{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "global_step": 23646,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.06343567622430855,
      "learning_rate": 4.8942738729594864e-05,
      "loss": 2.408385009765625,
      "step": 500
    },
    {
      "epoch": 0.1268713524486171,
      "learning_rate": 4.788547745918972e-05,
      "loss": 2.0042470703125,
      "step": 1000
    },
    {
      "epoch": 0.19030702867292565,
      "learning_rate": 4.682821618878457e-05,
      "loss": 1.8769989013671875,
      "step": 1500
    },
    {
      "epoch": 0.2537427048972342,
      "learning_rate": 4.5770954918379434e-05,
      "loss": 1.820308837890625,
      "step": 2000
    },
    {
      "epoch": 0.31717838112154273,
      "learning_rate": 4.4713693647974295e-05,
      "loss": 1.7414315185546876,
      "step": 2500
    },
    {
      "epoch": 0.3806140573458513,
      "learning_rate": 4.365643237756914e-05,
      "loss": 1.6673572998046875,
      "step": 3000
    },
    {
      "epoch": 0.44404973357015987,
      "learning_rate": 4.2599171107164004e-05,
      "loss": 1.5917396240234376,
      "step": 3500
    },
    {
      "epoch": 0.5074854097944684,
      "learning_rate": 4.1541909836758865e-05,
      "loss": 1.4989898681640625,
      "step": 4000
    },
    {
      "epoch": 0.570921086018777,
      "learning_rate": 4.048464856635372e-05,
      "loss": 1.4598778076171874,
      "step": 4500
    },
    {
      "epoch": 0.6343567622430855,
      "learning_rate": 3.9427387295948574e-05,
      "loss": 1.3672525634765624,
      "step": 5000
    },
    {
      "epoch": 0.697792438467394,
      "learning_rate": 3.8370126025543435e-05,
      "loss": 1.3376380615234376,
      "step": 5500
    },
    {
      "epoch": 0.7612281146917026,
      "learning_rate": 3.73128647551383e-05,
      "loss": 1.280780029296875,
      "step": 6000
    },
    {
      "epoch": 0.8246637909160112,
      "learning_rate": 3.6255603484733144e-05,
      "loss": 1.2319114990234374,
      "step": 6500
    },
    {
      "epoch": 0.8880994671403197,
      "learning_rate": 3.5198342214328006e-05,
      "loss": 1.1800660400390626,
      "step": 7000
    },
    {
      "epoch": 0.9515351433646283,
      "learning_rate": 3.414108094392287e-05,
      "loss": 1.12030810546875,
      "step": 7500
    },
    {
      "epoch": 1.0149708195889369,
      "learning_rate": 3.308381967351772e-05,
      "loss": 1.09913232421875,
      "step": 8000
    },
    {
      "epoch": 1.0784064958132453,
      "learning_rate": 3.2026558403112576e-05,
      "loss": 1.0072744750976563,
      "step": 8500
    },
    {
      "epoch": 1.141842172037554,
      "learning_rate": 3.096929713270744e-05,
      "loss": 1.0016256713867187,
      "step": 9000
    },
    {
      "epoch": 1.2052778482618625,
      "learning_rate": 2.9912035862302295e-05,
      "loss": 0.9350516357421875,
      "step": 9500
    },
    {
      "epoch": 1.268713524486171,
      "learning_rate": 2.885477459189715e-05,
      "loss": 0.9032111206054687,
      "step": 10000
    },
    {
      "epoch": 1.3321492007104796,
      "learning_rate": 2.7797513321492007e-05,
      "loss": 0.8644053344726562,
      "step": 10500
    },
    {
      "epoch": 1.395584876934788,
      "learning_rate": 2.674025205108687e-05,
      "loss": 0.8429204711914062,
      "step": 11000
    },
    {
      "epoch": 1.4590205531590967,
      "learning_rate": 2.5682990780681726e-05,
      "loss": 0.8093466186523437,
      "step": 11500
    },
    {
      "epoch": 1.5224562293834052,
      "learning_rate": 2.462572951027658e-05,
      "loss": 0.7872549438476563,
      "step": 12000
    },
    {
      "epoch": 1.5858919056077139,
      "learning_rate": 2.356846823987144e-05,
      "loss": 0.7391060180664063,
      "step": 12500
    },
    {
      "epoch": 1.6493275818320223,
      "learning_rate": 2.2511206969466293e-05,
      "loss": 0.72500390625,
      "step": 13000
    },
    {
      "epoch": 1.7127632580563308,
      "learning_rate": 2.1453945699061154e-05,
      "loss": 0.704537841796875,
      "step": 13500
    },
    {
      "epoch": 1.7761989342806395,
      "learning_rate": 2.039668442865601e-05,
      "loss": 0.675576171875,
      "step": 14000
    },
    {
      "epoch": 1.8396346105049481,
      "learning_rate": 1.933942315825087e-05,
      "loss": 0.65029541015625,
      "step": 14500
    },
    {
      "epoch": 1.9030702867292564,
      "learning_rate": 1.8282161887845725e-05,
      "loss": 0.6266005859375,
      "step": 15000
    },
    {
      "epoch": 1.966505962953565,
      "learning_rate": 1.7224900617440583e-05,
      "loss": 0.6266526489257812,
      "step": 15500
    },
    {
      "epoch": 2.0299416391778737,
      "learning_rate": 1.616763934703544e-05,
      "loss": 0.60100439453125,
      "step": 16000
    },
    {
      "epoch": 2.093377315402182,
      "learning_rate": 1.5110378076630297e-05,
      "loss": 0.5694144287109375,
      "step": 16500
    },
    {
      "epoch": 2.1568129916264906,
      "learning_rate": 1.4053116806225156e-05,
      "loss": 0.5693157958984375,
      "step": 17000
    },
    {
      "epoch": 2.2202486678507993,
      "learning_rate": 1.2995855535820012e-05,
      "loss": 0.5522522583007813,
      "step": 17500
    },
    {
      "epoch": 2.283684344075108,
      "learning_rate": 1.193859426541487e-05,
      "loss": 0.5293363647460938,
      "step": 18000
    },
    {
      "epoch": 2.3471200202994162,
      "learning_rate": 1.0881332995009726e-05,
      "loss": 0.52986669921875,
      "step": 18500
    },
    {
      "epoch": 2.410555696523725,
      "learning_rate": 9.824071724604584e-06,
      "loss": 0.5213316040039062,
      "step": 19000
    },
    {
      "epoch": 2.4739913727480336,
      "learning_rate": 8.766810454199442e-06,
      "loss": 0.5159844970703125,
      "step": 19500
    },
    {
      "epoch": 2.537427048972342,
      "learning_rate": 7.7095491837943e-06,
      "loss": 0.508293701171875,
      "step": 20000
    },
    {
      "epoch": 2.6008627251966505,
      "learning_rate": 6.652287913389158e-06,
      "loss": 0.4958880615234375,
      "step": 20500
    },
    {
      "epoch": 2.664298401420959,
      "learning_rate": 5.595026642984015e-06,
      "loss": 0.48646127319335936,
      "step": 21000
    },
    {
      "epoch": 2.727734077645268,
      "learning_rate": 4.537765372578872e-06,
      "loss": 0.4905679626464844,
      "step": 21500
    },
    {
      "epoch": 2.791169753869576,
      "learning_rate": 3.4805041021737297e-06,
      "loss": 0.4853497314453125,
      "step": 22000
    },
    {
      "epoch": 2.854605430093885,
      "learning_rate": 2.4232428317685867e-06,
      "loss": 0.48792453002929687,
      "step": 22500
    },
    {
      "epoch": 2.9180411063181935,
      "learning_rate": 1.3659815613634441e-06,
      "loss": 0.48739105224609375,
      "step": 23000
    },
    {
      "epoch": 2.9814767825425017,
      "learning_rate": 3.087202909583016e-07,
      "loss": 0.4768514404296875,
      "step": 23500
    },
    {
      "epoch": 3.0,
      "step": 23646,
      "train_runtime": 4566.2664,
      "train_samples_per_second": 5.178
    }
  ],
  "max_steps": 23646,
  "num_train_epochs": 3,
  "total_flos": 63460156661015040,
  "trial_name": null,
  "trial_params": null
}