{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.9977603583426653,
  "eval_steps": 2000,
  "global_step": 32112,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 9.342301943198804e-05,
      "grad_norm": 6.791104793548584,
      "learning_rate": 0.0,
      "loss": 2.1887,
      "step": 1
    },
    {
      "epoch": 0.18684603886397608,
      "grad_norm": 2.2691686153411865,
      "learning_rate": 1.1207250981125022e-05,
      "loss": 1.797,
      "step": 2000
    },
    {
      "epoch": 0.18684603886397608,
      "eval_cosine_accuracy": 0.8986994028091431,
      "eval_loss": 0.49471989274024963,
      "eval_runtime": 27.6086,
      "eval_samples_per_second": 342.538,
      "eval_steps_per_second": 1.34,
      "step": 2000
    },
    {
      "epoch": 0.37369207772795215,
      "grad_norm": 2.0151262283325195,
      "learning_rate": 2.242010839095496e-05,
      "loss": 1.1798,
      "step": 4000
    },
    {
      "epoch": 0.37369207772795215,
      "eval_cosine_accuracy": 0.9260864853858948,
      "eval_loss": 0.4586535394191742,
      "eval_runtime": 27.5216,
      "eval_samples_per_second": 343.62,
      "eval_steps_per_second": 1.344,
      "step": 4000
    },
    {
      "epoch": 0.5605381165919282,
      "grad_norm": 1.8293673992156982,
      "learning_rate": 2.9956752989128812e-05,
      "loss": 0.8731,
      "step": 6000
    },
    {
      "epoch": 0.5605381165919282,
      "eval_cosine_accuracy": 0.9477635622024536,
      "eval_loss": 0.41501784324645996,
      "eval_runtime": 27.2856,
      "eval_samples_per_second": 346.593,
      "eval_steps_per_second": 1.356,
      "step": 6000
    },
    {
      "epoch": 0.7473841554559043,
      "grad_norm": 1.68148934841156,
      "learning_rate": 2.928213765797858e-05,
      "loss": 0.7057,
      "step": 8000
    },
    {
      "epoch": 0.7473841554559043,
      "eval_cosine_accuracy": 0.9500898718833923,
      "eval_loss": 0.395673006772995,
      "eval_runtime": 27.3094,
      "eval_samples_per_second": 346.291,
      "eval_steps_per_second": 1.355,
      "step": 8000
    },
    {
      "epoch": 0.9342301943198804,
      "grad_norm": 1.6407605409622192,
      "learning_rate": 2.782368440395421e-05,
      "loss": 0.6096,
      "step": 10000
    },
    {
      "epoch": 0.9342301943198804,
      "eval_cosine_accuracy": 0.9434281587600708,
      "eval_loss": 0.37747085094451904,
      "eval_runtime": 27.4609,
      "eval_samples_per_second": 344.381,
      "eval_steps_per_second": 1.347,
      "step": 10000
    },
    {
      "epoch": 1.1209406494960805,
      "grad_norm": 1.4475024938583374,
      "learning_rate": 2.5662795115473793e-05,
      "loss": 0.5397,
      "step": 12000
    },
    {
      "epoch": 1.1209406494960805,
      "eval_cosine_accuracy": 0.958760678768158,
      "eval_loss": 0.36152419447898865,
      "eval_runtime": 27.3847,
      "eval_samples_per_second": 345.338,
      "eval_steps_per_second": 1.351,
      "step": 12000
    },
    {
      "epoch": 1.3075774542739829,
      "grad_norm": 1.596452236175537,
      "learning_rate": 2.2917388040385317e-05,
      "loss": 0.4996,
      "step": 14000
    },
    {
      "epoch": 1.3075774542739829,
      "eval_cosine_accuracy": 0.960769772529602,
      "eval_loss": 0.3521560728549957,
      "eval_runtime": 27.2239,
      "eval_samples_per_second": 347.379,
      "eval_steps_per_second": 1.359,
      "step": 14000
    },
    {
      "epoch": 1.494214259051885,
      "grad_norm": 1.4349610805511475,
      "learning_rate": 1.973629131160007e-05,
      "loss": 0.4737,
      "step": 16000
    },
    {
      "epoch": 1.494214259051885,
      "eval_cosine_accuracy": 0.9578090310096741,
      "eval_loss": 0.3508636951446533,
      "eval_runtime": 27.0893,
      "eval_samples_per_second": 349.105,
      "eval_steps_per_second": 1.366,
      "step": 16000
    },
    {
      "epoch": 1.6808510638297873,
      "grad_norm": 1.4522205591201782,
      "learning_rate": 1.6298808490859e-05,
      "loss": 0.4523,
      "step": 18000
    },
    {
      "epoch": 1.6808510638297873,
      "eval_cosine_accuracy": 0.9620386958122253,
      "eval_loss": 0.34350287914276123,
      "eval_runtime": 27.14,
      "eval_samples_per_second": 348.453,
      "eval_steps_per_second": 1.363,
      "step": 18000
    },
    {
      "epoch": 1.8674878686076894,
      "grad_norm": 2.92218017578125,
      "learning_rate": 1.2786714560275338e-05,
      "loss": 0.4411,
      "step": 20000
    },
    {
      "epoch": 1.8674878686076894,
      "eval_cosine_accuracy": 0.959395170211792,
      "eval_loss": 0.3393498957157135,
      "eval_runtime": 27.2509,
      "eval_samples_per_second": 347.034,
      "eval_steps_per_second": 1.358,
      "step": 20000
    },
    {
      "epoch": 2.0541246733855916,
      "grad_norm": 1.3841891288757324,
      "learning_rate": 9.397703795230084e-06,
      "loss": 0.4174,
      "step": 22000
    },
    {
      "epoch": 2.0541246733855916,
      "eval_cosine_accuracy": 0.9651052355766296,
      "eval_loss": 0.33905890583992004,
      "eval_runtime": 27.2071,
      "eval_samples_per_second": 347.593,
      "eval_steps_per_second": 1.36,
      "step": 22000
    },
    {
      "epoch": 2.2407614781634937,
      "grad_norm": 1.339237093925476,
      "learning_rate": 6.315802164522664e-06,
      "loss": 0.3957,
      "step": 24000
    },
    {
      "epoch": 2.2407614781634937,
      "eval_cosine_accuracy": 0.9655281901359558,
      "eval_loss": 0.3363184630870819,
      "eval_runtime": 27.2233,
      "eval_samples_per_second": 347.386,
      "eval_steps_per_second": 1.359,
      "step": 24000
    },
    {
      "epoch": 2.427398282941396,
      "grad_norm": 1.5053316354751587,
      "learning_rate": 3.7099547055673918e-06,
      "loss": 0.3955,
      "step": 26000
    },
    {
      "epoch": 2.427398282941396,
      "eval_cosine_accuracy": 0.9634133577346802,
      "eval_loss": 0.33477866649627686,
      "eval_runtime": 27.2189,
      "eval_samples_per_second": 347.442,
      "eval_steps_per_second": 1.359,
      "step": 26000
    },
    {
      "epoch": 2.6140350877192984,
      "grad_norm": 1.4208648204803467,
      "learning_rate": 1.7230099090697016e-06,
      "loss": 0.3923,
      "step": 28000
    },
    {
      "epoch": 2.6140350877192984,
      "eval_cosine_accuracy": 0.9642592668533325,
      "eval_loss": 0.33503878116607666,
      "eval_runtime": 27.6059,
      "eval_samples_per_second": 342.572,
      "eval_steps_per_second": 1.34,
      "step": 28000
    },
    {
      "epoch": 2.8006718924972005,
      "grad_norm": 1.3985047340393066,
      "learning_rate": 4.6388898771198173e-07,
      "loss": 0.3954,
      "step": 30000
    },
    {
      "epoch": 2.8006718924972005,
      "eval_cosine_accuracy": 0.9646822214126587,
      "eval_loss": 0.33804062008857727,
      "eval_runtime": 28.053,
      "eval_samples_per_second": 337.112,
      "eval_steps_per_second": 1.319,
      "step": 30000
    },
    {
      "epoch": 2.9873086972751026,
      "grad_norm": 1.5192722082138062,
      "learning_rate": 1.6149852696756284e-09,
      "loss": 0.3933,
      "step": 32000
    },
    {
      "epoch": 2.9873086972751026,
      "eval_cosine_accuracy": 0.9647879600524902,
      "eval_loss": 0.33818456530570984,
      "eval_runtime": 28.07,
      "eval_samples_per_second": 336.908,
      "eval_steps_per_second": 1.318,
      "step": 32000
    }
  ],
  "logging_steps": 2000,
  "max_steps": 32112,
  "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
}