{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.0,
  "eval_steps": 1000,
  "global_step": 13315,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0003755163349605708,
      "grad_norm": 6.263205528259277,
      "learning_rate": 0.0,
      "loss": 3.7128,
      "step": 1
    },
    {
      "epoch": 0.3755163349605708,
      "grad_norm": 3.8261115550994873,
      "learning_rate": 7.502816372512205e-06,
      "loss": 2.8207,
      "step": 1000
    },
    {
      "epoch": 0.3755163349605708,
      "eval_cosine_accuracy": 0.9523609280586243,
      "eval_loss": 1.359626054763794,
      "eval_runtime": 22.5746,
      "eval_samples_per_second": 421.226,
      "eval_steps_per_second": 1.683,
      "step": 1000
    },
    {
      "epoch": 0.7510326699211416,
      "grad_norm": 3.8652374744415283,
      "learning_rate": 1.5013143071723622e-05,
      "loss": 2.2515,
      "step": 2000
    },
    {
      "epoch": 0.7510326699211416,
      "eval_cosine_accuracy": 0.9578294157981873,
      "eval_loss": 1.3198553323745728,
      "eval_runtime": 22.5254,
      "eval_samples_per_second": 422.145,
      "eval_steps_per_second": 1.687,
      "step": 2000
    },
    {
      "epoch": 1.1265490048817124,
      "grad_norm": 3.8326590061187744,
      "learning_rate": 1.9372887720615848e-05,
      "loss": 1.8906,
      "step": 3000
    },
    {
      "epoch": 1.1265490048817124,
      "eval_cosine_accuracy": 0.9598275423049927,
      "eval_loss": 1.3083600997924805,
      "eval_runtime": 22.9203,
      "eval_samples_per_second": 414.873,
      "eval_steps_per_second": 1.658,
      "step": 3000
    },
    {
      "epoch": 1.5020653398422832,
      "grad_norm": 3.9242522716522217,
      "learning_rate": 1.7495306045812995e-05,
      "loss": 1.7993,
      "step": 4000
    },
    {
      "epoch": 1.5020653398422832,
      "eval_cosine_accuracy": 0.9630876183509827,
      "eval_loss": 1.272866129875183,
      "eval_runtime": 22.2844,
      "eval_samples_per_second": 426.712,
      "eval_steps_per_second": 1.705,
      "step": 4000
    },
    {
      "epoch": 1.8775816748028538,
      "grad_norm": 3.9837265014648438,
      "learning_rate": 1.561772437101014e-05,
      "loss": 1.6857,
      "step": 5000
    },
    {
      "epoch": 1.8775816748028538,
      "eval_cosine_accuracy": 0.9651908874511719,
      "eval_loss": 1.2809278964996338,
      "eval_runtime": 22.5564,
      "eval_samples_per_second": 421.565,
      "eval_steps_per_second": 1.685,
      "step": 5000
    },
    {
      "epoch": 2.253098009763425,
      "grad_norm": 3.77929425239563,
      "learning_rate": 1.3742020277882089e-05,
      "loss": 1.5251,
      "step": 6000
    },
    {
      "epoch": 2.253098009763425,
      "eval_cosine_accuracy": 0.9658218622207642,
      "eval_loss": 1.2626943588256836,
      "eval_runtime": 23.9918,
      "eval_samples_per_second": 396.343,
      "eval_steps_per_second": 1.584,
      "step": 6000
    },
    {
      "epoch": 2.6286143447239954,
      "grad_norm": 3.7737784385681152,
      "learning_rate": 1.1864438603079236e-05,
      "loss": 1.5447,
      "step": 7000
    },
    {
      "epoch": 2.6286143447239954,
      "eval_cosine_accuracy": 0.965927004814148,
      "eval_loss": 1.2667568922042847,
      "eval_runtime": 22.5788,
      "eval_samples_per_second": 421.147,
      "eval_steps_per_second": 1.683,
      "step": 7000
    },
    {
      "epoch": 3.0041306796845664,
      "grad_norm": 3.8111581802368164,
      "learning_rate": 9.988734509951183e-06,
      "loss": 1.4464,
      "step": 8000
    },
    {
      "epoch": 3.0041306796845664,
      "eval_cosine_accuracy": 0.9660322070121765,
      "eval_loss": 1.2620376348495483,
      "eval_runtime": 22.8802,
      "eval_samples_per_second": 415.599,
      "eval_steps_per_second": 1.661,
      "step": 8000
    },
    {
      "epoch": 3.379647014645137,
      "grad_norm": 3.9379682540893555,
      "learning_rate": 8.11115283514833e-06,
      "loss": 1.4583,
      "step": 9000
    },
    {
      "epoch": 3.379647014645137,
      "eval_cosine_accuracy": 0.9686612486839294,
      "eval_loss": 1.2404857873916626,
      "eval_runtime": 22.5549,
      "eval_samples_per_second": 421.593,
      "eval_steps_per_second": 1.685,
      "step": 9000
    },
    {
      "epoch": 3.755163349605708,
      "grad_norm": 3.93530011177063,
      "learning_rate": 6.233571160345475e-06,
      "loss": 1.4374,
      "step": 10000
    },
    {
      "epoch": 3.755163349605708,
      "eval_cosine_accuracy": 0.9690819382667542,
      "eval_loss": 1.2410143613815308,
      "eval_runtime": 22.5543,
      "eval_samples_per_second": 421.605,
      "eval_steps_per_second": 1.685,
      "step": 10000
    },
    {
      "epoch": 4.130679684566279,
      "grad_norm": 3.9264919757843018,
      "learning_rate": 4.357867067217424e-06,
      "loss": 1.3556,
      "step": 11000
    },
    {
      "epoch": 4.130679684566279,
      "eval_cosine_accuracy": 0.9686612486839294,
      "eval_loss": 1.2511132955551147,
      "eval_runtime": 24.1234,
      "eval_samples_per_second": 394.181,
      "eval_steps_per_second": 1.575,
      "step": 11000
    },
    {
      "epoch": 4.50619601952685,
      "grad_norm": 4.26756477355957,
      "learning_rate": 2.4802853924145703e-06,
      "loss": 1.3997,
      "step": 12000
    },
    {
      "epoch": 4.50619601952685,
      "eval_cosine_accuracy": 0.9689767360687256,
      "eval_loss": 1.236385464668274,
      "eval_runtime": 22.4588,
      "eval_samples_per_second": 423.397,
      "eval_steps_per_second": 1.692,
      "step": 12000
    },
    {
      "epoch": 4.88171235448742,
      "grad_norm": 3.928903102874756,
      "learning_rate": 6.027037176117161e-07,
      "loss": 1.3955,
      "step": 13000
    },
    {
      "epoch": 4.88171235448742,
      "eval_cosine_accuracy": 0.9692922234535217,
      "eval_loss": 1.2428208589553833,
      "eval_runtime": 22.676,
      "eval_samples_per_second": 419.342,
      "eval_steps_per_second": 1.676,
      "step": 13000
    }
  ],
  "logging_steps": 1000,
  "max_steps": 13315,
  "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": true
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 256,
  "trial_name": null,
  "trial_params": null
}