{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 20.0,
  "global_step": 40080,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.25,
      "learning_rate": 9.875249500998003e-05,
      "loss": 7.8671,
      "step": 500
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.750499001996009e-05,
      "loss": 7.0941,
      "step": 1000
    },
    {
      "epoch": 0.75,
      "learning_rate": 9.625748502994012e-05,
      "loss": 6.7944,
      "step": 1500
    },
    {
      "epoch": 1.0,
      "learning_rate": 9.500998003992016e-05,
      "loss": 6.526,
      "step": 2000
    },
    {
      "epoch": 1.25,
      "learning_rate": 9.37624750499002e-05,
      "loss": 6.2583,
      "step": 2500
    },
    {
      "epoch": 1.5,
      "learning_rate": 9.251497005988024e-05,
      "loss": 6.0239,
      "step": 3000
    },
    {
      "epoch": 1.75,
      "learning_rate": 9.126746506986029e-05,
      "loss": 5.7969,
      "step": 3500
    },
    {
      "epoch": 2.0,
      "learning_rate": 9.001996007984033e-05,
      "loss": 5.5919,
      "step": 4000
    },
    {
      "epoch": 2.25,
      "learning_rate": 8.877245508982036e-05,
      "loss": 5.3796,
      "step": 4500
    },
    {
      "epoch": 2.5,
      "learning_rate": 8.752495009980041e-05,
      "loss": 5.1887,
      "step": 5000
    },
    {
      "epoch": 2.74,
      "learning_rate": 8.627744510978044e-05,
      "loss": 5.0026,
      "step": 5500
    },
    {
      "epoch": 2.99,
      "learning_rate": 8.502994011976048e-05,
      "loss": 4.843,
      "step": 6000
    },
    {
      "epoch": 3.24,
      "learning_rate": 8.378243512974053e-05,
      "loss": 4.6736,
      "step": 6500
    },
    {
      "epoch": 3.49,
      "learning_rate": 8.253493013972057e-05,
      "loss": 4.553,
      "step": 7000
    },
    {
      "epoch": 3.74,
      "learning_rate": 8.12874251497006e-05,
      "loss": 4.4224,
      "step": 7500
    },
    {
      "epoch": 3.99,
      "learning_rate": 8.003992015968065e-05,
      "loss": 4.3097,
      "step": 8000
    },
    {
      "epoch": 4.24,
      "learning_rate": 7.879241516966068e-05,
      "loss": 4.1975,
      "step": 8500
    },
    {
      "epoch": 4.49,
      "learning_rate": 7.754491017964072e-05,
      "loss": 4.0994,
      "step": 9000
    },
    {
      "epoch": 4.74,
      "learning_rate": 7.629740518962077e-05,
      "loss": 4.0258,
      "step": 9500
    },
    {
      "epoch": 4.99,
      "learning_rate": 7.50499001996008e-05,
      "loss": 3.9443,
      "step": 10000
    },
    {
      "epoch": 5.24,
      "learning_rate": 7.380239520958085e-05,
      "loss": 3.8616,
      "step": 10500
    },
    {
      "epoch": 5.49,
      "learning_rate": 7.255489021956088e-05,
      "loss": 3.8086,
      "step": 11000
    },
    {
      "epoch": 5.74,
      "learning_rate": 7.130738522954092e-05,
      "loss": 3.7464,
      "step": 11500
    },
    {
      "epoch": 5.99,
      "learning_rate": 7.005988023952096e-05,
      "loss": 3.6925,
      "step": 12000
    },
    {
      "epoch": 6.24,
      "learning_rate": 6.8812375249501e-05,
      "loss": 3.6216,
      "step": 12500
    },
    {
      "epoch": 6.49,
      "learning_rate": 6.756487025948103e-05,
      "loss": 3.585,
      "step": 13000
    },
    {
      "epoch": 6.74,
      "learning_rate": 6.631736526946109e-05,
      "loss": 3.5383,
      "step": 13500
    },
    {
      "epoch": 6.99,
      "learning_rate": 6.506986027944112e-05,
      "loss": 3.4925,
      "step": 14000
    },
    {
      "epoch": 7.24,
      "learning_rate": 6.382235528942116e-05,
      "loss": 3.4524,
      "step": 14500
    },
    {
      "epoch": 7.49,
      "learning_rate": 6.25748502994012e-05,
      "loss": 3.3985,
      "step": 15000
    },
    {
      "epoch": 7.73,
      "learning_rate": 6.132734530938125e-05,
      "loss": 3.3684,
      "step": 15500
    },
    {
      "epoch": 7.98,
      "learning_rate": 6.007984031936128e-05,
      "loss": 3.3273,
      "step": 16000
    },
    {
      "epoch": 8.23,
      "learning_rate": 5.8832335329341316e-05,
      "loss": 3.2898,
      "step": 16500
    },
    {
      "epoch": 8.48,
      "learning_rate": 5.758483033932136e-05,
      "loss": 3.2554,
      "step": 17000
    },
    {
      "epoch": 8.73,
      "learning_rate": 5.6337325349301393e-05,
      "loss": 3.2215,
      "step": 17500
    },
    {
      "epoch": 8.98,
      "learning_rate": 5.508982035928144e-05,
      "loss": 3.2006,
      "step": 18000
    },
    {
      "epoch": 9.23,
      "learning_rate": 5.384231536926148e-05,
      "loss": 3.1627,
      "step": 18500
    },
    {
      "epoch": 9.48,
      "learning_rate": 5.259481037924152e-05,
      "loss": 3.1391,
      "step": 19000
    },
    {
      "epoch": 9.73,
      "learning_rate": 5.1347305389221555e-05,
      "loss": 3.1167,
      "step": 19500
    },
    {
      "epoch": 9.98,
      "learning_rate": 5.0099800399201604e-05,
      "loss": 3.108,
      "step": 20000
    },
    {
      "epoch": 10.23,
      "learning_rate": 4.885229540918164e-05,
      "loss": 3.0782,
      "step": 20500
    },
    {
      "epoch": 10.48,
      "learning_rate": 4.7604790419161675e-05,
      "loss": 3.054,
      "step": 21000
    },
    {
      "epoch": 10.73,
      "learning_rate": 4.635728542914172e-05,
      "loss": 3.0302,
      "step": 21500
    },
    {
      "epoch": 10.98,
      "learning_rate": 4.510978043912176e-05,
      "loss": 3.0129,
      "step": 22000
    },
    {
      "epoch": 11.23,
      "learning_rate": 4.3862275449101795e-05,
      "loss": 2.9978,
      "step": 22500
    },
    {
      "epoch": 11.48,
      "learning_rate": 4.261477045908184e-05,
      "loss": 2.9658,
      "step": 23000
    },
    {
      "epoch": 11.73,
      "learning_rate": 4.136726546906188e-05,
      "loss": 2.9536,
      "step": 23500
    },
    {
      "epoch": 11.98,
      "learning_rate": 4.0119760479041915e-05,
      "loss": 2.937,
      "step": 24000
    },
    {
      "epoch": 12.23,
      "learning_rate": 3.887225548902196e-05,
      "loss": 2.9144,
      "step": 24500
    },
    {
      "epoch": 12.48,
      "learning_rate": 3.7624750499002e-05,
      "loss": 2.9045,
      "step": 25000
    },
    {
      "epoch": 12.72,
      "learning_rate": 3.637724550898204e-05,
      "loss": 2.8814,
      "step": 25500
    },
    {
      "epoch": 12.97,
      "learning_rate": 3.512974051896208e-05,
      "loss": 2.8749,
      "step": 26000
    },
    {
      "epoch": 13.22,
      "learning_rate": 3.388223552894212e-05,
      "loss": 2.8333,
      "step": 26500
    },
    {
      "epoch": 13.47,
      "learning_rate": 3.263473053892216e-05,
      "loss": 2.8402,
      "step": 27000
    },
    {
      "epoch": 13.72,
      "learning_rate": 3.13872255489022e-05,
      "loss": 2.8363,
      "step": 27500
    },
    {
      "epoch": 13.97,
      "learning_rate": 3.013972055888224e-05,
      "loss": 2.8235,
      "step": 28000
    },
    {
      "epoch": 14.22,
      "learning_rate": 2.8892215568862274e-05,
      "loss": 2.8038,
      "step": 28500
    },
    {
      "epoch": 14.47,
      "learning_rate": 2.7644710578842313e-05,
      "loss": 2.7948,
      "step": 29000
    },
    {
      "epoch": 14.72,
      "learning_rate": 2.6397205588822355e-05,
      "loss": 2.7762,
      "step": 29500
    },
    {
      "epoch": 14.97,
      "learning_rate": 2.5149700598802394e-05,
      "loss": 2.7711,
      "step": 30000
    },
    {
      "epoch": 15.22,
      "learning_rate": 2.3902195608782436e-05,
      "loss": 2.7616,
      "step": 30500
    },
    {
      "epoch": 15.47,
      "learning_rate": 2.2654690618762475e-05,
      "loss": 2.7325,
      "step": 31000
    },
    {
      "epoch": 15.72,
      "learning_rate": 2.1407185628742514e-05,
      "loss": 2.7418,
      "step": 31500
    },
    {
      "epoch": 15.97,
      "learning_rate": 2.0159680638722556e-05,
      "loss": 2.7358,
      "step": 32000
    },
    {
      "epoch": 16.22,
      "learning_rate": 1.8912175648702595e-05,
      "loss": 2.7215,
      "step": 32500
    },
    {
      "epoch": 16.47,
      "learning_rate": 1.7664670658682637e-05,
      "loss": 2.7199,
      "step": 33000
    },
    {
      "epoch": 16.72,
      "learning_rate": 1.6417165668662676e-05,
      "loss": 2.7016,
      "step": 33500
    },
    {
      "epoch": 16.97,
      "learning_rate": 1.5169660678642716e-05,
      "loss": 2.6966,
      "step": 34000
    },
    {
      "epoch": 17.22,
      "learning_rate": 1.3922155688622754e-05,
      "loss": 2.6957,
      "step": 34500
    },
    {
      "epoch": 17.47,
      "learning_rate": 1.2674650698602794e-05,
      "loss": 2.7005,
      "step": 35000
    },
    {
      "epoch": 17.71,
      "learning_rate": 1.1427145708582835e-05,
      "loss": 2.6819,
      "step": 35500
    },
    {
      "epoch": 17.96,
      "learning_rate": 1.0179640718562875e-05,
      "loss": 2.6823,
      "step": 36000
    },
    {
      "epoch": 18.21,
      "learning_rate": 8.932135728542916e-06,
      "loss": 2.6879,
      "step": 36500
    },
    {
      "epoch": 18.46,
      "learning_rate": 7.684630738522954e-06,
      "loss": 2.6588,
      "step": 37000
    },
    {
      "epoch": 18.71,
      "learning_rate": 6.437125748502994e-06,
      "loss": 2.6521,
      "step": 37500
    },
    {
      "epoch": 18.96,
      "learning_rate": 5.189620758483034e-06,
      "loss": 2.6667,
      "step": 38000
    },
    {
      "epoch": 19.21,
      "learning_rate": 3.942115768463074e-06,
      "loss": 2.6727,
      "step": 38500
    },
    {
      "epoch": 19.46,
      "learning_rate": 2.6946107784431138e-06,
      "loss": 2.6702,
      "step": 39000
    },
    {
      "epoch": 19.71,
      "learning_rate": 1.4471057884231539e-06,
      "loss": 2.6484,
      "step": 39500
    },
    {
      "epoch": 19.96,
      "learning_rate": 1.996007984031936e-07,
      "loss": 2.6556,
      "step": 40000
    },
    {
      "epoch": 20.0,
      "step": 40080,
      "total_flos": 2362586579435520.0,
      "train_loss": 3.5848853368245197,
      "train_runtime": 32321.6149,
      "train_samples_per_second": 158.707,
      "train_steps_per_second": 1.24
    }
  ],
  "max_steps": 40080,
  "num_train_epochs": 20,
  "total_flos": 2362586579435520.0,
  "trial_name": null,
  "trial_params": null
}