{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.9840848806366047,
  "eval_steps": 500,
  "global_step": 40500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.04,
      "learning_rate": 4.938599076530111e-05,
      "loss": 0.5537,
      "step": 500
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.877198153060222e-05,
      "loss": 0.5484,
      "step": 1000
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.815797229590333e-05,
      "loss": 0.5563,
      "step": 1500
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.754396306120444e-05,
      "loss": 0.5339,
      "step": 2000
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.692995382650555e-05,
      "loss": 0.5478,
      "step": 2500
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.631594459180666e-05,
      "loss": 0.5462,
      "step": 3000
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.5701935357107775e-05,
      "loss": 0.5555,
      "step": 3500
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.508792612240889e-05,
      "loss": 0.5463,
      "step": 4000
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.4473916887709995e-05,
      "loss": 0.5379,
      "step": 4500
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.38599076530111e-05,
      "loss": 0.5305,
      "step": 5000
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.3245898418312215e-05,
      "loss": 0.5564,
      "step": 5500
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.263188918361332e-05,
      "loss": 0.5208,
      "step": 6000
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.2017879948914435e-05,
      "loss": 0.5264,
      "step": 6500
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.140387071421554e-05,
      "loss": 0.5595,
      "step": 7000
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.0789861479516655e-05,
      "loss": 0.556,
      "step": 7500
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.017585224481776e-05,
      "loss": 0.5514,
      "step": 8000
    },
    {
      "epoch": 0.63,
      "learning_rate": 3.9561843010118875e-05,
      "loss": 0.5334,
      "step": 8500
    },
    {
      "epoch": 0.66,
      "learning_rate": 3.894783377541998e-05,
      "loss": 0.5481,
      "step": 9000
    },
    {
      "epoch": 0.7,
      "learning_rate": 3.8333824540721095e-05,
      "loss": 0.5384,
      "step": 9500
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.771981530602221e-05,
      "loss": 0.5356,
      "step": 10000
    },
    {
      "epoch": 0.77,
      "learning_rate": 3.7105806071323314e-05,
      "loss": 0.5258,
      "step": 10500
    },
    {
      "epoch": 0.81,
      "learning_rate": 3.649179683662443e-05,
      "loss": 0.5356,
      "step": 11000
    },
    {
      "epoch": 0.85,
      "learning_rate": 3.5877787601925534e-05,
      "loss": 0.5282,
      "step": 11500
    },
    {
      "epoch": 0.88,
      "learning_rate": 3.526377836722665e-05,
      "loss": 0.5322,
      "step": 12000
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.4649769132527754e-05,
      "loss": 0.5364,
      "step": 12500
    },
    {
      "epoch": 0.96,
      "learning_rate": 3.403575989782887e-05,
      "loss": 0.5372,
      "step": 13000
    },
    {
      "epoch": 0.99,
      "learning_rate": 3.3421750663129974e-05,
      "loss": 0.5321,
      "step": 13500
    },
    {
      "epoch": 1.03,
      "learning_rate": 3.280774142843108e-05,
      "loss": 0.5254,
      "step": 14000
    },
    {
      "epoch": 1.07,
      "learning_rate": 3.2193732193732194e-05,
      "loss": 0.5168,
      "step": 14500
    },
    {
      "epoch": 1.11,
      "learning_rate": 3.15797229590333e-05,
      "loss": 0.5236,
      "step": 15000
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.0965713724334414e-05,
      "loss": 0.5276,
      "step": 15500
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.0351704489635524e-05,
      "loss": 0.511,
      "step": 16000
    },
    {
      "epoch": 1.22,
      "learning_rate": 2.9737695254936637e-05,
      "loss": 0.5144,
      "step": 16500
    },
    {
      "epoch": 1.25,
      "learning_rate": 2.9123686020237744e-05,
      "loss": 0.5215,
      "step": 17000
    },
    {
      "epoch": 1.29,
      "learning_rate": 2.8509676785538857e-05,
      "loss": 0.5143,
      "step": 17500
    },
    {
      "epoch": 1.33,
      "learning_rate": 2.7895667550839967e-05,
      "loss": 0.522,
      "step": 18000
    },
    {
      "epoch": 1.36,
      "learning_rate": 2.728165831614108e-05,
      "loss": 0.5258,
      "step": 18500
    },
    {
      "epoch": 1.4,
      "learning_rate": 2.6667649081442187e-05,
      "loss": 0.5236,
      "step": 19000
    },
    {
      "epoch": 1.44,
      "learning_rate": 2.6053639846743293e-05,
      "loss": 0.5124,
      "step": 19500
    },
    {
      "epoch": 1.47,
      "learning_rate": 2.5439630612044407e-05,
      "loss": 0.5134,
      "step": 20000
    },
    {
      "epoch": 1.51,
      "learning_rate": 2.4825621377345516e-05,
      "loss": 0.5123,
      "step": 20500
    },
    {
      "epoch": 1.55,
      "learning_rate": 2.4211612142646626e-05,
      "loss": 0.5157,
      "step": 21000
    },
    {
      "epoch": 1.58,
      "learning_rate": 2.3597602907947736e-05,
      "loss": 0.52,
      "step": 21500
    },
    {
      "epoch": 1.62,
      "learning_rate": 2.2983593673248846e-05,
      "loss": 0.5039,
      "step": 22000
    },
    {
      "epoch": 1.66,
      "learning_rate": 2.2369584438549956e-05,
      "loss": 0.5266,
      "step": 22500
    },
    {
      "epoch": 1.69,
      "learning_rate": 2.1755575203851066e-05,
      "loss": 0.5073,
      "step": 23000
    },
    {
      "epoch": 1.73,
      "learning_rate": 2.1141565969152176e-05,
      "loss": 0.5142,
      "step": 23500
    },
    {
      "epoch": 1.77,
      "learning_rate": 2.0527556734453286e-05,
      "loss": 0.5276,
      "step": 24000
    },
    {
      "epoch": 1.81,
      "learning_rate": 1.99135474997544e-05,
      "loss": 0.5156,
      "step": 24500
    },
    {
      "epoch": 1.84,
      "learning_rate": 1.929953826505551e-05,
      "loss": 0.5298,
      "step": 25000
    },
    {
      "epoch": 1.88,
      "learning_rate": 1.868552903035662e-05,
      "loss": 0.5183,
      "step": 25500
    },
    {
      "epoch": 1.92,
      "learning_rate": 1.8071519795657726e-05,
      "loss": 0.5152,
      "step": 26000
    },
    {
      "epoch": 1.95,
      "learning_rate": 1.7457510560958836e-05,
      "loss": 0.5025,
      "step": 26500
    },
    {
      "epoch": 1.99,
      "learning_rate": 1.6843501326259946e-05,
      "loss": 0.5113,
      "step": 27000
    },
    {
      "epoch": 2.03,
      "learning_rate": 1.622949209156106e-05,
      "loss": 0.512,
      "step": 27500
    },
    {
      "epoch": 2.06,
      "learning_rate": 1.561548285686217e-05,
      "loss": 0.5042,
      "step": 28000
    },
    {
      "epoch": 2.1,
      "learning_rate": 1.5001473622163279e-05,
      "loss": 0.5078,
      "step": 28500
    },
    {
      "epoch": 2.14,
      "learning_rate": 1.4387464387464389e-05,
      "loss": 0.5004,
      "step": 29000
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.3773455152765499e-05,
      "loss": 0.5042,
      "step": 29500
    },
    {
      "epoch": 2.21,
      "learning_rate": 1.315944591806661e-05,
      "loss": 0.4991,
      "step": 30000
    },
    {
      "epoch": 2.25,
      "learning_rate": 1.2545436683367717e-05,
      "loss": 0.5156,
      "step": 30500
    },
    {
      "epoch": 2.28,
      "learning_rate": 1.1931427448668829e-05,
      "loss": 0.5001,
      "step": 31000
    },
    {
      "epoch": 2.32,
      "learning_rate": 1.1317418213969938e-05,
      "loss": 0.4968,
      "step": 31500
    },
    {
      "epoch": 2.36,
      "learning_rate": 1.0703408979271048e-05,
      "loss": 0.5152,
      "step": 32000
    },
    {
      "epoch": 2.39,
      "learning_rate": 1.0089399744572158e-05,
      "loss": 0.5,
      "step": 32500
    },
    {
      "epoch": 2.43,
      "learning_rate": 9.47539050987327e-06,
      "loss": 0.5144,
      "step": 33000
    },
    {
      "epoch": 2.47,
      "learning_rate": 8.86138127517438e-06,
      "loss": 0.5103,
      "step": 33500
    },
    {
      "epoch": 2.51,
      "learning_rate": 8.247372040475488e-06,
      "loss": 0.5108,
      "step": 34000
    },
    {
      "epoch": 2.54,
      "learning_rate": 7.6333628057766e-06,
      "loss": 0.4941,
      "step": 34500
    },
    {
      "epoch": 2.58,
      "learning_rate": 7.01935357107771e-06,
      "loss": 0.5047,
      "step": 35000
    },
    {
      "epoch": 2.62,
      "learning_rate": 6.40534433637882e-06,
      "loss": 0.5011,
      "step": 35500
    },
    {
      "epoch": 2.65,
      "learning_rate": 5.7913351016799295e-06,
      "loss": 0.5069,
      "step": 36000
    },
    {
      "epoch": 2.69,
      "learning_rate": 5.17732586698104e-06,
      "loss": 0.5051,
      "step": 36500
    },
    {
      "epoch": 2.73,
      "learning_rate": 4.563316632282149e-06,
      "loss": 0.5186,
      "step": 37000
    },
    {
      "epoch": 2.76,
      "learning_rate": 3.94930739758326e-06,
      "loss": 0.5167,
      "step": 37500
    },
    {
      "epoch": 2.8,
      "learning_rate": 3.3352981628843697e-06,
      "loss": 0.5023,
      "step": 38000
    },
    {
      "epoch": 2.84,
      "learning_rate": 2.72128892818548e-06,
      "loss": 0.5044,
      "step": 38500
    },
    {
      "epoch": 2.87,
      "learning_rate": 2.1072796934865904e-06,
      "loss": 0.5128,
      "step": 39000
    },
    {
      "epoch": 2.91,
      "learning_rate": 1.4932704587877004e-06,
      "loss": 0.4992,
      "step": 39500
    },
    {
      "epoch": 2.95,
      "learning_rate": 8.792612240888103e-07,
      "loss": 0.5009,
      "step": 40000
    },
    {
      "epoch": 2.98,
      "learning_rate": 2.6525198938992043e-07,
      "loss": 0.5029,
      "step": 40500
    }
  ],
  "logging_steps": 500,
  "max_steps": 40716,
  "num_train_epochs": 3,
  "save_steps": 500,
  "total_flos": 1.0775395963109376e+17,
  "trial_name": null,
  "trial_params": null
}