{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 6.999625795185231,
  "eval_steps": 500,
  "global_step": 7014,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.02,
      "learning_rate": 4.9731096052490054e-05,
      "loss": 1.7709,
      "step": 100
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9462192104980105e-05,
      "loss": 1.6756,
      "step": 200
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.919328815747015e-05,
      "loss": 1.6585,
      "step": 300
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.892438420996021e-05,
      "loss": 1.6405,
      "step": 400
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.865548026245025e-05,
      "loss": 1.5871,
      "step": 500
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.838657631494031e-05,
      "loss": 1.5809,
      "step": 600
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8117672367430355e-05,
      "loss": 1.5877,
      "step": 700
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.7848768419920406e-05,
      "loss": 1.5475,
      "step": 800
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.757986447241046e-05,
      "loss": 1.5461,
      "step": 900
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.731096052490051e-05,
      "loss": 1.5553,
      "step": 1000
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.704205657739056e-05,
      "loss": 1.543,
      "step": 1100
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.6773152629880605e-05,
      "loss": 1.5603,
      "step": 1200
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.650424868237066e-05,
      "loss": 1.5441,
      "step": 1300
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.623534473486071e-05,
      "loss": 1.5416,
      "step": 1400
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.596644078735076e-05,
      "loss": 1.5159,
      "step": 1500
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.569753683984081e-05,
      "loss": 1.5225,
      "step": 1600
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.542863289233086e-05,
      "loss": 1.5281,
      "step": 1700
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.515972894482091e-05,
      "loss": 1.5128,
      "step": 1800
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.4890824997310964e-05,
      "loss": 1.5215,
      "step": 1900
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.4621921049801015e-05,
      "loss": 1.4975,
      "step": 2000
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.435301710229106e-05,
      "loss": 1.4772,
      "step": 2100
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.408411315478112e-05,
      "loss": 1.4776,
      "step": 2200
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.381520920727116e-05,
      "loss": 1.5157,
      "step": 2300
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.3546305259761214e-05,
      "loss": 1.4968,
      "step": 2400
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.3277401312251265e-05,
      "loss": 1.4835,
      "step": 2500
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.3008497364741317e-05,
      "loss": 1.4805,
      "step": 2600
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.273959341723137e-05,
      "loss": 1.4892,
      "step": 2700
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.247068946972142e-05,
      "loss": 1.4705,
      "step": 2800
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.220178552221147e-05,
      "loss": 1.4829,
      "step": 2900
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.193288157470152e-05,
      "loss": 1.4688,
      "step": 3000
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.1663977627191566e-05,
      "loss": 1.4711,
      "step": 3100
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.139507367968162e-05,
      "loss": 1.4813,
      "step": 3200
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.112616973217167e-05,
      "loss": 1.4541,
      "step": 3300
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.085726578466172e-05,
      "loss": 1.4574,
      "step": 3400
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.058836183715177e-05,
      "loss": 1.4634,
      "step": 3500
    },
    {
      "epoch": 3.59,
      "learning_rate": 1.407185628742515e-05,
      "loss": 1.2902,
      "step": 3600
    },
    {
      "epoch": 3.69,
      "learning_rate": 1.3073852295409184e-05,
      "loss": 1.2773,
      "step": 3700
    },
    {
      "epoch": 3.79,
      "learning_rate": 1.2075848303393214e-05,
      "loss": 1.2647,
      "step": 3800
    },
    {
      "epoch": 3.89,
      "learning_rate": 1.1077844311377246e-05,
      "loss": 1.2524,
      "step": 3900
    },
    {
      "epoch": 3.99,
      "learning_rate": 1.0079840319361278e-05,
      "loss": 1.2543,
      "step": 4000
    },
    {
      "epoch": 4.09,
      "learning_rate": 9.08183632734531e-06,
      "loss": 1.2499,
      "step": 4100
    },
    {
      "epoch": 4.19,
      "learning_rate": 8.083832335329342e-06,
      "loss": 1.2418,
      "step": 4200
    },
    {
      "epoch": 4.29,
      "learning_rate": 7.085828343313373e-06,
      "loss": 1.2379,
      "step": 4300
    },
    {
      "epoch": 4.39,
      "learning_rate": 6.0878243512974054e-06,
      "loss": 1.2325,
      "step": 4400
    },
    {
      "epoch": 4.49,
      "learning_rate": 5.0898203592814375e-06,
      "loss": 1.2246,
      "step": 4500
    },
    {
      "epoch": 4.59,
      "learning_rate": 4.091816367265469e-06,
      "loss": 1.2331,
      "step": 4600
    },
    {
      "epoch": 4.69,
      "learning_rate": 3.093812375249501e-06,
      "loss": 1.2236,
      "step": 4700
    },
    {
      "epoch": 4.79,
      "learning_rate": 2.095808383233533e-06,
      "loss": 1.2281,
      "step": 4800
    },
    {
      "epoch": 4.89,
      "learning_rate": 1.097804391217565e-06,
      "loss": 1.2254,
      "step": 4900
    },
    {
      "epoch": 4.99,
      "learning_rate": 9.98003992015968e-08,
      "loss": 1.225,
      "step": 5000
    },
    {
      "epoch": 5.09,
      "learning_rate": 1.364414029084688e-05,
      "loss": 1.2276,
      "step": 5100
    },
    {
      "epoch": 5.19,
      "learning_rate": 1.2931280296549758e-05,
      "loss": 1.2064,
      "step": 5200
    },
    {
      "epoch": 5.29,
      "learning_rate": 1.2218420302252638e-05,
      "loss": 1.2102,
      "step": 5300
    },
    {
      "epoch": 5.39,
      "learning_rate": 1.1505560307955517e-05,
      "loss": 1.2095,
      "step": 5400
    },
    {
      "epoch": 5.49,
      "learning_rate": 1.0792700313658398e-05,
      "loss": 1.2239,
      "step": 5500
    },
    {
      "epoch": 5.59,
      "learning_rate": 1.0079840319361278e-05,
      "loss": 1.216,
      "step": 5600
    },
    {
      "epoch": 5.69,
      "learning_rate": 9.366980325064158e-06,
      "loss": 1.2206,
      "step": 5700
    },
    {
      "epoch": 5.79,
      "learning_rate": 8.654120330767037e-06,
      "loss": 1.2133,
      "step": 5800
    },
    {
      "epoch": 5.89,
      "learning_rate": 7.941260336469917e-06,
      "loss": 1.2059,
      "step": 5900
    },
    {
      "epoch": 5.99,
      "learning_rate": 7.228400342172797e-06,
      "loss": 1.206,
      "step": 6000
    },
    {
      "epoch": 6.09,
      "learning_rate": 6.5155403478756775e-06,
      "loss": 1.1976,
      "step": 6100
    },
    {
      "epoch": 6.19,
      "learning_rate": 5.802680353578557e-06,
      "loss": 1.1966,
      "step": 6200
    },
    {
      "epoch": 6.29,
      "learning_rate": 5.0898203592814375e-06,
      "loss": 1.1992,
      "step": 6300
    },
    {
      "epoch": 6.39,
      "learning_rate": 4.376960364984317e-06,
      "loss": 1.1946,
      "step": 6400
    },
    {
      "epoch": 6.49,
      "learning_rate": 3.664100370687197e-06,
      "loss": 1.1879,
      "step": 6500
    },
    {
      "epoch": 6.59,
      "learning_rate": 2.951240376390077e-06,
      "loss": 1.1863,
      "step": 6600
    },
    {
      "epoch": 6.69,
      "learning_rate": 2.238380382092957e-06,
      "loss": 1.182,
      "step": 6700
    },
    {
      "epoch": 6.79,
      "learning_rate": 1.525520387795837e-06,
      "loss": 1.1914,
      "step": 6800
    },
    {
      "epoch": 6.89,
      "learning_rate": 8.12660393498717e-07,
      "loss": 1.1892,
      "step": 6900
    },
    {
      "epoch": 6.99,
      "learning_rate": 9.98003992015968e-08,
      "loss": 1.1943,
      "step": 7000
    },
    {
      "epoch": 7.0,
      "step": 7014,
      "total_flos": 1.1685618503900851e+17,
      "train_loss": 0.34533324811340976,
      "train_runtime": 2399.0328,
      "train_samples_per_second": 374.277,
      "train_steps_per_second": 2.924
    }
  ],
  "logging_steps": 100,
  "max_steps": 7014,
  "num_train_epochs": 7,
  "save_steps": 500,
  "total_flos": 1.1685618503900851e+17,
  "trial_name": null,
  "trial_params": null
}