{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.0,
  "eval_steps": 500,
  "global_step": 47275,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.05,
      "learning_rate": 4.9471179270227394e-05,
      "loss": 2.856,
      "step": 500
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8942358540454786e-05,
      "loss": 2.5599,
      "step": 1000
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.841353781068218e-05,
      "loss": 2.4166,
      "step": 1500
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.7884717080909575e-05,
      "loss": 2.3151,
      "step": 2000
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.735589635113697e-05,
      "loss": 2.2732,
      "step": 2500
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.682707562136436e-05,
      "loss": 2.2248,
      "step": 3000
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.629825489159175e-05,
      "loss": 2.1665,
      "step": 3500
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.576943416181915e-05,
      "loss": 2.1445,
      "step": 4000
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.524061343204653e-05,
      "loss": 2.1219,
      "step": 4500
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.471179270227393e-05,
      "loss": 2.0906,
      "step": 5000
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.418297197250132e-05,
      "loss": 2.0674,
      "step": 5500
    },
    {
      "epoch": 0.63,
      "learning_rate": 4.365415124272872e-05,
      "loss": 2.0508,
      "step": 6000
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.3125330512956106e-05,
      "loss": 2.0176,
      "step": 6500
    },
    {
      "epoch": 0.74,
      "learning_rate": 4.2596509783183505e-05,
      "loss": 1.9954,
      "step": 7000
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.2067689053410896e-05,
      "loss": 1.9745,
      "step": 7500
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.153886832363829e-05,
      "loss": 1.9776,
      "step": 8000
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.1011105235325226e-05,
      "loss": 1.9579,
      "step": 8500
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.048228450555262e-05,
      "loss": 1.9384,
      "step": 9000
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.9953463775780016e-05,
      "loss": 1.914,
      "step": 9500
    },
    {
      "epoch": 1.06,
      "learning_rate": 3.94246430460074e-05,
      "loss": 1.7926,
      "step": 10000
    },
    {
      "epoch": 1.11,
      "learning_rate": 3.8896879957694345e-05,
      "loss": 1.7862,
      "step": 10500
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.8369116869381283e-05,
      "loss": 1.7798,
      "step": 11000
    },
    {
      "epoch": 1.22,
      "learning_rate": 3.7841353781068215e-05,
      "loss": 1.7871,
      "step": 11500
    },
    {
      "epoch": 1.27,
      "learning_rate": 3.731253305129561e-05,
      "loss": 1.7872,
      "step": 12000
    },
    {
      "epoch": 1.32,
      "learning_rate": 3.6783712321523005e-05,
      "loss": 1.7669,
      "step": 12500
    },
    {
      "epoch": 1.37,
      "learning_rate": 3.62548915917504e-05,
      "loss": 1.7631,
      "step": 13000
    },
    {
      "epoch": 1.43,
      "learning_rate": 3.572607086197779e-05,
      "loss": 1.7672,
      "step": 13500
    },
    {
      "epoch": 1.48,
      "learning_rate": 3.5197250132205186e-05,
      "loss": 1.7741,
      "step": 14000
    },
    {
      "epoch": 1.53,
      "learning_rate": 3.466842940243258e-05,
      "loss": 1.7571,
      "step": 14500
    },
    {
      "epoch": 1.59,
      "learning_rate": 3.413960867265997e-05,
      "loss": 1.7444,
      "step": 15000
    },
    {
      "epoch": 1.64,
      "learning_rate": 3.361078794288736e-05,
      "loss": 1.742,
      "step": 15500
    },
    {
      "epoch": 1.69,
      "learning_rate": 3.30830248545743e-05,
      "loss": 1.734,
      "step": 16000
    },
    {
      "epoch": 1.75,
      "learning_rate": 3.25542041248017e-05,
      "loss": 1.7277,
      "step": 16500
    },
    {
      "epoch": 1.8,
      "learning_rate": 3.202538339502908e-05,
      "loss": 1.7379,
      "step": 17000
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.149656266525648e-05,
      "loss": 1.7307,
      "step": 17500
    },
    {
      "epoch": 1.9,
      "learning_rate": 3.096879957694342e-05,
      "loss": 1.7145,
      "step": 18000
    },
    {
      "epoch": 1.96,
      "learning_rate": 3.0439978847170807e-05,
      "loss": 1.7195,
      "step": 18500
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.99111581173982e-05,
      "loss": 1.6885,
      "step": 19000
    },
    {
      "epoch": 2.06,
      "learning_rate": 2.9382337387625597e-05,
      "loss": 1.5763,
      "step": 19500
    },
    {
      "epoch": 2.12,
      "learning_rate": 2.885351665785299e-05,
      "loss": 1.5877,
      "step": 20000
    },
    {
      "epoch": 2.17,
      "learning_rate": 2.832469592808038e-05,
      "loss": 1.5775,
      "step": 20500
    },
    {
      "epoch": 2.22,
      "learning_rate": 2.7795875198307775e-05,
      "loss": 1.584,
      "step": 21000
    },
    {
      "epoch": 2.27,
      "learning_rate": 2.726705446853517e-05,
      "loss": 1.5868,
      "step": 21500
    },
    {
      "epoch": 2.33,
      "learning_rate": 2.6738233738762558e-05,
      "loss": 1.583,
      "step": 22000
    },
    {
      "epoch": 2.38,
      "learning_rate": 2.6209413008989953e-05,
      "loss": 1.5835,
      "step": 22500
    },
    {
      "epoch": 2.43,
      "learning_rate": 2.568164992067689e-05,
      "loss": 1.5769,
      "step": 23000
    },
    {
      "epoch": 2.49,
      "learning_rate": 2.5152829190904286e-05,
      "loss": 1.565,
      "step": 23500
    },
    {
      "epoch": 2.54,
      "learning_rate": 2.4624008461131677e-05,
      "loss": 1.5741,
      "step": 24000
    },
    {
      "epoch": 2.59,
      "learning_rate": 2.409518773135907e-05,
      "loss": 1.5794,
      "step": 24500
    },
    {
      "epoch": 2.64,
      "learning_rate": 2.3566367001586464e-05,
      "loss": 1.548,
      "step": 25000
    },
    {
      "epoch": 2.7,
      "learning_rate": 2.3037546271813855e-05,
      "loss": 1.5791,
      "step": 25500
    },
    {
      "epoch": 2.75,
      "learning_rate": 2.2509783183500794e-05,
      "loss": 1.5628,
      "step": 26000
    },
    {
      "epoch": 2.8,
      "learning_rate": 2.198096245372819e-05,
      "loss": 1.5574,
      "step": 26500
    },
    {
      "epoch": 2.86,
      "learning_rate": 2.145214172395558e-05,
      "loss": 1.5635,
      "step": 27000
    },
    {
      "epoch": 2.91,
      "learning_rate": 2.0923320994182975e-05,
      "loss": 1.5682,
      "step": 27500
    },
    {
      "epoch": 2.96,
      "learning_rate": 2.0394500264410367e-05,
      "loss": 1.5423,
      "step": 28000
    },
    {
      "epoch": 3.01,
      "learning_rate": 1.986567953463776e-05,
      "loss": 1.5218,
      "step": 28500
    },
    {
      "epoch": 3.07,
      "learning_rate": 1.9336858804865153e-05,
      "loss": 1.4528,
      "step": 29000
    },
    {
      "epoch": 3.12,
      "learning_rate": 1.8808038075092545e-05,
      "loss": 1.4557,
      "step": 29500
    },
    {
      "epoch": 3.17,
      "learning_rate": 1.8280274986779483e-05,
      "loss": 1.4418,
      "step": 30000
    },
    {
      "epoch": 3.23,
      "learning_rate": 1.775251189846642e-05,
      "loss": 1.4528,
      "step": 30500
    },
    {
      "epoch": 3.28,
      "learning_rate": 1.7223691168693816e-05,
      "loss": 1.4406,
      "step": 31000
    },
    {
      "epoch": 3.33,
      "learning_rate": 1.6694870438921207e-05,
      "loss": 1.4369,
      "step": 31500
    },
    {
      "epoch": 3.38,
      "learning_rate": 1.6166049709148602e-05,
      "loss": 1.4498,
      "step": 32000
    },
    {
      "epoch": 3.44,
      "learning_rate": 1.5637228979375994e-05,
      "loss": 1.4466,
      "step": 32500
    },
    {
      "epoch": 3.49,
      "learning_rate": 1.5109465891062929e-05,
      "loss": 1.4383,
      "step": 33000
    },
    {
      "epoch": 3.54,
      "learning_rate": 1.4580645161290324e-05,
      "loss": 1.4497,
      "step": 33500
    },
    {
      "epoch": 3.6,
      "learning_rate": 1.4051824431517715e-05,
      "loss": 1.4426,
      "step": 34000
    },
    {
      "epoch": 3.65,
      "learning_rate": 1.352300370174511e-05,
      "loss": 1.4311,
      "step": 34500
    },
    {
      "epoch": 3.7,
      "learning_rate": 1.2994182971972502e-05,
      "loss": 1.4384,
      "step": 35000
    },
    {
      "epoch": 3.75,
      "learning_rate": 1.2465362242199895e-05,
      "loss": 1.4399,
      "step": 35500
    },
    {
      "epoch": 3.81,
      "learning_rate": 1.1936541512427288e-05,
      "loss": 1.4395,
      "step": 36000
    },
    {
      "epoch": 3.86,
      "learning_rate": 1.1407720782654681e-05,
      "loss": 1.428,
      "step": 36500
    },
    {
      "epoch": 3.91,
      "learning_rate": 1.0878900052882075e-05,
      "loss": 1.4439,
      "step": 37000
    },
    {
      "epoch": 3.97,
      "learning_rate": 1.0351136964569011e-05,
      "loss": 1.4361,
      "step": 37500
    },
    {
      "epoch": 4.02,
      "learning_rate": 9.822316234796404e-06,
      "loss": 1.4036,
      "step": 38000
    },
    {
      "epoch": 4.07,
      "learning_rate": 9.293495505023798e-06,
      "loss": 1.3662,
      "step": 38500
    },
    {
      "epoch": 4.12,
      "learning_rate": 8.76467477525119e-06,
      "loss": 1.3538,
      "step": 39000
    },
    {
      "epoch": 4.18,
      "learning_rate": 8.236911686938129e-06,
      "loss": 1.3632,
      "step": 39500
    },
    {
      "epoch": 4.23,
      "learning_rate": 7.708090957165522e-06,
      "loss": 1.3554,
      "step": 40000
    },
    {
      "epoch": 4.28,
      "learning_rate": 7.1792702273929146e-06,
      "loss": 1.3577,
      "step": 40500
    },
    {
      "epoch": 4.34,
      "learning_rate": 6.650449497620308e-06,
      "loss": 1.3537,
      "step": 41000
    },
    {
      "epoch": 4.39,
      "learning_rate": 6.1216287678477e-06,
      "loss": 1.3581,
      "step": 41500
    },
    {
      "epoch": 4.44,
      "learning_rate": 5.5928080380750935e-06,
      "loss": 1.3637,
      "step": 42000
    },
    {
      "epoch": 4.49,
      "learning_rate": 5.063987308302486e-06,
      "loss": 1.3432,
      "step": 42500
    },
    {
      "epoch": 4.55,
      "learning_rate": 4.535166578529878e-06,
      "loss": 1.3463,
      "step": 43000
    },
    {
      "epoch": 4.6,
      "learning_rate": 4.007403490216817e-06,
      "loss": 1.3481,
      "step": 43500
    },
    {
      "epoch": 4.65,
      "learning_rate": 3.4785827604442092e-06,
      "loss": 1.353,
      "step": 44000
    },
    {
      "epoch": 4.71,
      "learning_rate": 2.9508196721311478e-06,
      "loss": 1.3505,
      "step": 44500
    },
    {
      "epoch": 4.76,
      "learning_rate": 2.4219989423585406e-06,
      "loss": 1.3404,
      "step": 45000
    },
    {
      "epoch": 4.81,
      "learning_rate": 1.8931782125859334e-06,
      "loss": 1.3574,
      "step": 45500
    },
    {
      "epoch": 4.87,
      "learning_rate": 1.3643574828133265e-06,
      "loss": 1.3532,
      "step": 46000
    },
    {
      "epoch": 4.92,
      "learning_rate": 8.355367530407193e-07,
      "loss": 1.3357,
      "step": 46500
    },
    {
      "epoch": 4.97,
      "learning_rate": 3.077736647276573e-07,
      "loss": 1.3374,
      "step": 47000
    },
    {
      "epoch": 5.0,
      "step": 47275,
      "total_flos": 7.799978327959142e+17,
      "train_loss": 1.65651271519492,
      "train_runtime": 12815.2129,
      "train_samples_per_second": 236.094,
      "train_steps_per_second": 3.689
    }
  ],
  "logging_steps": 500,
  "max_steps": 47275,
  "num_train_epochs": 5,
  "save_steps": 10000000,
  "total_flos": 7.799978327959142e+17,
  "trial_name": null,
  "trial_params": null
}