{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "global_step": 61023,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.02,
      "learning_rate": 4.9590318404536e-05,
      "loss": 1.7769,
      "step": 500
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9180636809071987e-05,
      "loss": 0.7561,
      "step": 1000
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.877095521360799e-05,
      "loss": 0.694,
      "step": 1500
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8361273618143984e-05,
      "loss": 0.6611,
      "step": 2000
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.795159202267997e-05,
      "loss": 0.6388,
      "step": 2500
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.754191042721597e-05,
      "loss": 0.6214,
      "step": 3000
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.7132228831751964e-05,
      "loss": 0.6166,
      "step": 3500
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.672254723628796e-05,
      "loss": 0.615,
      "step": 4000
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.6312865640823955e-05,
      "loss": 0.6017,
      "step": 4500
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.590318404535995e-05,
      "loss": 0.596,
      "step": 5000
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.549350244989594e-05,
      "loss": 0.592,
      "step": 5500
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.5083820854431934e-05,
      "loss": 0.5848,
      "step": 6000
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.4674139258967936e-05,
      "loss": 0.581,
      "step": 6500
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.4264457663503925e-05,
      "loss": 0.5779,
      "step": 7000
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.385477606803992e-05,
      "loss": 0.5692,
      "step": 7500
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.3445094472575916e-05,
      "loss": 0.5604,
      "step": 8000
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.303541287711191e-05,
      "loss": 0.5536,
      "step": 8500
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.262573128164791e-05,
      "loss": 0.5418,
      "step": 9000
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.22160496861839e-05,
      "loss": 0.5366,
      "step": 9500
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.180636809071989e-05,
      "loss": 0.526,
      "step": 10000
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.1396686495255886e-05,
      "loss": 0.522,
      "step": 10500
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.098700489979188e-05,
      "loss": 0.5135,
      "step": 11000
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.057732330432788e-05,
      "loss": 0.5078,
      "step": 11500
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.016764170886387e-05,
      "loss": 0.5082,
      "step": 12000
    },
    {
      "epoch": 0.61,
      "learning_rate": 3.975796011339987e-05,
      "loss": 0.4986,
      "step": 12500
    },
    {
      "epoch": 0.64,
      "learning_rate": 3.9348278517935864e-05,
      "loss": 0.4907,
      "step": 13000
    },
    {
      "epoch": 0.66,
      "learning_rate": 3.893859692247186e-05,
      "loss": 0.4896,
      "step": 13500
    },
    {
      "epoch": 0.69,
      "learning_rate": 3.8528915327007854e-05,
      "loss": 0.4836,
      "step": 14000
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.811923373154384e-05,
      "loss": 0.4822,
      "step": 14500
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.770955213607984e-05,
      "loss": 0.4734,
      "step": 15000
    },
    {
      "epoch": 0.76,
      "learning_rate": 3.7299870540615834e-05,
      "loss": 0.4733,
      "step": 15500
    },
    {
      "epoch": 0.79,
      "learning_rate": 3.689018894515183e-05,
      "loss": 0.4721,
      "step": 16000
    },
    {
      "epoch": 0.81,
      "learning_rate": 3.6480507349687825e-05,
      "loss": 0.4689,
      "step": 16500
    },
    {
      "epoch": 0.84,
      "learning_rate": 3.607082575422382e-05,
      "loss": 0.4694,
      "step": 17000
    },
    {
      "epoch": 0.86,
      "learning_rate": 3.5661144158759816e-05,
      "loss": 0.4656,
      "step": 17500
    },
    {
      "epoch": 0.88,
      "learning_rate": 3.5251462563295804e-05,
      "loss": 0.4663,
      "step": 18000
    },
    {
      "epoch": 0.91,
      "learning_rate": 3.484178096783181e-05,
      "loss": 0.4671,
      "step": 18500
    },
    {
      "epoch": 0.93,
      "learning_rate": 3.4432099372367795e-05,
      "loss": 0.4623,
      "step": 19000
    },
    {
      "epoch": 0.96,
      "learning_rate": 3.402241777690379e-05,
      "loss": 0.4615,
      "step": 19500
    },
    {
      "epoch": 0.98,
      "learning_rate": 3.3612736181439786e-05,
      "loss": 0.4528,
      "step": 20000
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.320305458597578e-05,
      "loss": 0.4572,
      "step": 20500
    },
    {
      "epoch": 1.03,
      "learning_rate": 3.279337299051178e-05,
      "loss": 0.4524,
      "step": 21000
    },
    {
      "epoch": 1.06,
      "learning_rate": 3.238369139504777e-05,
      "loss": 0.4504,
      "step": 21500
    },
    {
      "epoch": 1.08,
      "learning_rate": 3.197400979958377e-05,
      "loss": 0.4563,
      "step": 22000
    },
    {
      "epoch": 1.11,
      "learning_rate": 3.156432820411976e-05,
      "loss": 0.4513,
      "step": 22500
    },
    {
      "epoch": 1.13,
      "learning_rate": 3.115464660865575e-05,
      "loss": 0.4503,
      "step": 23000
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.0744965013191754e-05,
      "loss": 0.4457,
      "step": 23500
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.0335283417727743e-05,
      "loss": 0.4488,
      "step": 24000
    },
    {
      "epoch": 1.2,
      "learning_rate": 2.992560182226374e-05,
      "loss": 0.4497,
      "step": 24500
    },
    {
      "epoch": 1.23,
      "learning_rate": 2.951592022679973e-05,
      "loss": 0.4481,
      "step": 25000
    },
    {
      "epoch": 1.25,
      "learning_rate": 2.9106238631335726e-05,
      "loss": 0.4439,
      "step": 25500
    },
    {
      "epoch": 1.28,
      "learning_rate": 2.8696557035871725e-05,
      "loss": 0.4393,
      "step": 26000
    },
    {
      "epoch": 1.3,
      "learning_rate": 2.8286875440407717e-05,
      "loss": 0.441,
      "step": 26500
    },
    {
      "epoch": 1.33,
      "learning_rate": 2.7877193844943712e-05,
      "loss": 0.4394,
      "step": 27000
    },
    {
      "epoch": 1.35,
      "learning_rate": 2.7467512249479704e-05,
      "loss": 0.4387,
      "step": 27500
    },
    {
      "epoch": 1.38,
      "learning_rate": 2.7057830654015696e-05,
      "loss": 0.4341,
      "step": 28000
    },
    {
      "epoch": 1.4,
      "learning_rate": 2.6648149058551695e-05,
      "loss": 0.4345,
      "step": 28500
    },
    {
      "epoch": 1.43,
      "learning_rate": 2.623846746308769e-05,
      "loss": 0.4368,
      "step": 29000
    },
    {
      "epoch": 1.45,
      "learning_rate": 2.5828785867623683e-05,
      "loss": 0.4348,
      "step": 29500
    },
    {
      "epoch": 1.47,
      "learning_rate": 2.5419104272159678e-05,
      "loss": 0.4324,
      "step": 30000
    },
    {
      "epoch": 1.5,
      "learning_rate": 2.500942267669567e-05,
      "loss": 0.4321,
      "step": 30500
    },
    {
      "epoch": 1.52,
      "learning_rate": 2.4599741081231666e-05,
      "loss": 0.4313,
      "step": 31000
    },
    {
      "epoch": 1.55,
      "learning_rate": 2.4190059485767664e-05,
      "loss": 0.4266,
      "step": 31500
    },
    {
      "epoch": 1.57,
      "learning_rate": 2.3780377890303656e-05,
      "loss": 0.4287,
      "step": 32000
    },
    {
      "epoch": 1.6,
      "learning_rate": 2.3370696294839652e-05,
      "loss": 0.4215,
      "step": 32500
    },
    {
      "epoch": 1.62,
      "learning_rate": 2.2961014699375647e-05,
      "loss": 0.4255,
      "step": 33000
    },
    {
      "epoch": 1.65,
      "learning_rate": 2.255133310391164e-05,
      "loss": 0.4258,
      "step": 33500
    },
    {
      "epoch": 1.67,
      "learning_rate": 2.2141651508447635e-05,
      "loss": 0.4255,
      "step": 34000
    },
    {
      "epoch": 1.7,
      "learning_rate": 2.173196991298363e-05,
      "loss": 0.4212,
      "step": 34500
    },
    {
      "epoch": 1.72,
      "learning_rate": 2.1322288317519626e-05,
      "loss": 0.4218,
      "step": 35000
    },
    {
      "epoch": 1.75,
      "learning_rate": 2.0912606722055618e-05,
      "loss": 0.4179,
      "step": 35500
    },
    {
      "epoch": 1.77,
      "learning_rate": 2.0502925126591613e-05,
      "loss": 0.4205,
      "step": 36000
    },
    {
      "epoch": 1.79,
      "learning_rate": 2.009324353112761e-05,
      "loss": 0.4183,
      "step": 36500
    },
    {
      "epoch": 1.82,
      "learning_rate": 1.9683561935663604e-05,
      "loss": 0.4175,
      "step": 37000
    },
    {
      "epoch": 1.84,
      "learning_rate": 1.92738803401996e-05,
      "loss": 0.4196,
      "step": 37500
    },
    {
      "epoch": 1.87,
      "learning_rate": 1.886419874473559e-05,
      "loss": 0.4183,
      "step": 38000
    },
    {
      "epoch": 1.89,
      "learning_rate": 1.8454517149271587e-05,
      "loss": 0.4148,
      "step": 38500
    },
    {
      "epoch": 1.92,
      "learning_rate": 1.8044835553807583e-05,
      "loss": 0.4103,
      "step": 39000
    },
    {
      "epoch": 1.94,
      "learning_rate": 1.7635153958343575e-05,
      "loss": 0.4167,
      "step": 39500
    },
    {
      "epoch": 1.97,
      "learning_rate": 1.722547236287957e-05,
      "loss": 0.4092,
      "step": 40000
    },
    {
      "epoch": 1.99,
      "learning_rate": 1.6815790767415565e-05,
      "loss": 0.4124,
      "step": 40500
    },
    {
      "epoch": 2.02,
      "learning_rate": 1.640610917195156e-05,
      "loss": 0.4107,
      "step": 41000
    },
    {
      "epoch": 2.04,
      "learning_rate": 1.5996427576487556e-05,
      "loss": 0.4117,
      "step": 41500
    },
    {
      "epoch": 2.06,
      "learning_rate": 1.558674598102355e-05,
      "loss": 0.4107,
      "step": 42000
    },
    {
      "epoch": 2.09,
      "learning_rate": 1.5177064385559544e-05,
      "loss": 0.4116,
      "step": 42500
    },
    {
      "epoch": 2.11,
      "learning_rate": 1.4767382790095538e-05,
      "loss": 0.4125,
      "step": 43000
    },
    {
      "epoch": 2.14,
      "learning_rate": 1.4357701194631535e-05,
      "loss": 0.4116,
      "step": 43500
    },
    {
      "epoch": 2.16,
      "learning_rate": 1.3948019599167528e-05,
      "loss": 0.4083,
      "step": 44000
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.353833800370352e-05,
      "loss": 0.4103,
      "step": 44500
    },
    {
      "epoch": 2.21,
      "learning_rate": 1.3128656408239518e-05,
      "loss": 0.4111,
      "step": 45000
    },
    {
      "epoch": 2.24,
      "learning_rate": 1.2718974812775511e-05,
      "loss": 0.4023,
      "step": 45500
    },
    {
      "epoch": 2.26,
      "learning_rate": 1.2309293217311505e-05,
      "loss": 0.4069,
      "step": 46000
    },
    {
      "epoch": 2.29,
      "learning_rate": 1.18996116218475e-05,
      "loss": 0.4063,
      "step": 46500
    },
    {
      "epoch": 2.31,
      "learning_rate": 1.1489930026383496e-05,
      "loss": 0.4078,
      "step": 47000
    },
    {
      "epoch": 2.34,
      "learning_rate": 1.108024843091949e-05,
      "loss": 0.4036,
      "step": 47500
    },
    {
      "epoch": 2.36,
      "learning_rate": 1.0670566835455484e-05,
      "loss": 0.4065,
      "step": 48000
    },
    {
      "epoch": 2.38,
      "learning_rate": 1.0260885239991479e-05,
      "loss": 0.403,
      "step": 48500
    },
    {
      "epoch": 2.41,
      "learning_rate": 9.851203644527474e-06,
      "loss": 0.3999,
      "step": 49000
    },
    {
      "epoch": 2.43,
      "learning_rate": 9.441522049063468e-06,
      "loss": 0.4095,
      "step": 49500
    },
    {
      "epoch": 2.46,
      "learning_rate": 9.031840453599464e-06,
      "loss": 0.4048,
      "step": 50000
    },
    {
      "epoch": 2.48,
      "learning_rate": 8.622158858135457e-06,
      "loss": 0.3998,
      "step": 50500
    },
    {
      "epoch": 2.51,
      "learning_rate": 8.212477262671451e-06,
      "loss": 0.4,
      "step": 51000
    },
    {
      "epoch": 2.53,
      "learning_rate": 7.802795667207447e-06,
      "loss": 0.4019,
      "step": 51500
    },
    {
      "epoch": 2.56,
      "learning_rate": 7.393114071743441e-06,
      "loss": 0.3953,
      "step": 52000
    },
    {
      "epoch": 2.58,
      "learning_rate": 6.983432476279437e-06,
      "loss": 0.3952,
      "step": 52500
    },
    {
      "epoch": 2.61,
      "learning_rate": 6.573750880815431e-06,
      "loss": 0.3996,
      "step": 53000
    },
    {
      "epoch": 2.63,
      "learning_rate": 6.164069285351425e-06,
      "loss": 0.4002,
      "step": 53500
    },
    {
      "epoch": 2.65,
      "learning_rate": 5.75438768988742e-06,
      "loss": 0.3983,
      "step": 54000
    },
    {
      "epoch": 2.68,
      "learning_rate": 5.344706094423414e-06,
      "loss": 0.394,
      "step": 54500
    },
    {
      "epoch": 2.7,
      "learning_rate": 4.935024498959409e-06,
      "loss": 0.3937,
      "step": 55000
    },
    {
      "epoch": 2.73,
      "learning_rate": 4.525342903495403e-06,
      "loss": 0.3955,
      "step": 55500
    },
    {
      "epoch": 2.75,
      "learning_rate": 4.115661308031398e-06,
      "loss": 0.4004,
      "step": 56000
    },
    {
      "epoch": 2.78,
      "learning_rate": 3.705979712567393e-06,
      "loss": 0.4024,
      "step": 56500
    },
    {
      "epoch": 2.8,
      "learning_rate": 3.296298117103387e-06,
      "loss": 0.3976,
      "step": 57000
    },
    {
      "epoch": 2.83,
      "learning_rate": 2.886616521639382e-06,
      "loss": 0.3932,
      "step": 57500
    },
    {
      "epoch": 2.85,
      "learning_rate": 2.4769349261753767e-06,
      "loss": 0.3955,
      "step": 58000
    },
    {
      "epoch": 2.88,
      "learning_rate": 2.0672533307113713e-06,
      "loss": 0.4007,
      "step": 58500
    },
    {
      "epoch": 2.9,
      "learning_rate": 1.6575717352473659e-06,
      "loss": 0.4005,
      "step": 59000
    },
    {
      "epoch": 2.93,
      "learning_rate": 1.2478901397833605e-06,
      "loss": 0.3935,
      "step": 59500
    },
    {
      "epoch": 2.95,
      "learning_rate": 8.382085443193551e-07,
      "loss": 0.3957,
      "step": 60000
    },
    {
      "epoch": 2.97,
      "learning_rate": 4.285269488553496e-07,
      "loss": 0.3972,
      "step": 60500
    },
    {
      "epoch": 3.0,
      "learning_rate": 1.8845353391344248e-08,
      "loss": 0.3934,
      "step": 61000
    },
    {
      "epoch": 3.0,
      "step": 61023,
      "total_flos": 3.740919470993169e+17,
      "train_loss": 0.46753893842028726,
      "train_runtime": 20527.2727,
      "train_samples_per_second": 95.125,
      "train_steps_per_second": 2.973
    }
  ],
  "max_steps": 61023,
  "num_train_epochs": 3,
  "total_flos": 3.740919470993169e+17,
  "trial_name": null,
  "trial_params": null
}