{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 4.0,
  "eval_steps": 1000,
  "global_step": 23120,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00017301038062283736,
      "grad_norm": 7.641876220703125,
      "learning_rate": 0.0,
      "loss": 3.1444,
      "step": 1
    },
    {
      "epoch": 0.17301038062283736,
      "grad_norm": 5.932699203491211,
      "learning_rate": 3.4532871972318343e-06,
      "loss": 2.8902,
      "step": 1000
    },
    {
      "epoch": 0.17301038062283736,
      "eval_cosine_accuracy": 0.9446839690208435,
      "eval_loss": 0.5921438336372375,
      "eval_runtime": 34.6127,
      "eval_samples_per_second": 274.726,
      "eval_steps_per_second": 2.167,
      "step": 1000
    },
    {
      "epoch": 0.3460207612456747,
      "grad_norm": 5.819000244140625,
      "learning_rate": 6.913494809688582e-06,
      "loss": 2.3137,
      "step": 2000
    },
    {
      "epoch": 0.3460207612456747,
      "eval_cosine_accuracy": 0.9536228775978088,
      "eval_loss": 0.5212574601173401,
      "eval_runtime": 34.0023,
      "eval_samples_per_second": 279.658,
      "eval_steps_per_second": 2.206,
      "step": 2000
    },
    {
      "epoch": 0.5190311418685121,
      "grad_norm": 6.0497145652771,
      "learning_rate": 1.0373702422145328e-05,
      "loss": 2.0646,
      "step": 3000
    },
    {
      "epoch": 0.5190311418685121,
      "eval_cosine_accuracy": 0.9586707353591919,
      "eval_loss": 0.48181453347206116,
      "eval_runtime": 34.1002,
      "eval_samples_per_second": 278.855,
      "eval_steps_per_second": 2.199,
      "step": 3000
    },
    {
      "epoch": 0.6920415224913494,
      "grad_norm": 8.783215522766113,
      "learning_rate": 1.3833910034602079e-05,
      "loss": 1.7373,
      "step": 4000
    },
    {
      "epoch": 0.6920415224913494,
      "eval_cosine_accuracy": 0.9621411561965942,
      "eval_loss": 0.45189711451530457,
      "eval_runtime": 34.4171,
      "eval_samples_per_second": 276.287,
      "eval_steps_per_second": 2.179,
      "step": 4000
    },
    {
      "epoch": 0.8650519031141869,
      "grad_norm": 9.1194486618042,
      "learning_rate": 1.7290657439446367e-05,
      "loss": 1.2623,
      "step": 5000
    },
    {
      "epoch": 0.8650519031141869,
      "eval_cosine_accuracy": 0.9628772735595703,
      "eval_loss": 0.43156424164772034,
      "eval_runtime": 34.3791,
      "eval_samples_per_second": 276.592,
      "eval_steps_per_second": 2.182,
      "step": 5000
    },
    {
      "epoch": 1.0380622837370241,
      "grad_norm": 5.145761013031006,
      "learning_rate": 1.981228373702422e-05,
      "loss": 1.1672,
      "step": 6000
    },
    {
      "epoch": 1.0380622837370241,
      "eval_cosine_accuracy": 0.9668734669685364,
      "eval_loss": 0.40927383303642273,
      "eval_runtime": 34.7652,
      "eval_samples_per_second": 273.52,
      "eval_steps_per_second": 2.157,
      "step": 6000
    },
    {
      "epoch": 1.2110726643598615,
      "grad_norm": 5.79741907119751,
      "learning_rate": 1.8947231833910037e-05,
      "loss": 1.5637,
      "step": 7000
    },
    {
      "epoch": 1.2110726643598615,
      "eval_cosine_accuracy": 0.9705542325973511,
      "eval_loss": 0.3972666263580322,
      "eval_runtime": 34.1048,
      "eval_samples_per_second": 278.817,
      "eval_steps_per_second": 2.199,
      "step": 7000
    },
    {
      "epoch": 1.3840830449826989,
      "grad_norm": 5.9991455078125,
      "learning_rate": 1.808304498269896e-05,
      "loss": 1.4262,
      "step": 8000
    },
    {
      "epoch": 1.3840830449826989,
      "eval_cosine_accuracy": 0.9701335430145264,
      "eval_loss": 0.38196370005607605,
      "eval_runtime": 34.3663,
      "eval_samples_per_second": 276.695,
      "eval_steps_per_second": 2.182,
      "step": 8000
    },
    {
      "epoch": 1.5570934256055362,
      "grad_norm": 5.515242576599121,
      "learning_rate": 1.7217993079584777e-05,
      "loss": 1.3676,
      "step": 9000
    },
    {
      "epoch": 1.5570934256055362,
      "eval_cosine_accuracy": 0.9711852073669434,
      "eval_loss": 0.37576621770858765,
      "eval_runtime": 34.1496,
      "eval_samples_per_second": 278.451,
      "eval_steps_per_second": 2.196,
      "step": 9000
    },
    {
      "epoch": 1.7301038062283736,
      "grad_norm": 7.644779682159424,
      "learning_rate": 1.6353806228373702e-05,
      "loss": 1.1239,
      "step": 10000
    },
    {
      "epoch": 1.7301038062283736,
      "eval_cosine_accuracy": 0.9715006947517395,
      "eval_loss": 0.3619973063468933,
      "eval_runtime": 34.2124,
      "eval_samples_per_second": 277.941,
      "eval_steps_per_second": 2.192,
      "step": 10000
    },
    {
      "epoch": 1.903114186851211,
      "grad_norm": 5.467769622802734,
      "learning_rate": 1.5488754325259517e-05,
      "loss": 0.8423,
      "step": 11000
    },
    {
      "epoch": 1.903114186851211,
      "eval_cosine_accuracy": 0.9704490303993225,
      "eval_loss": 0.359011709690094,
      "eval_runtime": 34.1623,
      "eval_samples_per_second": 278.348,
      "eval_steps_per_second": 2.195,
      "step": 11000
    },
    {
      "epoch": 2.0761245674740483,
      "grad_norm": 6.093100070953369,
      "learning_rate": 1.462370242214533e-05,
      "loss": 0.9595,
      "step": 12000
    },
    {
      "epoch": 2.0761245674740483,
      "eval_cosine_accuracy": 0.9734987616539001,
      "eval_loss": 0.35547682642936707,
      "eval_runtime": 34.4247,
      "eval_samples_per_second": 276.226,
      "eval_steps_per_second": 2.179,
      "step": 12000
    },
    {
      "epoch": 2.2491349480968856,
      "grad_norm": 5.501762390136719,
      "learning_rate": 1.3758650519031142e-05,
      "loss": 1.1842,
      "step": 13000
    },
    {
      "epoch": 2.2491349480968856,
      "eval_cosine_accuracy": 0.9738143086433411,
      "eval_loss": 0.3599579632282257,
      "eval_runtime": 34.3646,
      "eval_samples_per_second": 276.709,
      "eval_steps_per_second": 2.182,
      "step": 13000
    },
    {
      "epoch": 2.422145328719723,
      "grad_norm": 5.384144306182861,
      "learning_rate": 1.289446366782007e-05,
      "loss": 1.1436,
      "step": 14000
    },
    {
      "epoch": 2.422145328719723,
      "eval_cosine_accuracy": 0.9745504260063171,
      "eval_loss": 0.356321781873703,
      "eval_runtime": 34.079,
      "eval_samples_per_second": 279.028,
      "eval_steps_per_second": 2.201,
      "step": 14000
    },
    {
      "epoch": 2.595155709342561,
      "grad_norm": 5.746756076812744,
      "learning_rate": 1.2029411764705882e-05,
      "loss": 1.147,
      "step": 15000
    },
    {
      "epoch": 2.595155709342561,
      "eval_cosine_accuracy": 0.9736039638519287,
      "eval_loss": 0.34958595037460327,
      "eval_runtime": 34.8051,
      "eval_samples_per_second": 273.207,
      "eval_steps_per_second": 2.155,
      "step": 15000
    },
    {
      "epoch": 2.7681660899653977,
      "grad_norm": 5.618105411529541,
      "learning_rate": 1.116522491349481e-05,
      "loss": 0.8589,
      "step": 16000
    },
    {
      "epoch": 2.7681660899653977,
      "eval_cosine_accuracy": 0.9732884764671326,
      "eval_loss": 0.34324178099632263,
      "eval_runtime": 34.2693,
      "eval_samples_per_second": 277.478,
      "eval_steps_per_second": 2.189,
      "step": 16000
    },
    {
      "epoch": 2.9411764705882355,
      "grad_norm": 8.138910293579102,
      "learning_rate": 1.0300173010380624e-05,
      "loss": 0.6846,
      "step": 17000
    },
    {
      "epoch": 2.9411764705882355,
      "eval_cosine_accuracy": 0.9734987616539001,
      "eval_loss": 0.3368176221847534,
      "eval_runtime": 34.4801,
      "eval_samples_per_second": 275.782,
      "eval_steps_per_second": 2.175,
      "step": 17000
    },
    {
      "epoch": 3.114186851211073,
      "grad_norm": 6.119685649871826,
      "learning_rate": 9.435121107266438e-06,
      "loss": 0.9107,
      "step": 18000
    },
    {
      "epoch": 3.114186851211073,
      "eval_cosine_accuracy": 0.9750762581825256,
      "eval_loss": 0.34765371680259705,
      "eval_runtime": 34.2992,
      "eval_samples_per_second": 277.237,
      "eval_steps_per_second": 2.187,
      "step": 18000
    },
    {
      "epoch": 3.28719723183391,
      "grad_norm": 5.255776405334473,
      "learning_rate": 8.57006920415225e-06,
      "loss": 1.0392,
      "step": 19000
    },
    {
      "epoch": 3.28719723183391,
      "eval_cosine_accuracy": 0.9748659133911133,
      "eval_loss": 0.3496827185153961,
      "eval_runtime": 34.5553,
      "eval_samples_per_second": 275.182,
      "eval_steps_per_second": 2.17,
      "step": 19000
    },
    {
      "epoch": 3.4602076124567476,
      "grad_norm": 5.913624286651611,
      "learning_rate": 7.705882352941178e-06,
      "loss": 1.0226,
      "step": 20000
    },
    {
      "epoch": 3.4602076124567476,
      "eval_cosine_accuracy": 0.9758123755455017,
      "eval_loss": 0.34967276453971863,
      "eval_runtime": 33.8444,
      "eval_samples_per_second": 280.962,
      "eval_steps_per_second": 2.216,
      "step": 20000
    },
    {
      "epoch": 3.633217993079585,
      "grad_norm": 6.428199291229248,
      "learning_rate": 6.840830449826991e-06,
      "loss": 1.0175,
      "step": 21000
    },
    {
      "epoch": 3.633217993079585,
      "eval_cosine_accuracy": 0.9747607707977295,
      "eval_loss": 0.33775225281715393,
      "eval_runtime": 33.9663,
      "eval_samples_per_second": 279.954,
      "eval_steps_per_second": 2.208,
      "step": 21000
    },
    {
      "epoch": 3.8062283737024223,
      "grad_norm": 8.56866455078125,
      "learning_rate": 5.977508650519032e-06,
      "loss": 0.7316,
      "step": 22000
    },
    {
      "epoch": 3.8062283737024223,
      "eval_cosine_accuracy": 0.9751814007759094,
      "eval_loss": 0.33261120319366455,
      "eval_runtime": 33.9328,
      "eval_samples_per_second": 280.231,
      "eval_steps_per_second": 2.21,
      "step": 22000
    },
    {
      "epoch": 3.9792387543252596,
      "grad_norm": 9.559788703918457,
      "learning_rate": 5.112456747404845e-06,
      "loss": 0.6017,
      "step": 23000
    },
    {
      "epoch": 3.9792387543252596,
      "eval_cosine_accuracy": 0.9738143086433411,
      "eval_loss": 0.3310614228248596,
      "eval_runtime": 34.9039,
      "eval_samples_per_second": 272.434,
      "eval_steps_per_second": 2.149,
      "step": 23000
    }
  ],
  "logging_steps": 1000,
  "max_steps": 28900,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 128,
  "trial_name": null,
  "trial_params": null
}