{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "global_step": 68478,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.02,
      "learning_rate": 4.963491924413681e-05,
      "loss": 2.5929,
      "step": 500
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9269838488273604e-05,
      "loss": 2.2003,
      "step": 1000
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.890475773241041e-05,
      "loss": 2.0692,
      "step": 1500
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.853967697654722e-05,
      "loss": 1.9485,
      "step": 2000
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.817459622068401e-05,
      "loss": 1.9084,
      "step": 2500
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.780951546482082e-05,
      "loss": 1.8495,
      "step": 3000
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.744443470895763e-05,
      "loss": 1.8025,
      "step": 3500
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.707935395309443e-05,
      "loss": 1.7957,
      "step": 4000
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.671427319723123e-05,
      "loss": 1.7414,
      "step": 4500
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.634919244136804e-05,
      "loss": 1.7284,
      "step": 5000
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.598411168550484e-05,
      "loss": 1.7033,
      "step": 5500
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.561903092964164e-05,
      "loss": 1.705,
      "step": 6000
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.525395017377844e-05,
      "loss": 1.667,
      "step": 6500
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.4888869417915246e-05,
      "loss": 1.6606,
      "step": 7000
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.452378866205205e-05,
      "loss": 1.6179,
      "step": 7500
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.415870790618885e-05,
      "loss": 1.5977,
      "step": 8000
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.3793627150325655e-05,
      "loss": 1.5827,
      "step": 8500
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.3428546394462456e-05,
      "loss": 1.5621,
      "step": 9000
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.306346563859926e-05,
      "loss": 1.5675,
      "step": 9500
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.2698384882736064e-05,
      "loss": 1.5729,
      "step": 10000
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.2333304126872865e-05,
      "loss": 1.546,
      "step": 10500
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.1968223371009666e-05,
      "loss": 1.5546,
      "step": 11000
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.160314261514647e-05,
      "loss": 1.5163,
      "step": 11500
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.123806185928328e-05,
      "loss": 1.5097,
      "step": 12000
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.0872981103420075e-05,
      "loss": 1.5074,
      "step": 12500
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.050790034755688e-05,
      "loss": 1.4856,
      "step": 13000
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.014281959169369e-05,
      "loss": 1.5068,
      "step": 13500
    },
    {
      "epoch": 0.61,
      "learning_rate": 3.9777738835830483e-05,
      "loss": 1.5035,
      "step": 14000
    },
    {
      "epoch": 0.64,
      "learning_rate": 3.941265807996729e-05,
      "loss": 1.4676,
      "step": 14500
    },
    {
      "epoch": 0.66,
      "learning_rate": 3.90475773241041e-05,
      "loss": 1.4623,
      "step": 15000
    },
    {
      "epoch": 0.68,
      "learning_rate": 3.868249656824089e-05,
      "loss": 1.4472,
      "step": 15500
    },
    {
      "epoch": 0.7,
      "learning_rate": 3.83174158123777e-05,
      "loss": 1.447,
      "step": 16000
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.795233505651451e-05,
      "loss": 1.4557,
      "step": 16500
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.75872543006513e-05,
      "loss": 1.4439,
      "step": 17000
    },
    {
      "epoch": 0.77,
      "learning_rate": 3.722217354478811e-05,
      "loss": 1.4332,
      "step": 17500
    },
    {
      "epoch": 0.79,
      "learning_rate": 3.685709278892491e-05,
      "loss": 1.4365,
      "step": 18000
    },
    {
      "epoch": 0.81,
      "learning_rate": 3.649201203306172e-05,
      "loss": 1.4098,
      "step": 18500
    },
    {
      "epoch": 0.83,
      "learning_rate": 3.612693127719852e-05,
      "loss": 1.4125,
      "step": 19000
    },
    {
      "epoch": 0.85,
      "learning_rate": 3.576185052133532e-05,
      "loss": 1.4088,
      "step": 19500
    },
    {
      "epoch": 0.88,
      "learning_rate": 3.5396769765472126e-05,
      "loss": 1.3968,
      "step": 20000
    },
    {
      "epoch": 0.9,
      "learning_rate": 3.503168900960893e-05,
      "loss": 1.4125,
      "step": 20500
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.466660825374573e-05,
      "loss": 1.385,
      "step": 21000
    },
    {
      "epoch": 0.94,
      "learning_rate": 3.4301527497882535e-05,
      "loss": 1.3985,
      "step": 21500
    },
    {
      "epoch": 0.96,
      "learning_rate": 3.3936446742019336e-05,
      "loss": 1.3747,
      "step": 22000
    },
    {
      "epoch": 0.99,
      "learning_rate": 3.3571365986156136e-05,
      "loss": 1.3743,
      "step": 22500
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.3206285230292944e-05,
      "loss": 1.3065,
      "step": 23000
    },
    {
      "epoch": 1.03,
      "learning_rate": 3.2841204474429745e-05,
      "loss": 1.1883,
      "step": 23500
    },
    {
      "epoch": 1.05,
      "learning_rate": 3.2476123718566545e-05,
      "loss": 1.2017,
      "step": 24000
    },
    {
      "epoch": 1.07,
      "learning_rate": 3.211104296270335e-05,
      "loss": 1.1919,
      "step": 24500
    },
    {
      "epoch": 1.1,
      "learning_rate": 3.1745962206840153e-05,
      "loss": 1.1873,
      "step": 25000
    },
    {
      "epoch": 1.12,
      "learning_rate": 3.1380881450976954e-05,
      "loss": 1.2045,
      "step": 25500
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.101580069511376e-05,
      "loss": 1.1915,
      "step": 26000
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.065071993925057e-05,
      "loss": 1.1932,
      "step": 26500
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.0285639183387367e-05,
      "loss": 1.1793,
      "step": 27000
    },
    {
      "epoch": 1.2,
      "learning_rate": 2.992055842752417e-05,
      "loss": 1.2086,
      "step": 27500
    },
    {
      "epoch": 1.23,
      "learning_rate": 2.9555477671660975e-05,
      "loss": 1.2023,
      "step": 28000
    },
    {
      "epoch": 1.25,
      "learning_rate": 2.9190396915797775e-05,
      "loss": 1.1953,
      "step": 28500
    },
    {
      "epoch": 1.27,
      "learning_rate": 2.882531615993458e-05,
      "loss": 1.2049,
      "step": 29000
    },
    {
      "epoch": 1.29,
      "learning_rate": 2.8460235404071384e-05,
      "loss": 1.1931,
      "step": 29500
    },
    {
      "epoch": 1.31,
      "learning_rate": 2.8095154648208184e-05,
      "loss": 1.197,
      "step": 30000
    },
    {
      "epoch": 1.34,
      "learning_rate": 2.773007389234499e-05,
      "loss": 1.1704,
      "step": 30500
    },
    {
      "epoch": 1.36,
      "learning_rate": 2.7364993136481793e-05,
      "loss": 1.1548,
      "step": 31000
    },
    {
      "epoch": 1.38,
      "learning_rate": 2.6999912380618593e-05,
      "loss": 1.187,
      "step": 31500
    },
    {
      "epoch": 1.4,
      "learning_rate": 2.6634831624755397e-05,
      "loss": 1.1769,
      "step": 32000
    },
    {
      "epoch": 1.42,
      "learning_rate": 2.62697508688922e-05,
      "loss": 1.181,
      "step": 32500
    },
    {
      "epoch": 1.45,
      "learning_rate": 2.5904670113029006e-05,
      "loss": 1.1564,
      "step": 33000
    },
    {
      "epoch": 1.47,
      "learning_rate": 2.5539589357165806e-05,
      "loss": 1.1742,
      "step": 33500
    },
    {
      "epoch": 1.49,
      "learning_rate": 2.517450860130261e-05,
      "loss": 1.1651,
      "step": 34000
    },
    {
      "epoch": 1.51,
      "learning_rate": 2.480942784543941e-05,
      "loss": 1.1637,
      "step": 34500
    },
    {
      "epoch": 1.53,
      "learning_rate": 2.4444347089576215e-05,
      "loss": 1.1531,
      "step": 35000
    },
    {
      "epoch": 1.56,
      "learning_rate": 2.407926633371302e-05,
      "loss": 1.1703,
      "step": 35500
    },
    {
      "epoch": 1.58,
      "learning_rate": 2.371418557784982e-05,
      "loss": 1.1482,
      "step": 36000
    },
    {
      "epoch": 1.6,
      "learning_rate": 2.3349104821986624e-05,
      "loss": 1.1501,
      "step": 36500
    },
    {
      "epoch": 1.62,
      "learning_rate": 2.2984024066123428e-05,
      "loss": 1.1525,
      "step": 37000
    },
    {
      "epoch": 1.64,
      "learning_rate": 2.261894331026023e-05,
      "loss": 1.1576,
      "step": 37500
    },
    {
      "epoch": 1.66,
      "learning_rate": 2.2253862554397033e-05,
      "loss": 1.1518,
      "step": 38000
    },
    {
      "epoch": 1.69,
      "learning_rate": 2.1888781798533837e-05,
      "loss": 1.1594,
      "step": 38500
    },
    {
      "epoch": 1.71,
      "learning_rate": 2.152370104267064e-05,
      "loss": 1.1614,
      "step": 39000
    },
    {
      "epoch": 1.73,
      "learning_rate": 2.1158620286807442e-05,
      "loss": 1.185,
      "step": 39500
    },
    {
      "epoch": 1.75,
      "learning_rate": 2.0793539530944246e-05,
      "loss": 1.1399,
      "step": 40000
    },
    {
      "epoch": 1.77,
      "learning_rate": 2.042845877508105e-05,
      "loss": 1.1262,
      "step": 40500
    },
    {
      "epoch": 1.8,
      "learning_rate": 2.006337801921785e-05,
      "loss": 1.1458,
      "step": 41000
    },
    {
      "epoch": 1.82,
      "learning_rate": 1.9698297263354655e-05,
      "loss": 1.1379,
      "step": 41500
    },
    {
      "epoch": 1.84,
      "learning_rate": 1.933321650749146e-05,
      "loss": 1.1437,
      "step": 42000
    },
    {
      "epoch": 1.86,
      "learning_rate": 1.896813575162826e-05,
      "loss": 1.1161,
      "step": 42500
    },
    {
      "epoch": 1.88,
      "learning_rate": 1.8603054995765064e-05,
      "loss": 1.1205,
      "step": 43000
    },
    {
      "epoch": 1.91,
      "learning_rate": 1.8237974239901868e-05,
      "loss": 1.1422,
      "step": 43500
    },
    {
      "epoch": 1.93,
      "learning_rate": 1.787289348403867e-05,
      "loss": 1.1265,
      "step": 44000
    },
    {
      "epoch": 1.95,
      "learning_rate": 1.7507812728175473e-05,
      "loss": 1.0999,
      "step": 44500
    },
    {
      "epoch": 1.97,
      "learning_rate": 1.7142731972312277e-05,
      "loss": 1.11,
      "step": 45000
    },
    {
      "epoch": 1.99,
      "learning_rate": 1.677765121644908e-05,
      "loss": 1.1267,
      "step": 45500
    },
    {
      "epoch": 2.02,
      "learning_rate": 1.6412570460585882e-05,
      "loss": 1.0115,
      "step": 46000
    },
    {
      "epoch": 2.04,
      "learning_rate": 1.6047489704722686e-05,
      "loss": 0.9535,
      "step": 46500
    },
    {
      "epoch": 2.06,
      "learning_rate": 1.568240894885949e-05,
      "loss": 0.9458,
      "step": 47000
    },
    {
      "epoch": 2.08,
      "learning_rate": 1.531732819299629e-05,
      "loss": 0.9517,
      "step": 47500
    },
    {
      "epoch": 2.1,
      "learning_rate": 1.4952247437133093e-05,
      "loss": 0.9496,
      "step": 48000
    },
    {
      "epoch": 2.12,
      "learning_rate": 1.4587166681269897e-05,
      "loss": 0.9649,
      "step": 48500
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.42220859254067e-05,
      "loss": 0.934,
      "step": 49000
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.3857005169543506e-05,
      "loss": 0.9757,
      "step": 49500
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.3491924413680306e-05,
      "loss": 0.9349,
      "step": 50000
    },
    {
      "epoch": 2.21,
      "learning_rate": 1.3126843657817109e-05,
      "loss": 0.9548,
      "step": 50500
    },
    {
      "epoch": 2.23,
      "learning_rate": 1.2761762901953914e-05,
      "loss": 0.9426,
      "step": 51000
    },
    {
      "epoch": 2.26,
      "learning_rate": 1.2396682146090715e-05,
      "loss": 0.9543,
      "step": 51500
    },
    {
      "epoch": 2.28,
      "learning_rate": 1.203160139022752e-05,
      "loss": 0.9438,
      "step": 52000
    },
    {
      "epoch": 2.3,
      "learning_rate": 1.1666520634364322e-05,
      "loss": 0.963,
      "step": 52500
    },
    {
      "epoch": 2.32,
      "learning_rate": 1.1301439878501124e-05,
      "loss": 0.9334,
      "step": 53000
    },
    {
      "epoch": 2.34,
      "learning_rate": 1.0936359122637928e-05,
      "loss": 0.9526,
      "step": 53500
    },
    {
      "epoch": 2.37,
      "learning_rate": 1.057127836677473e-05,
      "loss": 0.9397,
      "step": 54000
    },
    {
      "epoch": 2.39,
      "learning_rate": 1.0206197610911535e-05,
      "loss": 0.9569,
      "step": 54500
    },
    {
      "epoch": 2.41,
      "learning_rate": 9.841116855048337e-06,
      "loss": 0.94,
      "step": 55000
    },
    {
      "epoch": 2.43,
      "learning_rate": 9.47603609918514e-06,
      "loss": 0.9345,
      "step": 55500
    },
    {
      "epoch": 2.45,
      "learning_rate": 9.110955343321944e-06,
      "loss": 0.933,
      "step": 56000
    },
    {
      "epoch": 2.48,
      "learning_rate": 8.745874587458746e-06,
      "loss": 0.9289,
      "step": 56500
    },
    {
      "epoch": 2.5,
      "learning_rate": 8.380793831595548e-06,
      "loss": 0.9581,
      "step": 57000
    },
    {
      "epoch": 2.52,
      "learning_rate": 8.015713075732353e-06,
      "loss": 0.9203,
      "step": 57500
    },
    {
      "epoch": 2.54,
      "learning_rate": 7.650632319869155e-06,
      "loss": 0.9539,
      "step": 58000
    },
    {
      "epoch": 2.56,
      "learning_rate": 7.285551564005959e-06,
      "loss": 0.9482,
      "step": 58500
    },
    {
      "epoch": 2.58,
      "learning_rate": 6.920470808142762e-06,
      "loss": 0.9206,
      "step": 59000
    },
    {
      "epoch": 2.61,
      "learning_rate": 6.555390052279564e-06,
      "loss": 0.9186,
      "step": 59500
    },
    {
      "epoch": 2.63,
      "learning_rate": 6.190309296416367e-06,
      "loss": 0.9168,
      "step": 60000
    },
    {
      "epoch": 2.65,
      "learning_rate": 5.825228540553171e-06,
      "loss": 0.9151,
      "step": 60500
    },
    {
      "epoch": 2.67,
      "learning_rate": 5.460147784689974e-06,
      "loss": 0.924,
      "step": 61000
    },
    {
      "epoch": 2.69,
      "learning_rate": 5.095067028826777e-06,
      "loss": 0.916,
      "step": 61500
    },
    {
      "epoch": 2.72,
      "learning_rate": 4.729986272963579e-06,
      "loss": 0.9311,
      "step": 62000
    },
    {
      "epoch": 2.74,
      "learning_rate": 4.364905517100383e-06,
      "loss": 0.9175,
      "step": 62500
    },
    {
      "epoch": 2.76,
      "learning_rate": 3.999824761237186e-06,
      "loss": 0.9121,
      "step": 63000
    },
    {
      "epoch": 2.78,
      "learning_rate": 3.634744005373989e-06,
      "loss": 0.9226,
      "step": 63500
    },
    {
      "epoch": 2.8,
      "learning_rate": 3.269663249510792e-06,
      "loss": 0.9335,
      "step": 64000
    },
    {
      "epoch": 2.83,
      "learning_rate": 2.9045824936475953e-06,
      "loss": 0.9104,
      "step": 64500
    },
    {
      "epoch": 2.85,
      "learning_rate": 2.539501737784398e-06,
      "loss": 0.9197,
      "step": 65000
    },
    {
      "epoch": 2.87,
      "learning_rate": 2.1744209819212013e-06,
      "loss": 0.8976,
      "step": 65500
    },
    {
      "epoch": 2.89,
      "learning_rate": 1.8093402260580042e-06,
      "loss": 0.9401,
      "step": 66000
    },
    {
      "epoch": 2.91,
      "learning_rate": 1.444259470194807e-06,
      "loss": 0.9265,
      "step": 66500
    },
    {
      "epoch": 2.94,
      "learning_rate": 1.07917871433161e-06,
      "loss": 0.9255,
      "step": 67000
    },
    {
      "epoch": 2.96,
      "learning_rate": 7.140979584684132e-07,
      "loss": 0.9099,
      "step": 67500
    },
    {
      "epoch": 2.98,
      "learning_rate": 3.490172026052163e-07,
      "loss": 0.9033,
      "step": 68000
    },
    {
      "epoch": 3.0,
      "step": 68478,
      "total_flos": 9.218136679415808e+16,
      "train_loss": 1.233144689885432,
      "train_runtime": 17748.5921,
      "train_samples_per_second": 38.582,
      "train_steps_per_second": 3.858
    }
  ],
  "max_steps": 68478,
  "num_train_epochs": 3,
  "total_flos": 9.218136679415808e+16,
  "trial_name": null,
  "trial_params": null
}