{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.4356265074078327,
  "global_step": 12500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 4.968098002934984e-05,
      "loss": 0.8326,
      "step": 100
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.936196005869967e-05,
      "loss": 0.8028,
      "step": 200
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9042940088049516e-05,
      "loss": 0.7492,
      "step": 300
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.8723920117399354e-05,
      "loss": 0.7413,
      "step": 400
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.840490014674919e-05,
      "loss": 0.7788,
      "step": 500
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.808588017609902e-05,
      "loss": 0.6938,
      "step": 600
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.776686020544887e-05,
      "loss": 0.7157,
      "step": 700
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.7447840234798705e-05,
      "loss": 0.7606,
      "step": 800
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.7128820264148536e-05,
      "loss": 0.7222,
      "step": 900
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.6809800293498373e-05,
      "loss": 0.6986,
      "step": 1000
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.649078032284822e-05,
      "loss": 0.7202,
      "step": 1100
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.617176035219805e-05,
      "loss": 0.7456,
      "step": 1200
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.585274038154789e-05,
      "loss": 0.6753,
      "step": 1300
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.5533720410897725e-05,
      "loss": 0.6983,
      "step": 1400
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.521470044024756e-05,
      "loss": 0.7113,
      "step": 1500
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.48956804695974e-05,
      "loss": 0.7059,
      "step": 1600
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.457666049894724e-05,
      "loss": 0.691,
      "step": 1700
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.4257640528297076e-05,
      "loss": 0.752,
      "step": 1800
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.393862055764691e-05,
      "loss": 0.681,
      "step": 1900
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.361960058699675e-05,
      "loss": 0.7501,
      "step": 2000
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.330058061634659e-05,
      "loss": 0.6956,
      "step": 2100
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.298156064569642e-05,
      "loss": 0.6996,
      "step": 2200
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.266254067504626e-05,
      "loss": 0.6891,
      "step": 2300
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.2343520704396096e-05,
      "loss": 0.7419,
      "step": 2400
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.202450073374593e-05,
      "loss": 0.6979,
      "step": 2500
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.170548076309577e-05,
      "loss": 0.6425,
      "step": 2600
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.138646079244561e-05,
      "loss": 0.7243,
      "step": 2700
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.106744082179545e-05,
      "loss": 0.7348,
      "step": 2800
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.0748420851145285e-05,
      "loss": 0.6352,
      "step": 2900
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.042940088049512e-05,
      "loss": 0.6469,
      "step": 3000
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.011038090984496e-05,
      "loss": 0.718,
      "step": 3100
    },
    {
      "epoch": 0.37,
      "learning_rate": 3.979136093919479e-05,
      "loss": 0.7129,
      "step": 3200
    },
    {
      "epoch": 0.38,
      "learning_rate": 3.9472340968544636e-05,
      "loss": 0.672,
      "step": 3300
    },
    {
      "epoch": 0.39,
      "learning_rate": 3.9153320997894473e-05,
      "loss": 0.6616,
      "step": 3400
    },
    {
      "epoch": 0.4,
      "learning_rate": 3.8834301027244304e-05,
      "loss": 0.7105,
      "step": 3500
    },
    {
      "epoch": 0.41,
      "learning_rate": 3.851528105659414e-05,
      "loss": 0.7299,
      "step": 3600
    },
    {
      "epoch": 0.42,
      "learning_rate": 3.819626108594398e-05,
      "loss": 0.7022,
      "step": 3700
    },
    {
      "epoch": 0.44,
      "learning_rate": 3.787724111529382e-05,
      "loss": 0.6225,
      "step": 3800
    },
    {
      "epoch": 0.45,
      "learning_rate": 3.7558221144643656e-05,
      "loss": 0.6461,
      "step": 3900
    },
    {
      "epoch": 0.46,
      "learning_rate": 3.723920117399349e-05,
      "loss": 0.6935,
      "step": 4000
    },
    {
      "epoch": 0.47,
      "learning_rate": 3.692018120334333e-05,
      "loss": 0.671,
      "step": 4100
    },
    {
      "epoch": 0.48,
      "learning_rate": 3.660116123269317e-05,
      "loss": 0.6818,
      "step": 4200
    },
    {
      "epoch": 0.49,
      "learning_rate": 3.628214126204301e-05,
      "loss": 0.6343,
      "step": 4300
    },
    {
      "epoch": 0.51,
      "learning_rate": 3.5963121291392844e-05,
      "loss": 0.6623,
      "step": 4400
    },
    {
      "epoch": 0.52,
      "learning_rate": 3.5644101320742675e-05,
      "loss": 0.6896,
      "step": 4500
    },
    {
      "epoch": 0.53,
      "learning_rate": 3.532508135009251e-05,
      "loss": 0.6735,
      "step": 4600
    },
    {
      "epoch": 0.54,
      "learning_rate": 3.500606137944236e-05,
      "loss": 0.6723,
      "step": 4700
    },
    {
      "epoch": 0.55,
      "learning_rate": 3.4687041408792196e-05,
      "loss": 0.6224,
      "step": 4800
    },
    {
      "epoch": 0.56,
      "learning_rate": 3.4368021438142027e-05,
      "loss": 0.6902,
      "step": 4900
    },
    {
      "epoch": 0.57,
      "learning_rate": 3.4049001467491864e-05,
      "loss": 0.6264,
      "step": 5000
    },
    {
      "epoch": 0.59,
      "learning_rate": 3.372998149684171e-05,
      "loss": 0.6278,
      "step": 5100
    },
    {
      "epoch": 0.6,
      "learning_rate": 3.341096152619154e-05,
      "loss": 0.653,
      "step": 5200
    },
    {
      "epoch": 0.61,
      "learning_rate": 3.309194155554138e-05,
      "loss": 0.6709,
      "step": 5300
    },
    {
      "epoch": 0.62,
      "learning_rate": 3.2772921584891215e-05,
      "loss": 0.7207,
      "step": 5400
    },
    {
      "epoch": 0.63,
      "learning_rate": 3.245390161424105e-05,
      "loss": 0.671,
      "step": 5500
    },
    {
      "epoch": 0.64,
      "learning_rate": 3.213488164359089e-05,
      "loss": 0.7015,
      "step": 5600
    },
    {
      "epoch": 0.65,
      "learning_rate": 3.181586167294073e-05,
      "loss": 0.6896,
      "step": 5700
    },
    {
      "epoch": 0.67,
      "learning_rate": 3.1496841702290567e-05,
      "loss": 0.6674,
      "step": 5800
    },
    {
      "epoch": 0.68,
      "learning_rate": 3.11778217316404e-05,
      "loss": 0.6784,
      "step": 5900
    },
    {
      "epoch": 0.69,
      "learning_rate": 3.085880176099024e-05,
      "loss": 0.7048,
      "step": 6000
    },
    {
      "epoch": 0.7,
      "learning_rate": 3.053978179034008e-05,
      "loss": 0.6401,
      "step": 6100
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.0220761819689914e-05,
      "loss": 0.7202,
      "step": 6200
    },
    {
      "epoch": 0.72,
      "learning_rate": 2.990174184903975e-05,
      "loss": 0.6472,
      "step": 6300
    },
    {
      "epoch": 0.74,
      "learning_rate": 2.958272187838959e-05,
      "loss": 0.6601,
      "step": 6400
    },
    {
      "epoch": 0.75,
      "learning_rate": 2.9263701907739428e-05,
      "loss": 0.6788,
      "step": 6500
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.8944681937089262e-05,
      "loss": 0.745,
      "step": 6600
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.86256619664391e-05,
      "loss": 0.6654,
      "step": 6700
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.8306641995788934e-05,
      "loss": 0.6988,
      "step": 6800
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.7987622025138775e-05,
      "loss": 0.6084,
      "step": 6900
    },
    {
      "epoch": 0.8,
      "learning_rate": 2.7668602054488613e-05,
      "loss": 0.6551,
      "step": 7000
    },
    {
      "epoch": 0.82,
      "learning_rate": 2.7349582083838448e-05,
      "loss": 0.6264,
      "step": 7100
    },
    {
      "epoch": 0.83,
      "learning_rate": 2.7030562113188285e-05,
      "loss": 0.6456,
      "step": 7200
    },
    {
      "epoch": 0.84,
      "learning_rate": 2.6711542142538127e-05,
      "loss": 0.7103,
      "step": 7300
    },
    {
      "epoch": 0.85,
      "learning_rate": 2.639252217188796e-05,
      "loss": 0.7238,
      "step": 7400
    },
    {
      "epoch": 0.86,
      "learning_rate": 2.60735022012378e-05,
      "loss": 0.6286,
      "step": 7500
    },
    {
      "epoch": 0.87,
      "learning_rate": 2.5754482230587633e-05,
      "loss": 0.6284,
      "step": 7600
    },
    {
      "epoch": 0.88,
      "learning_rate": 2.5435462259937474e-05,
      "loss": 0.6169,
      "step": 7700
    },
    {
      "epoch": 0.9,
      "learning_rate": 2.5116442289287312e-05,
      "loss": 0.6334,
      "step": 7800
    },
    {
      "epoch": 0.91,
      "learning_rate": 2.4797422318637146e-05,
      "loss": 0.6573,
      "step": 7900
    },
    {
      "epoch": 0.92,
      "learning_rate": 2.4478402347986984e-05,
      "loss": 0.6026,
      "step": 8000
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.4159382377336822e-05,
      "loss": 0.6836,
      "step": 8100
    },
    {
      "epoch": 0.94,
      "learning_rate": 2.384036240668666e-05,
      "loss": 0.6093,
      "step": 8200
    },
    {
      "epoch": 0.95,
      "learning_rate": 2.3521342436036498e-05,
      "loss": 0.6603,
      "step": 8300
    },
    {
      "epoch": 0.96,
      "learning_rate": 2.3202322465386332e-05,
      "loss": 0.6312,
      "step": 8400
    },
    {
      "epoch": 0.98,
      "learning_rate": 2.2883302494736173e-05,
      "loss": 0.6312,
      "step": 8500
    },
    {
      "epoch": 0.99,
      "learning_rate": 2.2564282524086008e-05,
      "loss": 0.6278,
      "step": 8600
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.224526255343585e-05,
      "loss": 0.6115,
      "step": 8700
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.1926242582785683e-05,
      "loss": 0.6666,
      "step": 8800
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.160722261213552e-05,
      "loss": 0.6047,
      "step": 8900
    },
    {
      "epoch": 1.03,
      "learning_rate": 2.128820264148536e-05,
      "loss": 0.6174,
      "step": 9000
    },
    {
      "epoch": 1.05,
      "learning_rate": 2.0969182670835193e-05,
      "loss": 0.5977,
      "step": 9100
    },
    {
      "epoch": 1.06,
      "learning_rate": 2.0650162700185034e-05,
      "loss": 0.553,
      "step": 9200
    },
    {
      "epoch": 1.07,
      "learning_rate": 2.033114272953487e-05,
      "loss": 0.6447,
      "step": 9300
    },
    {
      "epoch": 1.08,
      "learning_rate": 2.0012122758884706e-05,
      "loss": 0.5979,
      "step": 9400
    },
    {
      "epoch": 1.09,
      "learning_rate": 1.9693102788234544e-05,
      "loss": 0.6234,
      "step": 9500
    },
    {
      "epoch": 1.1,
      "learning_rate": 1.9374082817584382e-05,
      "loss": 0.5939,
      "step": 9600
    },
    {
      "epoch": 1.11,
      "learning_rate": 1.905506284693422e-05,
      "loss": 0.6481,
      "step": 9700
    },
    {
      "epoch": 1.13,
      "learning_rate": 1.8736042876284058e-05,
      "loss": 0.592,
      "step": 9800
    },
    {
      "epoch": 1.14,
      "learning_rate": 1.8417022905633892e-05,
      "loss": 0.6491,
      "step": 9900
    },
    {
      "epoch": 1.15,
      "learning_rate": 1.8098002934983733e-05,
      "loss": 0.5978,
      "step": 10000
    },
    {
      "epoch": 1.16,
      "learning_rate": 1.7778982964333567e-05,
      "loss": 0.6034,
      "step": 10100
    },
    {
      "epoch": 1.17,
      "learning_rate": 1.7459962993683405e-05,
      "loss": 0.6037,
      "step": 10200
    },
    {
      "epoch": 1.18,
      "learning_rate": 1.7140943023033243e-05,
      "loss": 0.6251,
      "step": 10300
    },
    {
      "epoch": 1.19,
      "learning_rate": 1.6821923052383077e-05,
      "loss": 0.581,
      "step": 10400
    },
    {
      "epoch": 1.21,
      "learning_rate": 1.650290308173292e-05,
      "loss": 0.5919,
      "step": 10500
    },
    {
      "epoch": 1.22,
      "learning_rate": 1.6183883111082753e-05,
      "loss": 0.6073,
      "step": 10600
    },
    {
      "epoch": 1.23,
      "learning_rate": 1.5864863140432594e-05,
      "loss": 0.6273,
      "step": 10700
    },
    {
      "epoch": 1.24,
      "learning_rate": 1.554584316978243e-05,
      "loss": 0.6198,
      "step": 10800
    },
    {
      "epoch": 1.25,
      "learning_rate": 1.5226823199132268e-05,
      "loss": 0.5993,
      "step": 10900
    },
    {
      "epoch": 1.26,
      "learning_rate": 1.4907803228482104e-05,
      "loss": 0.5519,
      "step": 11000
    },
    {
      "epoch": 1.27,
      "learning_rate": 1.4588783257831942e-05,
      "loss": 0.6152,
      "step": 11100
    },
    {
      "epoch": 1.29,
      "learning_rate": 1.4269763287181778e-05,
      "loss": 0.6251,
      "step": 11200
    },
    {
      "epoch": 1.3,
      "learning_rate": 1.3950743316531614e-05,
      "loss": 0.6747,
      "step": 11300
    },
    {
      "epoch": 1.31,
      "learning_rate": 1.3631723345881454e-05,
      "loss": 0.6253,
      "step": 11400
    },
    {
      "epoch": 1.32,
      "learning_rate": 1.331270337523129e-05,
      "loss": 0.6009,
      "step": 11500
    },
    {
      "epoch": 1.33,
      "learning_rate": 1.2993683404581127e-05,
      "loss": 0.6019,
      "step": 11600
    },
    {
      "epoch": 1.34,
      "learning_rate": 1.2674663433930964e-05,
      "loss": 0.6286,
      "step": 11700
    },
    {
      "epoch": 1.36,
      "learning_rate": 1.2355643463280801e-05,
      "loss": 0.5796,
      "step": 11800
    },
    {
      "epoch": 1.37,
      "learning_rate": 1.2036623492630639e-05,
      "loss": 0.6077,
      "step": 11900
    },
    {
      "epoch": 1.38,
      "learning_rate": 1.1717603521980477e-05,
      "loss": 0.6186,
      "step": 12000
    },
    {
      "epoch": 1.39,
      "learning_rate": 1.1398583551330313e-05,
      "loss": 0.6632,
      "step": 12100
    },
    {
      "epoch": 1.4,
      "learning_rate": 1.107956358068015e-05,
      "loss": 0.5796,
      "step": 12200
    },
    {
      "epoch": 1.41,
      "learning_rate": 1.0760543610029989e-05,
      "loss": 0.5951,
      "step": 12300
    },
    {
      "epoch": 1.42,
      "learning_rate": 1.0441523639379826e-05,
      "loss": 0.5743,
      "step": 12400
    },
    {
      "epoch": 1.44,
      "learning_rate": 1.0122503668729662e-05,
      "loss": 0.6705,
      "step": 12500
    }
  ],
  "max_steps": 15673,
  "num_train_epochs": 2,
  "total_flos": 4.939903709604864e+16,
  "trial_name": null,
  "trial_params": null
}