{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.9997518764344644,
  "global_step": 16120,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.12,
      "learning_rate": 4.8449131513647645e-05,
      "loss": 7.5778,
      "step": 500
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.689826302729529e-05,
      "loss": 5.6514,
      "step": 1000
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.534739454094293e-05,
      "loss": 4.753,
      "step": 1500
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.379652605459057e-05,
      "loss": 4.1203,
      "step": 2000
    },
    {
      "epoch": 0.62,
      "learning_rate": 4.2245657568238216e-05,
      "loss": 3.6995,
      "step": 2500
    },
    {
      "epoch": 0.74,
      "learning_rate": 4.069478908188586e-05,
      "loss": 3.4027,
      "step": 3000
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.91439205955335e-05,
      "loss": 3.1636,
      "step": 3500
    },
    {
      "epoch": 0.99,
      "learning_rate": 3.7593052109181144e-05,
      "loss": 2.9862,
      "step": 4000
    },
    {
      "epoch": 1.12,
      "learning_rate": 3.6042183622828787e-05,
      "loss": 2.8469,
      "step": 4500
    },
    {
      "epoch": 1.24,
      "learning_rate": 3.449131513647643e-05,
      "loss": 2.7431,
      "step": 5000
    },
    {
      "epoch": 1.36,
      "learning_rate": 3.294044665012407e-05,
      "loss": 2.6552,
      "step": 5500
    },
    {
      "epoch": 1.49,
      "learning_rate": 3.1389578163771715e-05,
      "loss": 2.5856,
      "step": 6000
    },
    {
      "epoch": 1.61,
      "learning_rate": 2.9838709677419357e-05,
      "loss": 2.5195,
      "step": 6500
    },
    {
      "epoch": 1.74,
      "learning_rate": 2.8287841191067e-05,
      "loss": 2.4654,
      "step": 7000
    },
    {
      "epoch": 1.86,
      "learning_rate": 2.6736972704714643e-05,
      "loss": 2.4238,
      "step": 7500
    },
    {
      "epoch": 1.98,
      "learning_rate": 2.5186104218362282e-05,
      "loss": 2.3821,
      "step": 8000
    },
    {
      "epoch": 2.11,
      "learning_rate": 2.3635235732009928e-05,
      "loss": 2.3322,
      "step": 8500
    },
    {
      "epoch": 2.23,
      "learning_rate": 2.208436724565757e-05,
      "loss": 2.3004,
      "step": 9000
    },
    {
      "epoch": 2.36,
      "learning_rate": 2.053349875930521e-05,
      "loss": 2.2745,
      "step": 9500
    },
    {
      "epoch": 2.48,
      "learning_rate": 1.8982630272952853e-05,
      "loss": 2.2494,
      "step": 10000
    },
    {
      "epoch": 2.61,
      "learning_rate": 1.74317617866005e-05,
      "loss": 2.2309,
      "step": 10500
    },
    {
      "epoch": 2.73,
      "learning_rate": 1.588089330024814e-05,
      "loss": 2.2102,
      "step": 11000
    },
    {
      "epoch": 2.85,
      "learning_rate": 1.4330024813895782e-05,
      "loss": 2.1912,
      "step": 11500
    },
    {
      "epoch": 2.98,
      "learning_rate": 1.2779156327543427e-05,
      "loss": 2.1743,
      "step": 12000
    },
    {
      "epoch": 3.1,
      "learning_rate": 1.1228287841191068e-05,
      "loss": 2.1509,
      "step": 12500
    },
    {
      "epoch": 3.23,
      "learning_rate": 9.67741935483871e-06,
      "loss": 2.1385,
      "step": 13000
    },
    {
      "epoch": 3.35,
      "learning_rate": 8.126550868486353e-06,
      "loss": 2.1343,
      "step": 13500
    },
    {
      "epoch": 3.47,
      "learning_rate": 6.575682382133995e-06,
      "loss": 2.1247,
      "step": 14000
    },
    {
      "epoch": 3.6,
      "learning_rate": 5.0248138957816385e-06,
      "loss": 2.1188,
      "step": 14500
    },
    {
      "epoch": 3.72,
      "learning_rate": 3.4739454094292807e-06,
      "loss": 2.1198,
      "step": 15000
    },
    {
      "epoch": 3.85,
      "learning_rate": 1.9230769230769234e-06,
      "loss": 2.1102,
      "step": 15500
    },
    {
      "epoch": 3.97,
      "learning_rate": 3.7220843672456576e-07,
      "loss": 2.106,
      "step": 16000
    },
    {
      "epoch": 4.0,
      "step": 16120,
      "total_flos": 4.540075318875832e+17,
      "train_loss": 2.836493848511954,
      "train_runtime": 50326.1237,
      "train_samples_per_second": 307.517,
      "train_steps_per_second": 0.32
    }
  ],
  "max_steps": 16120,
  "num_train_epochs": 4,
  "total_flos": 4.540075318875832e+17,
  "trial_name": null,
  "trial_params": null
}