate": 0.0,
      "loss": 2.9555,
      "step": 1
    },
    {
      "epoch": 0.22251891410769917,
      "grad_norm": 4.010181903839111,
      "learning_rate": 6.672972070768888e-06,
      "loss": 2.5193,
      "step": 2000
    },
    {
      "epoch": 0.22251891410769917,
      "eval_cosine_accuracy": 0.9481027126312256,
      "eval_loss": 1.1309852600097656,
      "eval_runtime": 23.41,
      "eval_samples_per_second": 404.144,
      "eval_steps_per_second": 1.581,
      "step": 2000
    },
    {
      "epoch": 0.44503782821539833,
      "grad_norm": 4.1902337074279785,
      "learning_rate": 1.3349282296650718e-05,
      "loss": 1.9678,
      "step": 4000
    },
    {
      "epoch": 0.44503782821539833,
      "eval_cosine_accuracy": 0.949688196182251,
      "eval_loss": 1.11049485206604,
      "eval_runtime": 24.2815,
      "eval_samples_per_second": 389.638,
      "eval_steps_per_second": 1.524,
      "step": 4000
    },
    {
      "epoch": 0.6675567423230975,
      "grad_norm": 4.1325764656066895,
      "learning_rate": 2.002559252253255e-05,
      "loss": 1.6578,
      "step": 6000
    },
    {
      "epoch": 0.6675567423230975,
      "eval_cosine_accuracy": 0.9524363279342651,
      "eval_loss": 1.1041761636734009,
      "eval_runtime": 23.4441,
      "eval_samples_per_second": 403.556,
      "eval_steps_per_second": 1.578,
      "step": 6000
    },
    {
      "epoch": 0.8900756564307967,
      "grad_norm": 3.8412773609161377,
      "learning_rate": 2.6691888283075553e-05,
      "loss": 1.4707,
      "step": 8000
    },
    {
      "epoch": 0.8900756564307967,
      "eval_cosine_accuracy": 0.9552901387214661,
      "eval_loss": 1.0991708040237427,
      "eval_runtime": 27.9755,
      "eval_samples_per_second": 338.189,
      "eval_steps_per_second": 1.323,
      "step": 8000
    },
    {
      "epoch": 1.1124944419742107,
      "grad_norm": 3.709298849105835,
      "learning_rate": 2.8316181787840018e-05,
      "loss": 1.412,
      "step": 10000
    },
    {
      "epoch": 1.1124944419742107,
      "eval_cosine_accuracy": 0.9583553671836853,
      "eval_loss": 1.084265947341919,
      "eval_runtime": 25.082,
      "eval_samples_per_second": 377.203,
      "eval_steps_per_second": 1.475,
      "step": 10000
    },
    {
      "epoch": 1.3348154735437974,
      "grad_norm": 3.361589193344116,
      "learning_rate": 2.4980252544918508e-05,
      "loss": 1.2352,
      "step": 12000
    },
    {
      "epoch": 1.3348154735437974,
      "eval_cosine_accuracy": 0.9597294330596924,
      "eval_loss": 1.0563029050827026,
      "eval_runtime": 23.5949,
      "eval_samples_per_second": 400.976,
      "eval_steps_per_second": 1.568,
      "step": 12000
    },
    {
      "epoch": 1.5571365051133839,
      "grad_norm": 3.4214119911193848,
      "learning_rate": 2.1644323301996995e-05,
      "loss": 1.1677,
      "step": 14000
    },
    {
      "epoch": 1.5571365051133839,
      "eval_cosine_accuracy": 0.9624775648117065,
      "eval_loss": 1.0454884767532349,
      "eval_runtime": 23.641,
      "eval_samples_per_second": 400.195,
      "eval_steps_per_second": 1.565,
      "step": 14000
    },
    {
      "epoch": 1.7794575366829704,
      "grad_norm": 3.2034318447113037,
      "learning_rate": 1.8306725260054514e-05,
      "loss": 1.1204,
      "step": 16000
    },
    {
      "epoch": 1.7794575366829704,
      "eval_cosine_accuracy": 0.9643800854682922,
      "eval_loss": 1.0180429220199585,
      "eval_runtime": 24.0838,
      "eval_samples_per_second": 392.836,
      "eval_steps_per_second": 1.536,
      "step": 16000
    },
    {
      "epoch": 2.001778568252557,
      "grad_norm": 3.1798715591430664,
      "learning_rate": 1.4970796017133005e-05,
      "loss": 1.0841,
      "step": 18000
    },
    {
      "epoch": 2.001778568252557,
      "eval_cosine_accuracy": 0.9650142788887024,
      "eval_loss": 1.0156272649765015,
      "eval_runtime": 23.4255,
      "eval_samples_per_second": 403.876,
      "eval_steps_per_second": 1.579,
      "step": 18000
    },
    {
      "epoch": 2.2240995998221433,
      "grad_norm": 3.4516661167144775,
      "learning_rate": 1.1634866774211493e-05,
      "loss": 1.0393,
      "step": 20000
    },
    {
      "epoch": 2.2240995998221433,
      "eval_cosine_accuracy": 0.9641686677932739,
      "eval_loss": 1.0275158882141113,
      "eval_runtime": 23.8864,
      "eval_samples_per_second": 396.083,
      "eval_steps_per_second": 1.549,
      "step": 20000
    },
    {
      "epoch": 2.44642063139173,
      "grad_norm": 3.4377176761627197,
      "learning_rate": 8.298937531289982e-06,
      "loss": 1.0227,
      "step": 22000
    },
    {
      "epoch": 2.44642063139173,
      "eval_cosine_accuracy": 0.9646971821784973,
      "eval_loss": 1.0210366249084473,
      "eval_runtime": 24.1039,
      "eval_samples_per_second": 392.51,
      "eval_steps_per_second": 1.535,
      "step": 22000
    },
    {
      "epoch": 2.668741662961316,
      "grad_norm": 3.480964183807373,
      "learning_rate": 4.9613394893475e-06,
      "loss": 1.0137,
      "step": 24000
    },
    {
      "epoch": 2.668741662961316,
      "eval_cosine_accuracy": 0.9654370546340942,
      "eval_loss": 1.0162781476974487,
      "eval_runtime": 23.9155,
      "eval_samples_per_second": 395.601,
      "eval_steps_per_second": 1.547,
      "step": 24000
    },
    {
      "epoch": 2.891062694530903,
      "grad_norm": 3.2681243419647217,
      "learning_rate": 1.6254102464259888e-06,
      "loss": 1.0091,
      "step": 26000
    },
    {
      "epoch": 2.891062694530903,
      "eval_cosine_accuracy": 0.9663883447647095,
      "eval_loss": 1.019175410270691,
      "eval_runtime": 23.5992,
      "eval_samples_per_second": 400.904,
      "eval_steps_per_second": 1.568,
      "step": 26000
    }
  ],
  "logging_steps": 2000,
  "max_steps": 26964,
  "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": 0.0,
  "train_batch_size": 256,
  "trial_name": null,
  "trial_params": null
}