{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.9992356201031916,
  "eval_steps": 1000,
  "global_step": 15693,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00019116803670426305,
      "grad_norm": 4.81204080581665,
      "learning_rate": 0.0,
      "loss": 2.4739,
      "step": 1
    },
    {
      "epoch": 0.19116803670426305,
      "grad_norm": 3.8444881439208984,
      "learning_rate": 1.9089171974522293e-05,
      "loss": 1.904,
      "step": 1000
    },
    {
      "epoch": 0.19116803670426305,
      "eval_cosine_accuracy": 0.9566870927810669,
      "eval_loss": 1.0658318996429443,
      "eval_runtime": 24.2832,
      "eval_samples_per_second": 389.817,
      "eval_steps_per_second": 1.524,
      "step": 1000
    },
    {
      "epoch": 0.3823360734085261,
      "grad_norm": 3.6171822547912598,
      "learning_rate": 2.9088720526800256e-05,
      "loss": 1.4035,
      "step": 2000
    },
    {
      "epoch": 0.3823360734085261,
      "eval_cosine_accuracy": 0.9651383757591248,
      "eval_loss": 0.9598055481910706,
      "eval_runtime": 28.2382,
      "eval_samples_per_second": 335.22,
      "eval_steps_per_second": 1.31,
      "step": 2000
    },
    {
      "epoch": 0.5735041101127891,
      "grad_norm": 3.25337290763855,
      "learning_rate": 2.696452595057707e-05,
      "loss": 1.1812,
      "step": 3000
    },
    {
      "epoch": 0.5735041101127891,
      "eval_cosine_accuracy": 0.9709486365318298,
      "eval_loss": 0.9080163836479187,
      "eval_runtime": 23.2786,
      "eval_samples_per_second": 406.639,
      "eval_steps_per_second": 1.589,
      "step": 3000
    },
    {
      "epoch": 0.7646721468170522,
      "grad_norm": 3.5734429359436035,
      "learning_rate": 2.4840331374353893e-05,
      "loss": 1.1325,
      "step": 4000
    },
    {
      "epoch": 0.7646721468170522,
      "eval_cosine_accuracy": 0.9728502035140991,
      "eval_loss": 0.8635017275810242,
      "eval_runtime": 23.4355,
      "eval_samples_per_second": 403.918,
      "eval_steps_per_second": 1.579,
      "step": 4000
    },
    {
      "epoch": 0.9558401835213153,
      "grad_norm": 2.7804830074310303,
      "learning_rate": 2.2718260992706933e-05,
      "loss": 1.6713,
      "step": 5000
    },
    {
      "epoch": 0.9558401835213153,
      "eval_cosine_accuracy": 0.9725332856178284,
      "eval_loss": 0.8610901236534119,
      "eval_runtime": 23.4275,
      "eval_samples_per_second": 404.055,
      "eval_steps_per_second": 1.579,
      "step": 5000
    },
    {
      "epoch": 1.1469520351614753,
      "grad_norm": 3.1853649616241455,
      "learning_rate": 2.0596190611059974e-05,
      "loss": 1.5416,
      "step": 6000
    },
    {
      "epoch": 1.1469520351614753,
      "eval_cosine_accuracy": 0.9736953377723694,
      "eval_loss": 0.835780918598175,
      "eval_runtime": 23.4889,
      "eval_samples_per_second": 402.999,
      "eval_steps_per_second": 1.575,
      "step": 6000
    },
    {
      "epoch": 1.3380470093636538,
      "grad_norm": 3.137946367263794,
      "learning_rate": 1.847199603483679e-05,
      "loss": 1.1634,
      "step": 7000
    },
    {
      "epoch": 1.3380470093636538,
      "eval_cosine_accuracy": 0.9752799272537231,
      "eval_loss": 0.8124368190765381,
      "eval_runtime": 23.5481,
      "eval_samples_per_second": 401.986,
      "eval_steps_per_second": 1.571,
      "step": 7000
    },
    {
      "epoch": 1.5291419835658322,
      "grad_norm": 2.9267895221710205,
      "learning_rate": 1.634780145861361e-05,
      "loss": 1.1446,
      "step": 8000
    },
    {
      "epoch": 1.5291419835658322,
      "eval_cosine_accuracy": 0.9762307405471802,
      "eval_loss": 0.8198033571243286,
      "eval_runtime": 23.482,
      "eval_samples_per_second": 403.117,
      "eval_steps_per_second": 1.576,
      "step": 8000
    },
    {
      "epoch": 1.7202369577680106,
      "grad_norm": 2.9279608726501465,
      "learning_rate": 1.4223606882390427e-05,
      "loss": 1.1326,
      "step": 9000
    },
    {
      "epoch": 1.7202369577680106,
      "eval_cosine_accuracy": 0.9760194420814514,
      "eval_loss": 0.8059930801391602,
      "eval_runtime": 23.4446,
      "eval_samples_per_second": 403.761,
      "eval_steps_per_second": 1.578,
      "step": 9000
    },
    {
      "epoch": 1.9113319319701891,
      "grad_norm": 7.512048244476318,
      "learning_rate": 1.2103660695319691e-05,
      "loss": 1.1111,
      "step": 10000
    },
    {
      "epoch": 1.9113319319701891,
      "eval_cosine_accuracy": 0.9765476584434509,
      "eval_loss": 0.8110432028770447,
      "eval_runtime": 23.4361,
      "eval_samples_per_second": 403.906,
      "eval_steps_per_second": 1.579,
      "step": 10000
    },
    {
      "epoch": 2.1024269061723677,
      "grad_norm": 2.9181814193725586,
      "learning_rate": 9.979466119096509e-06,
      "loss": 1.062,
      "step": 11000
    },
    {
      "epoch": 2.1024269061723677,
      "eval_cosine_accuracy": 0.9778153300285339,
      "eval_loss": 0.8091529011726379,
      "eval_runtime": 23.4212,
      "eval_samples_per_second": 404.164,
      "eval_steps_per_second": 1.58,
      "step": 11000
    },
    {
      "epoch": 2.2935218803745463,
      "grad_norm": 4.17739725112915,
      "learning_rate": 7.855271542873328e-06,
      "loss": 1.0553,
      "step": 12000
    },
    {
      "epoch": 2.2935218803745463,
      "eval_cosine_accuracy": 0.9770758748054504,
      "eval_loss": 0.8072733879089355,
      "eval_runtime": 23.4866,
      "eval_samples_per_second": 403.038,
      "eval_steps_per_second": 1.575,
      "step": 12000
    },
    {
      "epoch": 2.484616854576725,
      "grad_norm": 3.0207948684692383,
      "learning_rate": 5.731076966650145e-06,
      "loss": 1.0477,
      "step": 13000
    },
    {
      "epoch": 2.484616854576725,
      "eval_cosine_accuracy": 0.9777097105979919,
      "eval_loss": 0.8018991351127625,
      "eval_runtime": 23.3704,
      "eval_samples_per_second": 405.042,
      "eval_steps_per_second": 1.583,
      "step": 13000
    },
    {
      "epoch": 2.675711828778903,
      "grad_norm": 4.817632675170898,
      "learning_rate": 3.6090065850031863e-06,
      "loss": 1.0424,
      "step": 14000
    },
    {
      "epoch": 2.675711828778903,
      "eval_cosine_accuracy": 0.9778153300285339,
      "eval_loss": 0.801213800907135,
      "eval_runtime": 23.5864,
      "eval_samples_per_second": 401.334,
      "eval_steps_per_second": 1.569,
      "step": 14000
    },
    {
      "epoch": 2.8668068029810816,
      "grad_norm": 2.7821929454803467,
      "learning_rate": 1.4869362033562276e-06,
      "loss": 1.0563,
      "step": 15000
    },
    {
      "epoch": 2.8668068029810816,
      "eval_cosine_accuracy": 0.9774984121322632,
      "eval_loss": 0.800908088684082,
      "eval_runtime": 23.551,
      "eval_samples_per_second": 401.936,
      "eval_steps_per_second": 1.571,
      "step": 15000
    }
  ],
  "logging_steps": 1000,
  "max_steps": 15693,
  "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
}