{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.998699609882965,
  "eval_steps": 1000,
  "global_step": 18444,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00016265452179570593,
      "grad_norm": 6.144397258758545,
      "learning_rate": 0.0,
      "loss": 3.8954,
      "step": 1
    },
    {
      "epoch": 0.16265452179570591,
      "grad_norm": 4.344635009765625,
      "learning_rate": 1.624390243902439e-05,
      "loss": 3.0884,
      "step": 1000
    },
    {
      "epoch": 0.16265452179570591,
      "eval_cosine_accuracy": 0.9446439743041992,
      "eval_loss": 1.014482855796814,
      "eval_runtime": 39.0277,
      "eval_samples_per_second": 242.545,
      "eval_steps_per_second": 0.948,
      "step": 1000
    },
    {
      "epoch": 0.32530904359141183,
      "grad_norm": 4.419331073760986,
      "learning_rate": 2.9721669980119285e-05,
      "loss": 2.3127,
      "step": 2000
    },
    {
      "epoch": 0.32530904359141183,
      "eval_cosine_accuracy": 0.9529896378517151,
      "eval_loss": 0.9663239121437073,
      "eval_runtime": 32.6394,
      "eval_samples_per_second": 290.018,
      "eval_steps_per_second": 1.134,
      "step": 2000
    },
    {
      "epoch": 0.48796356538711777,
      "grad_norm": 4.359193325042725,
      "learning_rate": 2.7914332188686065e-05,
      "loss": 2.192,
      "step": 3000
    },
    {
      "epoch": 0.48796356538711777,
      "eval_cosine_accuracy": 0.9564757943153381,
      "eval_loss": 0.921445369720459,
      "eval_runtime": 31.8747,
      "eval_samples_per_second": 296.975,
      "eval_steps_per_second": 1.161,
      "step": 3000
    },
    {
      "epoch": 0.6506180871828237,
      "grad_norm": 4.2792558670043945,
      "learning_rate": 2.6106994397252848e-05,
      "loss": 2.466,
      "step": 4000
    },
    {
      "epoch": 0.6506180871828237,
      "eval_cosine_accuracy": 0.9553137421607971,
      "eval_loss": 0.8910627365112305,
      "eval_runtime": 31.5443,
      "eval_samples_per_second": 300.086,
      "eval_steps_per_second": 1.173,
      "step": 4000
    },
    {
      "epoch": 0.8132726089785296,
      "grad_norm": 3.616072177886963,
      "learning_rate": 2.430146394361106e-05,
      "loss": 3.0285,
      "step": 5000
    },
    {
      "epoch": 0.8132726089785296,
      "eval_cosine_accuracy": 0.947284996509552,
      "eval_loss": 0.8859251141548157,
      "eval_runtime": 31.4273,
      "eval_samples_per_second": 301.203,
      "eval_steps_per_second": 1.177,
      "step": 5000
    },
    {
      "epoch": 0.9759271307742355,
      "grad_norm": 3.35726261138916,
      "learning_rate": 2.249412615217784e-05,
      "loss": 3.3706,
      "step": 6000
    },
    {
      "epoch": 0.9759271307742355,
      "eval_cosine_accuracy": 0.9297485947608948,
      "eval_loss": 0.9370076060295105,
      "eval_runtime": 31.4172,
      "eval_samples_per_second": 301.3,
      "eval_steps_per_second": 1.178,
      "step": 6000
    },
    {
      "epoch": 1.1384915474642392,
      "grad_norm": 4.022605895996094,
      "learning_rate": 2.0690403036327487e-05,
      "loss": 2.4494,
      "step": 7000
    },
    {
      "epoch": 1.1384915474642392,
      "eval_cosine_accuracy": 0.9583773612976074,
      "eval_loss": 0.8522692918777466,
      "eval_runtime": 31.3226,
      "eval_samples_per_second": 302.21,
      "eval_steps_per_second": 1.181,
      "step": 7000
    },
    {
      "epoch": 1.301040312093628,
      "grad_norm": 3.716705799102783,
      "learning_rate": 1.888306524489427e-05,
      "loss": 2.3159,
      "step": 8000
    },
    {
      "epoch": 1.301040312093628,
      "eval_cosine_accuracy": 0.9596450328826904,
      "eval_loss": 0.8640613555908203,
      "eval_runtime": 31.1134,
      "eval_samples_per_second": 304.242,
      "eval_steps_per_second": 1.189,
      "step": 8000
    },
    {
      "epoch": 1.4635890767230169,
      "grad_norm": 3.850571632385254,
      "learning_rate": 1.7077534791252487e-05,
      "loss": 2.2839,
      "step": 9000
    },
    {
      "epoch": 1.4635890767230169,
      "eval_cosine_accuracy": 0.961229681968689,
      "eval_loss": 0.8384360074996948,
      "eval_runtime": 31.1908,
      "eval_samples_per_second": 303.487,
      "eval_steps_per_second": 1.186,
      "step": 9000
    },
    {
      "epoch": 1.6261378413524057,
      "grad_norm": 3.9032530784606934,
      "learning_rate": 1.5270196999819267e-05,
      "loss": 2.2567,
      "step": 10000
    },
    {
      "epoch": 1.6261378413524057,
      "eval_cosine_accuracy": 0.9607014656066895,
      "eval_loss": 0.8450990319252014,
      "eval_runtime": 31.4124,
      "eval_samples_per_second": 301.346,
      "eval_steps_per_second": 1.178,
      "step": 10000
    },
    {
      "epoch": 1.7886866059817945,
      "grad_norm": 3.5230298042297363,
      "learning_rate": 1.3462859208386047e-05,
      "loss": 2.2833,
      "step": 11000
    },
    {
      "epoch": 1.7886866059817945,
      "eval_cosine_accuracy": 0.9608070850372314,
      "eval_loss": 0.8417729139328003,
      "eval_runtime": 31.1564,
      "eval_samples_per_second": 303.822,
      "eval_steps_per_second": 1.188,
      "step": 11000
    },
    {
      "epoch": 1.9512353706111834,
      "grad_norm": 3.5231220722198486,
      "learning_rate": 1.1657328754744262e-05,
      "loss": 2.2157,
      "step": 12000
    },
    {
      "epoch": 1.9512353706111834,
      "eval_cosine_accuracy": 0.9621804356575012,
      "eval_loss": 0.8389899134635925,
      "eval_runtime": 31.256,
      "eval_samples_per_second": 302.853,
      "eval_steps_per_second": 1.184,
      "step": 12000
    },
    {
      "epoch": 2.113784135240572,
      "grad_norm": 3.622410297393799,
      "learning_rate": 9.849990963311043e-06,
      "loss": 2.1555,
      "step": 13000
    },
    {
      "epoch": 2.113784135240572,
      "eval_cosine_accuracy": 0.962602972984314,
      "eval_loss": 0.8353741765022278,
      "eval_runtime": 31.4278,
      "eval_samples_per_second": 301.198,
      "eval_steps_per_second": 1.177,
      "step": 13000
    },
    {
      "epoch": 2.276332899869961,
      "grad_norm": 3.7428393363952637,
      "learning_rate": 8.044460509669258e-06,
      "loss": 2.1388,
      "step": 14000
    },
    {
      "epoch": 2.276332899869961,
      "eval_cosine_accuracy": 0.9630255699157715,
      "eval_loss": 0.838468074798584,
      "eval_runtime": 42.9003,
      "eval_samples_per_second": 220.651,
      "eval_steps_per_second": 0.862,
      "step": 14000
    },
    {
      "epoch": 2.4388816644993496,
      "grad_norm": 6.022229194641113,
      "learning_rate": 6.237122718236038e-06,
      "loss": 2.1736,
      "step": 15000
    },
    {
      "epoch": 2.4388816644993496,
      "eval_cosine_accuracy": 0.9629199504852295,
      "eval_loss": 0.8393190503120422,
      "eval_runtime": 31.3182,
      "eval_samples_per_second": 302.253,
      "eval_steps_per_second": 1.181,
      "step": 15000
    },
    {
      "epoch": 2.6014304291287385,
      "grad_norm": 3.2880027294158936,
      "learning_rate": 4.431592264594253e-06,
      "loss": 2.1658,
      "step": 16000
    },
    {
      "epoch": 2.6014304291287385,
      "eval_cosine_accuracy": 0.9617578983306885,
      "eval_loss": 0.8351490497589111,
      "eval_runtime": 31.3949,
      "eval_samples_per_second": 301.514,
      "eval_steps_per_second": 1.179,
      "step": 16000
    },
    {
      "epoch": 2.7639791937581273,
      "grad_norm": 3.7715585231781006,
      "learning_rate": 2.624254473161034e-06,
      "loss": 2.1363,
      "step": 17000
    },
    {
      "epoch": 2.7639791937581273,
      "eval_cosine_accuracy": 0.9630255699157715,
      "eval_loss": 0.8291847705841064,
      "eval_runtime": 31.3887,
      "eval_samples_per_second": 301.573,
      "eval_steps_per_second": 1.179,
      "step": 17000
    },
    {
      "epoch": 2.926527958387516,
      "grad_norm": 3.582385778427124,
      "learning_rate": 8.187240195192481e-07,
      "loss": 2.1611,
      "step": 18000
    },
    {
      "epoch": 2.926527958387516,
      "eval_cosine_accuracy": 0.9633424878120422,
      "eval_loss": 0.8309187889099121,
      "eval_runtime": 31.3265,
      "eval_samples_per_second": 302.172,
      "eval_steps_per_second": 1.181,
      "step": 18000
    }
  ],
  "logging_steps": 1000,
  "max_steps": 18444,
  "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
}