{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.9984492116826056,
  "eval_steps": 1000,
  "global_step": 15468,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0002585983966899405,
      "grad_norm": 6.478072643280029,
      "learning_rate": 0.0,
      "loss": 3.5134,
      "step": 1
    },
    {
      "epoch": 0.2585983966899405,
      "grad_norm": 3.769953966140747,
      "learning_rate": 1.9372979961215257e-05,
      "loss": 2.5294,
      "step": 1000
    },
    {
      "epoch": 0.2585983966899405,
      "eval_cosine_accuracy": 0.9476074576377869,
      "eval_loss": 1.1219995021820068,
      "eval_runtime": 26.9302,
      "eval_samples_per_second": 351.539,
      "eval_steps_per_second": 1.374,
      "step": 1000
    },
    {
      "epoch": 0.517196793379881,
      "grad_norm": 3.698908567428589,
      "learning_rate": 2.90259320451117e-05,
      "loss": 1.84,
      "step": 2000
    },
    {
      "epoch": 0.517196793379881,
      "eval_cosine_accuracy": 0.9596493244171143,
      "eval_loss": 1.035656452178955,
      "eval_runtime": 25.5286,
      "eval_samples_per_second": 370.839,
      "eval_steps_per_second": 1.449,
      "step": 2000
    },
    {
      "epoch": 0.7757951900698216,
      "grad_norm": 3.6514859199523926,
      "learning_rate": 2.6870914445801306e-05,
      "loss": 1.6007,
      "step": 3000
    },
    {
      "epoch": 0.7757951900698216,
      "eval_cosine_accuracy": 0.9655646085739136,
      "eval_loss": 0.969273030757904,
      "eval_runtime": 26.2028,
      "eval_samples_per_second": 361.297,
      "eval_steps_per_second": 1.412,
      "step": 3000
    },
    {
      "epoch": 1.0343758077022487,
      "grad_norm": 3.4506077766418457,
      "learning_rate": 2.4715896846490915e-05,
      "loss": 2.0429,
      "step": 4000
    },
    {
      "epoch": 1.0343758077022487,
      "eval_cosine_accuracy": 0.9675715565681458,
      "eval_loss": 0.9276238679885864,
      "eval_runtime": 25.874,
      "eval_samples_per_second": 365.889,
      "eval_steps_per_second": 1.43,
      "step": 4000
    },
    {
      "epoch": 1.292840527268028,
      "grad_norm": 3.4195048809051514,
      "learning_rate": 2.256303426477983e-05,
      "loss": 1.5438,
      "step": 5000
    },
    {
      "epoch": 1.292840527268028,
      "eval_cosine_accuracy": 0.9688391089439392,
      "eval_loss": 0.8985922336578369,
      "eval_runtime": 25.7504,
      "eval_samples_per_second": 367.644,
      "eval_steps_per_second": 1.437,
      "step": 5000
    },
    {
      "epoch": 1.5513052468338073,
      "grad_norm": 3.425976276397705,
      "learning_rate": 2.0408016665469435e-05,
      "loss": 1.5027,
      "step": 6000
    },
    {
      "epoch": 1.5513052468338073,
      "eval_cosine_accuracy": 0.9702123403549194,
      "eval_loss": 0.8979971408843994,
      "eval_runtime": 28.0719,
      "eval_samples_per_second": 337.241,
      "eval_steps_per_second": 1.318,
      "step": 6000
    },
    {
      "epoch": 1.8097699663995863,
      "grad_norm": 3.0993404388427734,
      "learning_rate": 1.825515408375835e-05,
      "loss": 1.4302,
      "step": 7000
    },
    {
      "epoch": 1.8097699663995863,
      "eval_cosine_accuracy": 0.9708461165428162,
      "eval_loss": 0.9006130695343018,
      "eval_runtime": 26.6511,
      "eval_samples_per_second": 355.22,
      "eval_steps_per_second": 1.388,
      "step": 7000
    },
    {
      "epoch": 2.0682346859653657,
      "grad_norm": 3.259557008743286,
      "learning_rate": 1.6100136484447955e-05,
      "loss": 1.4145,
      "step": 8000
    },
    {
      "epoch": 2.0682346859653657,
      "eval_cosine_accuracy": 0.9703179597854614,
      "eval_loss": 0.899039089679718,
      "eval_runtime": 26.4114,
      "eval_samples_per_second": 358.444,
      "eval_steps_per_second": 1.401,
      "step": 8000
    },
    {
      "epoch": 2.326699405531145,
      "grad_norm": 3.6806280612945557,
      "learning_rate": 1.3947273902736874e-05,
      "loss": 1.3572,
      "step": 9000
    },
    {
      "epoch": 2.326699405531145,
      "eval_cosine_accuracy": 0.9706348180770874,
      "eval_loss": 0.8929062485694885,
      "eval_runtime": 26.9498,
      "eval_samples_per_second": 351.283,
      "eval_steps_per_second": 1.373,
      "step": 9000
    },
    {
      "epoch": 2.585164125096924,
      "grad_norm": 3.2950358390808105,
      "learning_rate": 1.1792256303426479e-05,
      "loss": 1.3533,
      "step": 10000
    },
    {
      "epoch": 2.585164125096924,
      "eval_cosine_accuracy": 0.9734868407249451,
      "eval_loss": 0.8817673325538635,
      "eval_runtime": 26.1748,
      "eval_samples_per_second": 361.683,
      "eval_steps_per_second": 1.414,
      "step": 10000
    },
    {
      "epoch": 2.8436288446627036,
      "grad_norm": 3.5843701362609863,
      "learning_rate": 9.641548739314706e-06,
      "loss": 1.3183,
      "step": 11000
    },
    {
      "epoch": 2.8436288446627036,
      "eval_cosine_accuracy": 0.9726418256759644,
      "eval_loss": 0.8857349753379822,
      "eval_runtime": 27.1597,
      "eval_samples_per_second": 348.568,
      "eval_steps_per_second": 1.362,
      "step": 11000
    },
    {
      "epoch": 3.1020935642284826,
      "grad_norm": 3.425356149673462,
      "learning_rate": 7.48653114000431e-06,
      "loss": 1.3243,
      "step": 12000
    },
    {
      "epoch": 3.1020935642284826,
      "eval_cosine_accuracy": 0.9745431542396545,
      "eval_loss": 0.8805464506149292,
      "eval_runtime": 26.6713,
      "eval_samples_per_second": 354.95,
      "eval_steps_per_second": 1.387,
      "step": 12000
    },
    {
      "epoch": 3.360558283794262,
      "grad_norm": 3.2024552822113037,
      "learning_rate": 5.331513540693916e-06,
      "loss": 1.2964,
      "step": 13000
    },
    {
      "epoch": 3.360558283794262,
      "eval_cosine_accuracy": 0.9733812212944031,
      "eval_loss": 0.8850848078727722,
      "eval_runtime": 26.7557,
      "eval_samples_per_second": 353.831,
      "eval_steps_per_second": 1.383,
      "step": 13000
    },
    {
      "epoch": 3.6190230033600415,
      "grad_norm": 3.8241848945617676,
      "learning_rate": 3.1786509589828317e-06,
      "loss": 1.2724,
      "step": 14000
    },
    {
      "epoch": 3.6190230033600415,
      "eval_cosine_accuracy": 0.9738037586212158,
      "eval_loss": 0.8803183436393738,
      "eval_runtime": 26.7358,
      "eval_samples_per_second": 354.094,
      "eval_steps_per_second": 1.384,
      "step": 14000
    },
    {
      "epoch": 3.877487722925821,
      "grad_norm": 3.636179208755493,
      "learning_rate": 1.0236333596724372e-06,
      "loss": 1.2631,
      "step": 15000
    },
    {
      "epoch": 3.877487722925821,
      "eval_cosine_accuracy": 0.9725362062454224,
      "eval_loss": 0.8834013938903809,
      "eval_runtime": 27.3454,
      "eval_samples_per_second": 346.201,
      "eval_steps_per_second": 1.353,
      "step": 15000
    }
  ],
  "logging_steps": 1000,
  "max_steps": 15468,
  "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
}