{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.767357481089724,
  "global_step": 45000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.03,
      "learning_rate": 4.948752639239079e-05,
      "loss": 7.0829,
      "step": 500
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.897505278478159e-05,
      "loss": 6.3316,
      "step": 1000
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.846257917717238e-05,
      "loss": 5.9867,
      "step": 1500
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.795010556956317e-05,
      "loss": 5.7109,
      "step": 2000
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.743763196195396e-05,
      "loss": 5.374,
      "step": 2500
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.6925158354344756e-05,
      "loss": 4.9609,
      "step": 3000
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.6412684746735545e-05,
      "loss": 4.576,
      "step": 3500
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.5900211139126335e-05,
      "loss": 4.2963,
      "step": 4000
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.5387737531517125e-05,
      "loss": 4.0911,
      "step": 4500
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.487526392390792e-05,
      "loss": 3.9354,
      "step": 5000
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.436279031629871e-05,
      "loss": 3.7992,
      "step": 5500
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.38503167086895e-05,
      "loss": 3.6783,
      "step": 6000
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.333784310108029e-05,
      "loss": 3.5666,
      "step": 6500
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.282536949347109e-05,
      "loss": 3.472,
      "step": 7000
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.231289588586188e-05,
      "loss": 3.3692,
      "step": 7500
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.180042227825267e-05,
      "loss": 3.2752,
      "step": 8000
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.128794867064346e-05,
      "loss": 3.1837,
      "step": 8500
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.0775475063034255e-05,
      "loss": 3.1124,
      "step": 9000
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.0263001455425045e-05,
      "loss": 3.0376,
      "step": 9500
    },
    {
      "epoch": 0.61,
      "learning_rate": 3.9750527847815835e-05,
      "loss": 2.9563,
      "step": 10000
    },
    {
      "epoch": 0.65,
      "learning_rate": 3.923805424020663e-05,
      "loss": 2.892,
      "step": 10500
    },
    {
      "epoch": 0.68,
      "learning_rate": 3.872558063259742e-05,
      "loss": 2.8688,
      "step": 11000
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.821310702498822e-05,
      "loss": 2.8093,
      "step": 11500
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.770063341737901e-05,
      "loss": 2.7535,
      "step": 12000
    },
    {
      "epoch": 0.77,
      "learning_rate": 3.71881598097698e-05,
      "loss": 2.6915,
      "step": 12500
    },
    {
      "epoch": 0.8,
      "learning_rate": 3.6675686202160595e-05,
      "loss": 2.6404,
      "step": 13000
    },
    {
      "epoch": 0.83,
      "learning_rate": 3.6163212594551385e-05,
      "loss": 2.5972,
      "step": 13500
    },
    {
      "epoch": 0.86,
      "learning_rate": 3.5650738986942175e-05,
      "loss": 2.5741,
      "step": 14000
    },
    {
      "epoch": 0.89,
      "learning_rate": 3.513826537933297e-05,
      "loss": 2.5309,
      "step": 14500
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.462579177172376e-05,
      "loss": 2.5045,
      "step": 15000
    },
    {
      "epoch": 0.95,
      "learning_rate": 3.411331816411455e-05,
      "loss": 2.4531,
      "step": 15500
    },
    {
      "epoch": 0.98,
      "learning_rate": 3.360084455650534e-05,
      "loss": 2.4361,
      "step": 16000
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.308837094889614e-05,
      "loss": 2.3951,
      "step": 16500
    },
    {
      "epoch": 1.05,
      "learning_rate": 3.257589734128693e-05,
      "loss": 2.3564,
      "step": 17000
    },
    {
      "epoch": 1.08,
      "learning_rate": 3.206342373367772e-05,
      "loss": 2.3385,
      "step": 17500
    },
    {
      "epoch": 1.11,
      "learning_rate": 3.155095012606851e-05,
      "loss": 2.3167,
      "step": 18000
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.1038476518459304e-05,
      "loss": 2.285,
      "step": 18500
    },
    {
      "epoch": 1.17,
      "learning_rate": 3.0526002910850094e-05,
      "loss": 2.2793,
      "step": 19000
    },
    {
      "epoch": 1.2,
      "learning_rate": 3.0013529303240884e-05,
      "loss": 2.2349,
      "step": 19500
    },
    {
      "epoch": 1.23,
      "learning_rate": 2.9501055695631674e-05,
      "loss": 2.2214,
      "step": 20000
    },
    {
      "epoch": 1.26,
      "learning_rate": 2.898858208802247e-05,
      "loss": 2.1906,
      "step": 20500
    },
    {
      "epoch": 1.29,
      "learning_rate": 2.847610848041326e-05,
      "loss": 2.1662,
      "step": 21000
    },
    {
      "epoch": 1.32,
      "learning_rate": 2.796363487280405e-05,
      "loss": 2.1746,
      "step": 21500
    },
    {
      "epoch": 1.35,
      "learning_rate": 2.745116126519484e-05,
      "loss": 2.1394,
      "step": 22000
    },
    {
      "epoch": 1.38,
      "learning_rate": 2.6938687657585637e-05,
      "loss": 2.1186,
      "step": 22500
    },
    {
      "epoch": 1.41,
      "learning_rate": 2.6426214049976427e-05,
      "loss": 2.0967,
      "step": 23000
    },
    {
      "epoch": 1.45,
      "learning_rate": 2.5913740442367217e-05,
      "loss": 2.0976,
      "step": 23500
    },
    {
      "epoch": 1.48,
      "learning_rate": 2.5401266834758007e-05,
      "loss": 2.0787,
      "step": 24000
    },
    {
      "epoch": 1.51,
      "learning_rate": 2.48887932271488e-05,
      "loss": 2.0739,
      "step": 24500
    },
    {
      "epoch": 1.54,
      "learning_rate": 2.4376319619539594e-05,
      "loss": 2.0548,
      "step": 25000
    },
    {
      "epoch": 1.57,
      "learning_rate": 2.3863846011930387e-05,
      "loss": 2.0068,
      "step": 25500
    },
    {
      "epoch": 1.6,
      "learning_rate": 2.3351372404321177e-05,
      "loss": 2.0166,
      "step": 26000
    },
    {
      "epoch": 1.63,
      "learning_rate": 2.283889879671197e-05,
      "loss": 1.9973,
      "step": 26500
    },
    {
      "epoch": 1.66,
      "learning_rate": 2.232642518910276e-05,
      "loss": 1.9874,
      "step": 27000
    },
    {
      "epoch": 1.69,
      "learning_rate": 2.1813951581493554e-05,
      "loss": 1.9941,
      "step": 27500
    },
    {
      "epoch": 1.72,
      "learning_rate": 2.1301477973884347e-05,
      "loss": 1.9653,
      "step": 28000
    },
    {
      "epoch": 1.75,
      "learning_rate": 2.078900436627514e-05,
      "loss": 1.9548,
      "step": 28500
    },
    {
      "epoch": 1.78,
      "learning_rate": 2.027653075866593e-05,
      "loss": 1.9431,
      "step": 29000
    },
    {
      "epoch": 1.81,
      "learning_rate": 1.9764057151056724e-05,
      "loss": 1.9327,
      "step": 29500
    },
    {
      "epoch": 1.84,
      "learning_rate": 1.9251583543447514e-05,
      "loss": 1.9279,
      "step": 30000
    },
    {
      "epoch": 1.88,
      "learning_rate": 1.8739109935838307e-05,
      "loss": 1.9013,
      "step": 30500
    },
    {
      "epoch": 1.91,
      "learning_rate": 1.8226636328229097e-05,
      "loss": 1.8981,
      "step": 31000
    },
    {
      "epoch": 1.94,
      "learning_rate": 1.771416272061989e-05,
      "loss": 1.9058,
      "step": 31500
    },
    {
      "epoch": 1.97,
      "learning_rate": 1.720168911301068e-05,
      "loss": 1.8961,
      "step": 32000
    },
    {
      "epoch": 2.0,
      "learning_rate": 1.6689215505401473e-05,
      "loss": 1.8763,
      "step": 32500
    },
    {
      "epoch": 2.03,
      "learning_rate": 1.6176741897792263e-05,
      "loss": 1.8504,
      "step": 33000
    },
    {
      "epoch": 2.06,
      "learning_rate": 1.5664268290183057e-05,
      "loss": 1.8633,
      "step": 33500
    },
    {
      "epoch": 2.09,
      "learning_rate": 1.5151794682573847e-05,
      "loss": 1.8527,
      "step": 34000
    },
    {
      "epoch": 2.12,
      "learning_rate": 1.463932107496464e-05,
      "loss": 1.837,
      "step": 34500
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.4126847467355431e-05,
      "loss": 1.8145,
      "step": 35000
    },
    {
      "epoch": 2.18,
      "learning_rate": 1.3614373859746223e-05,
      "loss": 1.8368,
      "step": 35500
    },
    {
      "epoch": 2.21,
      "learning_rate": 1.3101900252137015e-05,
      "loss": 1.8084,
      "step": 36000
    },
    {
      "epoch": 2.24,
      "learning_rate": 1.2589426644527808e-05,
      "loss": 1.8235,
      "step": 36500
    },
    {
      "epoch": 2.28,
      "learning_rate": 1.20769530369186e-05,
      "loss": 1.8042,
      "step": 37000
    },
    {
      "epoch": 2.31,
      "learning_rate": 1.1564479429309391e-05,
      "loss": 1.7855,
      "step": 37500
    },
    {
      "epoch": 2.34,
      "learning_rate": 1.1052005821700183e-05,
      "loss": 1.7904,
      "step": 38000
    },
    {
      "epoch": 2.37,
      "learning_rate": 1.0539532214090975e-05,
      "loss": 1.7749,
      "step": 38500
    },
    {
      "epoch": 2.4,
      "learning_rate": 1.0027058606481766e-05,
      "loss": 1.785,
      "step": 39000
    },
    {
      "epoch": 2.43,
      "learning_rate": 9.514584998872558e-06,
      "loss": 1.7599,
      "step": 39500
    },
    {
      "epoch": 2.46,
      "learning_rate": 9.00211139126335e-06,
      "loss": 1.774,
      "step": 40000
    },
    {
      "epoch": 2.49,
      "learning_rate": 8.489637783654141e-06,
      "loss": 1.7442,
      "step": 40500
    },
    {
      "epoch": 2.52,
      "learning_rate": 7.977164176044934e-06,
      "loss": 1.7571,
      "step": 41000
    },
    {
      "epoch": 2.55,
      "learning_rate": 7.464690568435725e-06,
      "loss": 1.751,
      "step": 41500
    },
    {
      "epoch": 2.58,
      "learning_rate": 6.952216960826518e-06,
      "loss": 1.7367,
      "step": 42000
    },
    {
      "epoch": 2.61,
      "learning_rate": 6.439743353217309e-06,
      "loss": 1.7437,
      "step": 42500
    },
    {
      "epoch": 2.64,
      "learning_rate": 5.927269745608102e-06,
      "loss": 1.7467,
      "step": 43000
    },
    {
      "epoch": 2.68,
      "learning_rate": 5.414796137998893e-06,
      "loss": 1.7417,
      "step": 43500
    },
    {
      "epoch": 2.71,
      "learning_rate": 4.902322530389685e-06,
      "loss": 1.7392,
      "step": 44000
    },
    {
      "epoch": 2.74,
      "learning_rate": 4.389848922780477e-06,
      "loss": 1.7313,
      "step": 44500
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.877375315171269e-06,
      "loss": 1.745,
      "step": 45000
    }
  ],
  "max_steps": 48783,
  "num_train_epochs": 3,
  "total_flos": 9.548707870117478e+16,
  "trial_name": null,
  "trial_params": null
}