{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "global_step": 48159,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.03,
      "learning_rate": 4.94808862310264e-05,
      "loss": 0.6922,
      "step": 500
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.8961772462052786e-05,
      "loss": 0.3808,
      "step": 1000
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.844265869307918e-05,
      "loss": 0.3117,
      "step": 1500
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.792354492410557e-05,
      "loss": 0.2758,
      "step": 2000
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.7404431155131964e-05,
      "loss": 0.251,
      "step": 2500
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.688531738615835e-05,
      "loss": 0.2316,
      "step": 3000
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.636620361718475e-05,
      "loss": 0.2138,
      "step": 3500
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.5847089848211136e-05,
      "loss": 0.1987,
      "step": 4000
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.5327976079237525e-05,
      "loss": 0.1947,
      "step": 4500
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.480886231026392e-05,
      "loss": 0.1797,
      "step": 5000
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.428974854129031e-05,
      "loss": 0.1747,
      "step": 5500
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.37706347723167e-05,
      "loss": 0.1664,
      "step": 6000
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.325152100334309e-05,
      "loss": 0.1611,
      "step": 6500
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.273240723436949e-05,
      "loss": 0.1529,
      "step": 7000
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.2213293465395875e-05,
      "loss": 0.149,
      "step": 7500
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.169417969642227e-05,
      "loss": 0.1505,
      "step": 8000
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.1175065927448665e-05,
      "loss": 0.1459,
      "step": 8500
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.0655952158475054e-05,
      "loss": 0.1406,
      "step": 9000
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.013683838950145e-05,
      "loss": 0.1359,
      "step": 9500
    },
    {
      "epoch": 0.62,
      "learning_rate": 3.961772462052784e-05,
      "loss": 0.1326,
      "step": 10000
    },
    {
      "epoch": 0.65,
      "learning_rate": 3.909861085155423e-05,
      "loss": 0.132,
      "step": 10500
    },
    {
      "epoch": 0.69,
      "learning_rate": 3.857949708258062e-05,
      "loss": 0.13,
      "step": 11000
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.806038331360701e-05,
      "loss": 0.124,
      "step": 11500
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.7541269544633404e-05,
      "loss": 0.1242,
      "step": 12000
    },
    {
      "epoch": 0.78,
      "learning_rate": 3.702215577565979e-05,
      "loss": 0.1244,
      "step": 12500
    },
    {
      "epoch": 0.81,
      "learning_rate": 3.650304200668619e-05,
      "loss": 0.1216,
      "step": 13000
    },
    {
      "epoch": 0.84,
      "learning_rate": 3.5983928237712576e-05,
      "loss": 0.1177,
      "step": 13500
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.546481446873897e-05,
      "loss": 0.1176,
      "step": 14000
    },
    {
      "epoch": 0.9,
      "learning_rate": 3.494570069976536e-05,
      "loss": 0.1144,
      "step": 14500
    },
    {
      "epoch": 0.93,
      "learning_rate": 3.4426586930791755e-05,
      "loss": 0.1127,
      "step": 15000
    },
    {
      "epoch": 0.97,
      "learning_rate": 3.390747316181814e-05,
      "loss": 0.1111,
      "step": 15500
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.338835939284454e-05,
      "loss": 0.1099,
      "step": 16000
    },
    {
      "epoch": 1.03,
      "learning_rate": 3.286924562387093e-05,
      "loss": 0.1009,
      "step": 16500
    },
    {
      "epoch": 1.06,
      "learning_rate": 3.235013185489732e-05,
      "loss": 0.1,
      "step": 17000
    },
    {
      "epoch": 1.09,
      "learning_rate": 3.183101808592372e-05,
      "loss": 0.0976,
      "step": 17500
    },
    {
      "epoch": 1.12,
      "learning_rate": 3.1311904316950105e-05,
      "loss": 0.0953,
      "step": 18000
    },
    {
      "epoch": 1.15,
      "learning_rate": 3.079279054797649e-05,
      "loss": 0.0953,
      "step": 18500
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.0273676779002885e-05,
      "loss": 0.0978,
      "step": 19000
    },
    {
      "epoch": 1.21,
      "learning_rate": 2.975456301002928e-05,
      "loss": 0.0945,
      "step": 19500
    },
    {
      "epoch": 1.25,
      "learning_rate": 2.9235449241055672e-05,
      "loss": 0.0962,
      "step": 20000
    },
    {
      "epoch": 1.28,
      "learning_rate": 2.8716335472082064e-05,
      "loss": 0.0936,
      "step": 20500
    },
    {
      "epoch": 1.31,
      "learning_rate": 2.8197221703108455e-05,
      "loss": 0.0952,
      "step": 21000
    },
    {
      "epoch": 1.34,
      "learning_rate": 2.7678107934134844e-05,
      "loss": 0.0927,
      "step": 21500
    },
    {
      "epoch": 1.37,
      "learning_rate": 2.715899416516124e-05,
      "loss": 0.0913,
      "step": 22000
    },
    {
      "epoch": 1.4,
      "learning_rate": 2.6639880396187627e-05,
      "loss": 0.0921,
      "step": 22500
    },
    {
      "epoch": 1.43,
      "learning_rate": 2.6120766627214022e-05,
      "loss": 0.0895,
      "step": 23000
    },
    {
      "epoch": 1.46,
      "learning_rate": 2.5601652858240418e-05,
      "loss": 0.0869,
      "step": 23500
    },
    {
      "epoch": 1.5,
      "learning_rate": 2.5082539089266806e-05,
      "loss": 0.0898,
      "step": 24000
    },
    {
      "epoch": 1.53,
      "learning_rate": 2.4563425320293194e-05,
      "loss": 0.0867,
      "step": 24500
    },
    {
      "epoch": 1.56,
      "learning_rate": 2.4044311551319586e-05,
      "loss": 0.0889,
      "step": 25000
    },
    {
      "epoch": 1.59,
      "learning_rate": 2.352519778234598e-05,
      "loss": 0.0858,
      "step": 25500
    },
    {
      "epoch": 1.62,
      "learning_rate": 2.3006084013372373e-05,
      "loss": 0.0856,
      "step": 26000
    },
    {
      "epoch": 1.65,
      "learning_rate": 2.2486970244398765e-05,
      "loss": 0.0847,
      "step": 26500
    },
    {
      "epoch": 1.68,
      "learning_rate": 2.1967856475425156e-05,
      "loss": 0.0861,
      "step": 27000
    },
    {
      "epoch": 1.71,
      "learning_rate": 2.1448742706451548e-05,
      "loss": 0.0847,
      "step": 27500
    },
    {
      "epoch": 1.74,
      "learning_rate": 2.0929628937477936e-05,
      "loss": 0.0819,
      "step": 28000
    },
    {
      "epoch": 1.78,
      "learning_rate": 2.0410515168504328e-05,
      "loss": 0.0819,
      "step": 28500
    },
    {
      "epoch": 1.81,
      "learning_rate": 1.989140139953072e-05,
      "loss": 0.0856,
      "step": 29000
    },
    {
      "epoch": 1.84,
      "learning_rate": 1.9372287630557115e-05,
      "loss": 0.0828,
      "step": 29500
    },
    {
      "epoch": 1.87,
      "learning_rate": 1.8853173861583507e-05,
      "loss": 0.0832,
      "step": 30000
    },
    {
      "epoch": 1.9,
      "learning_rate": 1.83340600926099e-05,
      "loss": 0.0823,
      "step": 30500
    },
    {
      "epoch": 1.93,
      "learning_rate": 1.781494632363629e-05,
      "loss": 0.0808,
      "step": 31000
    },
    {
      "epoch": 1.96,
      "learning_rate": 1.7295832554662682e-05,
      "loss": 0.0811,
      "step": 31500
    },
    {
      "epoch": 1.99,
      "learning_rate": 1.677671878568907e-05,
      "loss": 0.084,
      "step": 32000
    },
    {
      "epoch": 2.02,
      "learning_rate": 1.6257605016715462e-05,
      "loss": 0.0769,
      "step": 32500
    },
    {
      "epoch": 2.06,
      "learning_rate": 1.5738491247741857e-05,
      "loss": 0.0733,
      "step": 33000
    },
    {
      "epoch": 2.09,
      "learning_rate": 1.5219377478768249e-05,
      "loss": 0.0724,
      "step": 33500
    },
    {
      "epoch": 2.12,
      "learning_rate": 1.470026370979464e-05,
      "loss": 0.0747,
      "step": 34000
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.418114994082103e-05,
      "loss": 0.0715,
      "step": 34500
    },
    {
      "epoch": 2.18,
      "learning_rate": 1.3662036171847422e-05,
      "loss": 0.0736,
      "step": 35000
    },
    {
      "epoch": 2.21,
      "learning_rate": 1.3142922402873814e-05,
      "loss": 0.0745,
      "step": 35500
    },
    {
      "epoch": 2.24,
      "learning_rate": 1.2623808633900206e-05,
      "loss": 0.0732,
      "step": 36000
    },
    {
      "epoch": 2.27,
      "learning_rate": 1.2104694864926598e-05,
      "loss": 0.0738,
      "step": 36500
    },
    {
      "epoch": 2.3,
      "learning_rate": 1.158558109595299e-05,
      "loss": 0.0739,
      "step": 37000
    },
    {
      "epoch": 2.34,
      "learning_rate": 1.1066467326979381e-05,
      "loss": 0.0703,
      "step": 37500
    },
    {
      "epoch": 2.37,
      "learning_rate": 1.0547353558005773e-05,
      "loss": 0.072,
      "step": 38000
    },
    {
      "epoch": 2.4,
      "learning_rate": 1.0028239789032165e-05,
      "loss": 0.0723,
      "step": 38500
    },
    {
      "epoch": 2.43,
      "learning_rate": 9.509126020058556e-06,
      "loss": 0.0734,
      "step": 39000
    },
    {
      "epoch": 2.46,
      "learning_rate": 8.990012251084948e-06,
      "loss": 0.0702,
      "step": 39500
    },
    {
      "epoch": 2.49,
      "learning_rate": 8.47089848211134e-06,
      "loss": 0.0712,
      "step": 40000
    },
    {
      "epoch": 2.52,
      "learning_rate": 7.951784713137732e-06,
      "loss": 0.0713,
      "step": 40500
    },
    {
      "epoch": 2.55,
      "learning_rate": 7.432670944164123e-06,
      "loss": 0.0719,
      "step": 41000
    },
    {
      "epoch": 2.59,
      "learning_rate": 6.913557175190514e-06,
      "loss": 0.0718,
      "step": 41500
    },
    {
      "epoch": 2.62,
      "learning_rate": 6.394443406216908e-06,
      "loss": 0.0709,
      "step": 42000
    },
    {
      "epoch": 2.65,
      "learning_rate": 5.8753296372432985e-06,
      "loss": 0.07,
      "step": 42500
    },
    {
      "epoch": 2.68,
      "learning_rate": 5.35621586826969e-06,
      "loss": 0.068,
      "step": 43000
    },
    {
      "epoch": 2.71,
      "learning_rate": 4.837102099296082e-06,
      "loss": 0.0714,
      "step": 43500
    },
    {
      "epoch": 2.74,
      "learning_rate": 4.317988330322474e-06,
      "loss": 0.0719,
      "step": 44000
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.7988745613488655e-06,
      "loss": 0.0678,
      "step": 44500
    },
    {
      "epoch": 2.8,
      "learning_rate": 3.2797607923752568e-06,
      "loss": 0.0693,
      "step": 45000
    },
    {
      "epoch": 2.83,
      "learning_rate": 2.760647023401649e-06,
      "loss": 0.0677,
      "step": 45500
    },
    {
      "epoch": 2.87,
      "learning_rate": 2.2415332544280407e-06,
      "loss": 0.0711,
      "step": 46000
    },
    {
      "epoch": 2.9,
      "learning_rate": 1.7224194854544324e-06,
      "loss": 0.0692,
      "step": 46500
    },
    {
      "epoch": 2.93,
      "learning_rate": 1.203305716480824e-06,
      "loss": 0.0692,
      "step": 47000
    },
    {
      "epoch": 2.96,
      "learning_rate": 6.841919475072158e-07,
      "loss": 0.0699,
      "step": 47500
    },
    {
      "epoch": 2.99,
      "learning_rate": 1.6507817853360744e-07,
      "loss": 0.0711,
      "step": 48000
    },
    {
      "epoch": 3.0,
      "step": 48159,
      "total_flos": 5.191245227822285e+16,
      "train_loss": 0.11467816326249321,
      "train_runtime": 7321.5952,
      "train_samples_per_second": 210.473,
      "train_steps_per_second": 6.578
    }
  ],
  "max_steps": 48159,
  "num_train_epochs": 3,
  "total_flos": 5.191245227822285e+16,
  "trial_name": null,
  "trial_params": null
}