{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 4.0,
  "eval_steps": 1000,
  "global_step": 16570,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0002414875633904854,
      "grad_norm": 7.886749267578125,
      "learning_rate": 0.0,
      "loss": 3.6713,
      "step": 1
    },
    {
      "epoch": 0.24148756339048538,
      "grad_norm": 6.218736171722412,
      "learning_rate": 9.647513278609368e-06,
      "loss": 2.8183,
      "step": 1000
    },
    {
      "epoch": 0.24148756339048538,
      "eval_cosine_accuracy": 0.9434220194816589,
      "eval_loss": 0.585781455039978,
      "eval_runtime": 34.9645,
      "eval_samples_per_second": 271.962,
      "eval_steps_per_second": 2.145,
      "step": 1000
    },
    {
      "epoch": 0.48297512678097076,
      "grad_norm": 6.299063682556152,
      "learning_rate": 1.9304683727667796e-05,
      "loss": 2.1179,
      "step": 2000
    },
    {
      "epoch": 0.48297512678097076,
      "eval_cosine_accuracy": 0.9497318267822266,
      "eval_loss": 0.5328050851821899,
      "eval_runtime": 34.2691,
      "eval_samples_per_second": 277.48,
      "eval_steps_per_second": 2.189,
      "step": 2000
    },
    {
      "epoch": 0.7244626901714561,
      "grad_norm": 9.58321475982666,
      "learning_rate": 1.9003971235376196e-05,
      "loss": 1.4826,
      "step": 3000
    },
    {
      "epoch": 0.7244626901714561,
      "eval_cosine_accuracy": 0.9538332223892212,
      "eval_loss": 0.49323785305023193,
      "eval_runtime": 34.0684,
      "eval_samples_per_second": 279.115,
      "eval_steps_per_second": 2.201,
      "step": 3000
    },
    {
      "epoch": 0.9659502535619415,
      "grad_norm": 15.6340970993042,
      "learning_rate": 1.79317376838038e-05,
      "loss": 0.949,
      "step": 4000
    },
    {
      "epoch": 0.9659502535619415,
      "eval_cosine_accuracy": 0.9546745419502258,
      "eval_loss": 0.4723624587059021,
      "eval_runtime": 33.7991,
      "eval_samples_per_second": 281.339,
      "eval_steps_per_second": 2.219,
      "step": 4000
    },
    {
      "epoch": 1.2073376780111031,
      "grad_norm": 6.6801862716674805,
      "learning_rate": 1.6858430825372976e-05,
      "loss": 1.1823,
      "step": 5000
    },
    {
      "epoch": 1.2073376780111031,
      "eval_cosine_accuracy": 0.960037887096405,
      "eval_loss": 0.4632853865623474,
      "eval_runtime": 34.2938,
      "eval_samples_per_second": 277.28,
      "eval_steps_per_second": 2.187,
      "step": 5000
    },
    {
      "epoch": 1.4487086652184407,
      "grad_norm": 6.39561653137207,
      "learning_rate": 1.578619727380058e-05,
      "loss": 1.1665,
      "step": 6000
    },
    {
      "epoch": 1.4487086652184407,
      "eval_cosine_accuracy": 0.9617204666137695,
      "eval_loss": 0.4432311952114105,
      "eval_runtime": 35.5061,
      "eval_samples_per_second": 267.813,
      "eval_steps_per_second": 2.112,
      "step": 6000
    },
    {
      "epoch": 1.6900796524257784,
      "grad_norm": 6.977872371673584,
      "learning_rate": 1.4712890415369757e-05,
      "loss": 1.1042,
      "step": 7000
    },
    {
      "epoch": 1.6900796524257784,
      "eval_cosine_accuracy": 0.9625617861747742,
      "eval_loss": 0.43880006670951843,
      "eval_runtime": 34.1311,
      "eval_samples_per_second": 278.603,
      "eval_steps_per_second": 2.197,
      "step": 7000
    },
    {
      "epoch": 1.9314506396331161,
      "grad_norm": 7.896651268005371,
      "learning_rate": 1.3640656863797362e-05,
      "loss": 1.0525,
      "step": 8000
    },
    {
      "epoch": 1.9314506396331161,
      "eval_cosine_accuracy": 0.9643495678901672,
      "eval_loss": 0.4344501495361328,
      "eval_runtime": 34.0153,
      "eval_samples_per_second": 279.55,
      "eval_steps_per_second": 2.205,
      "step": 8000
    },
    {
      "epoch": 2.1728216268404537,
      "grad_norm": 6.305137634277344,
      "learning_rate": 1.2567350005366534e-05,
      "loss": 0.9752,
      "step": 9000
    },
    {
      "epoch": 2.1728216268404537,
      "eval_cosine_accuracy": 0.9641392230987549,
      "eval_loss": 0.43460169434547424,
      "eval_runtime": 33.9573,
      "eval_samples_per_second": 280.028,
      "eval_steps_per_second": 2.209,
      "step": 9000
    },
    {
      "epoch": 2.4141926140477916,
      "grad_norm": 6.064307689666748,
      "learning_rate": 1.1495116453794142e-05,
      "loss": 0.9177,
      "step": 10000
    },
    {
      "epoch": 2.4141926140477916,
      "eval_cosine_accuracy": 0.9636133909225464,
      "eval_loss": 0.4276249408721924,
      "eval_runtime": 34.07,
      "eval_samples_per_second": 279.102,
      "eval_steps_per_second": 2.201,
      "step": 10000
    },
    {
      "epoch": 2.655563601255129,
      "grad_norm": 11.2512788772583,
      "learning_rate": 1.0421809595363315e-05,
      "loss": 0.9044,
      "step": 11000
    },
    {
      "epoch": 2.655563601255129,
      "eval_cosine_accuracy": 0.9652960300445557,
      "eval_loss": 0.42561274766921997,
      "eval_runtime": 34.6345,
      "eval_samples_per_second": 274.553,
      "eval_steps_per_second": 2.165,
      "step": 11000
    },
    {
      "epoch": 2.8969345884624667,
      "grad_norm": 5.970104694366455,
      "learning_rate": 9.34850273693249e-06,
      "loss": 0.8924,
      "step": 12000
    },
    {
      "epoch": 2.8969345884624667,
      "eval_cosine_accuracy": 0.9664528369903564,
      "eval_loss": 0.42230626940727234,
      "eval_runtime": 33.7931,
      "eval_samples_per_second": 281.389,
      "eval_steps_per_second": 2.219,
      "step": 12000
    },
    {
      "epoch": 3.1383055756698046,
      "grad_norm": 6.617201805114746,
      "learning_rate": 8.275195878501664e-06,
      "loss": 0.8378,
      "step": 13000
    },
    {
      "epoch": 3.1383055756698046,
      "eval_cosine_accuracy": 0.9656115174293518,
      "eval_loss": 0.42513710260391235,
      "eval_runtime": 33.8615,
      "eval_samples_per_second": 280.821,
      "eval_steps_per_second": 2.215,
      "step": 13000
    },
    {
      "epoch": 3.379676562877142,
      "grad_norm": 5.853105545043945,
      "learning_rate": 7.201889020070839e-06,
      "loss": 0.831,
      "step": 14000
    },
    {
      "epoch": 3.379676562877142,
      "eval_cosine_accuracy": 0.9662424921989441,
      "eval_loss": 0.42466071248054504,
      "eval_runtime": 33.8525,
      "eval_samples_per_second": 280.895,
      "eval_steps_per_second": 2.215,
      "step": 14000
    },
    {
      "epoch": 3.62104755008448,
      "grad_norm": 5.566030025482178,
      "learning_rate": 6.128582161640013e-06,
      "loss": 0.8012,
      "step": 15000
    },
    {
      "epoch": 3.62104755008448,
      "eval_cosine_accuracy": 0.9660322070121765,
      "eval_loss": 0.4249342978000641,
      "eval_runtime": 34.3475,
      "eval_samples_per_second": 276.847,
      "eval_steps_per_second": 2.184,
      "step": 15000
    },
    {
      "epoch": 3.8624185372918176,
      "grad_norm": 5.444942474365234,
      "learning_rate": 5.056348610067619e-06,
      "loss": 0.7952,
      "step": 16000
    },
    {
      "epoch": 3.8624185372918176,
      "eval_cosine_accuracy": 0.9661373496055603,
      "eval_loss": 0.42102375626564026,
      "eval_runtime": 33.7901,
      "eval_samples_per_second": 281.414,
      "eval_steps_per_second": 2.22,
      "step": 16000
    }
  ],
  "logging_steps": 1000,
  "max_steps": 20705,
  "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
}