{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.0,
  "global_step": 41475,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.06,
      "learning_rate": 4.939722724532851e-05,
      "loss": 2.1591,
      "step": 500
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8794454490657024e-05,
      "loss": 1.9533,
      "step": 1000
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.8191681735985535e-05,
      "loss": 1.8939,
      "step": 1500
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.7588908981314046e-05,
      "loss": 1.8664,
      "step": 2000
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.6986136226642556e-05,
      "loss": 1.8392,
      "step": 2500
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.638336347197107e-05,
      "loss": 1.8177,
      "step": 3000
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.5780590717299585e-05,
      "loss": 1.7998,
      "step": 3500
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.5177817962628096e-05,
      "loss": 1.7895,
      "step": 4000
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.45750452079566e-05,
      "loss": 1.7798,
      "step": 4500
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.397227245328511e-05,
      "loss": 1.7647,
      "step": 5000
    },
    {
      "epoch": 0.66,
      "learning_rate": 4.336949969861363e-05,
      "loss": 1.7568,
      "step": 5500
    },
    {
      "epoch": 0.72,
      "learning_rate": 4.276672694394214e-05,
      "loss": 1.7502,
      "step": 6000
    },
    {
      "epoch": 0.78,
      "learning_rate": 4.216395418927065e-05,
      "loss": 1.7346,
      "step": 6500
    },
    {
      "epoch": 0.84,
      "learning_rate": 4.1561181434599153e-05,
      "loss": 1.7177,
      "step": 7000
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.095840867992767e-05,
      "loss": 1.7246,
      "step": 7500
    },
    {
      "epoch": 0.96,
      "learning_rate": 4.035563592525618e-05,
      "loss": 1.7119,
      "step": 8000
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.975286317058469e-05,
      "loss": 1.7028,
      "step": 8500
    },
    {
      "epoch": 1.08,
      "learning_rate": 3.9150090415913203e-05,
      "loss": 1.685,
      "step": 9000
    },
    {
      "epoch": 1.15,
      "learning_rate": 3.8547317661241714e-05,
      "loss": 1.6779,
      "step": 9500
    },
    {
      "epoch": 1.21,
      "learning_rate": 3.7944544906570225e-05,
      "loss": 1.6613,
      "step": 10000
    },
    {
      "epoch": 1.27,
      "learning_rate": 3.7341772151898736e-05,
      "loss": 1.6646,
      "step": 10500
    },
    {
      "epoch": 1.33,
      "learning_rate": 3.6738999397227247e-05,
      "loss": 1.6491,
      "step": 11000
    },
    {
      "epoch": 1.39,
      "learning_rate": 3.613622664255576e-05,
      "loss": 1.6525,
      "step": 11500
    },
    {
      "epoch": 1.45,
      "learning_rate": 3.553345388788427e-05,
      "loss": 1.648,
      "step": 12000
    },
    {
      "epoch": 1.51,
      "learning_rate": 3.493068113321278e-05,
      "loss": 1.6482,
      "step": 12500
    },
    {
      "epoch": 1.57,
      "learning_rate": 3.432790837854129e-05,
      "loss": 1.6386,
      "step": 13000
    },
    {
      "epoch": 1.63,
      "learning_rate": 3.37251356238698e-05,
      "loss": 1.6421,
      "step": 13500
    },
    {
      "epoch": 1.69,
      "learning_rate": 3.312236286919831e-05,
      "loss": 1.6305,
      "step": 14000
    },
    {
      "epoch": 1.75,
      "learning_rate": 3.251959011452683e-05,
      "loss": 1.6255,
      "step": 14500
    },
    {
      "epoch": 1.81,
      "learning_rate": 3.191681735985534e-05,
      "loss": 1.6254,
      "step": 15000
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.1314044605183844e-05,
      "loss": 1.6217,
      "step": 15500
    },
    {
      "epoch": 1.93,
      "learning_rate": 3.0711271850512355e-05,
      "loss": 1.6082,
      "step": 16000
    },
    {
      "epoch": 1.99,
      "learning_rate": 3.010849909584087e-05,
      "loss": 1.6026,
      "step": 16500
    },
    {
      "epoch": 2.05,
      "learning_rate": 2.9505726341169383e-05,
      "loss": 1.5867,
      "step": 17000
    },
    {
      "epoch": 2.11,
      "learning_rate": 2.8902953586497894e-05,
      "loss": 1.5732,
      "step": 17500
    },
    {
      "epoch": 2.17,
      "learning_rate": 2.83001808318264e-05,
      "loss": 1.5877,
      "step": 18000
    },
    {
      "epoch": 2.23,
      "learning_rate": 2.7697408077154912e-05,
      "loss": 1.5735,
      "step": 18500
    },
    {
      "epoch": 2.29,
      "learning_rate": 2.7094635322483426e-05,
      "loss": 1.5707,
      "step": 19000
    },
    {
      "epoch": 2.35,
      "learning_rate": 2.6491862567811937e-05,
      "loss": 1.5633,
      "step": 19500
    },
    {
      "epoch": 2.41,
      "learning_rate": 2.5889089813140448e-05,
      "loss": 1.5716,
      "step": 20000
    },
    {
      "epoch": 2.47,
      "learning_rate": 2.5286317058468955e-05,
      "loss": 1.5516,
      "step": 20500
    },
    {
      "epoch": 2.53,
      "learning_rate": 2.468354430379747e-05,
      "loss": 1.5582,
      "step": 21000
    },
    {
      "epoch": 2.59,
      "learning_rate": 2.408077154912598e-05,
      "loss": 1.5618,
      "step": 21500
    },
    {
      "epoch": 2.65,
      "learning_rate": 2.347799879445449e-05,
      "loss": 1.545,
      "step": 22000
    },
    {
      "epoch": 2.71,
      "learning_rate": 2.2875226039783005e-05,
      "loss": 1.5472,
      "step": 22500
    },
    {
      "epoch": 2.77,
      "learning_rate": 2.2272453285111512e-05,
      "loss": 1.5445,
      "step": 23000
    },
    {
      "epoch": 2.83,
      "learning_rate": 2.1669680530440027e-05,
      "loss": 1.5458,
      "step": 23500
    },
    {
      "epoch": 2.89,
      "learning_rate": 2.1066907775768534e-05,
      "loss": 1.5449,
      "step": 24000
    },
    {
      "epoch": 2.95,
      "learning_rate": 2.0464135021097048e-05,
      "loss": 1.5378,
      "step": 24500
    },
    {
      "epoch": 3.01,
      "learning_rate": 1.986136226642556e-05,
      "loss": 1.5272,
      "step": 25000
    },
    {
      "epoch": 3.07,
      "learning_rate": 1.925858951175407e-05,
      "loss": 1.5207,
      "step": 25500
    },
    {
      "epoch": 3.13,
      "learning_rate": 1.865581675708258e-05,
      "loss": 1.5153,
      "step": 26000
    },
    {
      "epoch": 3.19,
      "learning_rate": 1.8053044002411095e-05,
      "loss": 1.5062,
      "step": 26500
    },
    {
      "epoch": 3.25,
      "learning_rate": 1.7450271247739602e-05,
      "loss": 1.5083,
      "step": 27000
    },
    {
      "epoch": 3.32,
      "learning_rate": 1.6847498493068113e-05,
      "loss": 1.5091,
      "step": 27500
    },
    {
      "epoch": 3.38,
      "learning_rate": 1.6244725738396624e-05,
      "loss": 1.4977,
      "step": 28000
    },
    {
      "epoch": 3.44,
      "learning_rate": 1.5641952983725134e-05,
      "loss": 1.4937,
      "step": 28500
    },
    {
      "epoch": 3.5,
      "learning_rate": 1.5039180229053649e-05,
      "loss": 1.5031,
      "step": 29000
    },
    {
      "epoch": 3.56,
      "learning_rate": 1.4436407474382158e-05,
      "loss": 1.5006,
      "step": 29500
    },
    {
      "epoch": 3.62,
      "learning_rate": 1.383363471971067e-05,
      "loss": 1.4899,
      "step": 30000
    },
    {
      "epoch": 3.68,
      "learning_rate": 1.3230861965039181e-05,
      "loss": 1.4942,
      "step": 30500
    },
    {
      "epoch": 3.74,
      "learning_rate": 1.2628089210367692e-05,
      "loss": 1.48,
      "step": 31000
    },
    {
      "epoch": 3.8,
      "learning_rate": 1.2025316455696203e-05,
      "loss": 1.4784,
      "step": 31500
    },
    {
      "epoch": 3.86,
      "learning_rate": 1.1422543701024713e-05,
      "loss": 1.4768,
      "step": 32000
    },
    {
      "epoch": 3.92,
      "learning_rate": 1.0819770946353224e-05,
      "loss": 1.4776,
      "step": 32500
    },
    {
      "epoch": 3.98,
      "learning_rate": 1.0216998191681737e-05,
      "loss": 1.4759,
      "step": 33000
    },
    {
      "epoch": 4.04,
      "learning_rate": 9.614225437010247e-06,
      "loss": 1.4794,
      "step": 33500
    },
    {
      "epoch": 4.1,
      "learning_rate": 9.011452682338758e-06,
      "loss": 1.4614,
      "step": 34000
    },
    {
      "epoch": 4.16,
      "learning_rate": 8.408679927667269e-06,
      "loss": 1.4605,
      "step": 34500
    },
    {
      "epoch": 4.22,
      "learning_rate": 7.805907172995782e-06,
      "loss": 1.4724,
      "step": 35000
    },
    {
      "epoch": 4.28,
      "learning_rate": 7.203134418324292e-06,
      "loss": 1.4625,
      "step": 35500
    },
    {
      "epoch": 4.34,
      "learning_rate": 6.600361663652803e-06,
      "loss": 1.4623,
      "step": 36000
    },
    {
      "epoch": 4.4,
      "learning_rate": 5.997588908981315e-06,
      "loss": 1.4485,
      "step": 36500
    },
    {
      "epoch": 4.46,
      "learning_rate": 5.3948161543098255e-06,
      "loss": 1.4532,
      "step": 37000
    },
    {
      "epoch": 4.52,
      "learning_rate": 4.792043399638336e-06,
      "loss": 1.4435,
      "step": 37500
    },
    {
      "epoch": 4.58,
      "learning_rate": 4.189270644966848e-06,
      "loss": 1.4503,
      "step": 38000
    },
    {
      "epoch": 4.64,
      "learning_rate": 3.5864978902953588e-06,
      "loss": 1.4512,
      "step": 38500
    },
    {
      "epoch": 4.7,
      "learning_rate": 2.98372513562387e-06,
      "loss": 1.4388,
      "step": 39000
    },
    {
      "epoch": 4.76,
      "learning_rate": 2.3809523809523808e-06,
      "loss": 1.4554,
      "step": 39500
    },
    {
      "epoch": 4.82,
      "learning_rate": 1.778179626280892e-06,
      "loss": 1.4481,
      "step": 40000
    },
    {
      "epoch": 4.88,
      "learning_rate": 1.1754068716094034e-06,
      "loss": 1.4525,
      "step": 40500
    },
    {
      "epoch": 4.94,
      "learning_rate": 5.726341169379144e-07,
      "loss": 1.4436,
      "step": 41000
    },
    {
      "epoch": 5.0,
      "step": 41475,
      "total_flos": 6.56610253452288e+16,
      "train_loss": 1.5926296373474231,
      "train_runtime": 22304.2482,
      "train_samples_per_second": 11.156,
      "train_steps_per_second": 1.86
    }
  ],
  "max_steps": 41475,
  "num_train_epochs": 5,
  "total_flos": 6.56610253452288e+16,
  "trial_name": null,
  "trial_params": null
}