{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "global_step": 58899,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.03,
      "learning_rate": 4.95755445763086e-05,
      "loss": 2.797,
      "step": 500
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.91510891526172e-05,
      "loss": 2.0557,
      "step": 1000
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.872663372892579e-05,
      "loss": 1.8501,
      "step": 1500
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.830217830523439e-05,
      "loss": 1.7462,
      "step": 2000
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.787772288154298e-05,
      "loss": 1.6583,
      "step": 2500
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.745326745785158e-05,
      "loss": 1.5985,
      "step": 3000
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.7028812034160174e-05,
      "loss": 1.5383,
      "step": 3500
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.660435661046877e-05,
      "loss": 1.5078,
      "step": 4000
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.617990118677737e-05,
      "loss": 1.4759,
      "step": 4500
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.575544576308596e-05,
      "loss": 1.4643,
      "step": 5000
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.533099033939456e-05,
      "loss": 1.4383,
      "step": 5500
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.490653491570315e-05,
      "loss": 1.4008,
      "step": 6000
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.4482079492011754e-05,
      "loss": 1.3803,
      "step": 6500
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.4057624068320345e-05,
      "loss": 1.3634,
      "step": 7000
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.363316864462894e-05,
      "loss": 1.351,
      "step": 7500
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.320871322093754e-05,
      "loss": 1.3307,
      "step": 8000
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.278425779724614e-05,
      "loss": 1.3303,
      "step": 8500
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.235980237355473e-05,
      "loss": 1.3105,
      "step": 9000
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.193534694986333e-05,
      "loss": 1.3139,
      "step": 9500
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.1510891526171925e-05,
      "loss": 1.2975,
      "step": 10000
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.1086436102480516e-05,
      "loss": 1.2893,
      "step": 10500
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.0661980678789114e-05,
      "loss": 1.2754,
      "step": 11000
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.023752525509771e-05,
      "loss": 1.2749,
      "step": 11500
    },
    {
      "epoch": 0.61,
      "learning_rate": 3.981306983140631e-05,
      "loss": 1.2569,
      "step": 12000
    },
    {
      "epoch": 0.64,
      "learning_rate": 3.93886144077149e-05,
      "loss": 1.2517,
      "step": 12500
    },
    {
      "epoch": 0.66,
      "learning_rate": 3.8964158984023505e-05,
      "loss": 1.2547,
      "step": 13000
    },
    {
      "epoch": 0.69,
      "learning_rate": 3.8539703560332096e-05,
      "loss": 1.2295,
      "step": 13500
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.8115248136640694e-05,
      "loss": 1.2121,
      "step": 14000
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.7690792712949285e-05,
      "loss": 1.2246,
      "step": 14500
    },
    {
      "epoch": 0.76,
      "learning_rate": 3.726633728925788e-05,
      "loss": 1.2127,
      "step": 15000
    },
    {
      "epoch": 0.79,
      "learning_rate": 3.684188186556648e-05,
      "loss": 1.1937,
      "step": 15500
    },
    {
      "epoch": 0.81,
      "learning_rate": 3.641742644187507e-05,
      "loss": 1.191,
      "step": 16000
    },
    {
      "epoch": 0.84,
      "learning_rate": 3.5992971018183676e-05,
      "loss": 1.1785,
      "step": 16500
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.556851559449227e-05,
      "loss": 1.1734,
      "step": 17000
    },
    {
      "epoch": 0.89,
      "learning_rate": 3.5144060170800865e-05,
      "loss": 1.1964,
      "step": 17500
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.4719604747109456e-05,
      "loss": 1.1786,
      "step": 18000
    },
    {
      "epoch": 0.94,
      "learning_rate": 3.429514932341806e-05,
      "loss": 1.1424,
      "step": 18500
    },
    {
      "epoch": 0.97,
      "learning_rate": 3.387069389972665e-05,
      "loss": 1.1677,
      "step": 19000
    },
    {
      "epoch": 0.99,
      "learning_rate": 3.344623847603525e-05,
      "loss": 1.1628,
      "step": 19500
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.302178305234385e-05,
      "loss": 1.0756,
      "step": 20000
    },
    {
      "epoch": 1.04,
      "learning_rate": 3.259732762865244e-05,
      "loss": 1.0654,
      "step": 20500
    },
    {
      "epoch": 1.07,
      "learning_rate": 3.2172872204961036e-05,
      "loss": 1.0666,
      "step": 21000
    },
    {
      "epoch": 1.1,
      "learning_rate": 3.174841678126963e-05,
      "loss": 1.0594,
      "step": 21500
    },
    {
      "epoch": 1.12,
      "learning_rate": 3.132396135757823e-05,
      "loss": 1.0642,
      "step": 22000
    },
    {
      "epoch": 1.15,
      "learning_rate": 3.089950593388682e-05,
      "loss": 1.0594,
      "step": 22500
    },
    {
      "epoch": 1.17,
      "learning_rate": 3.0475050510195417e-05,
      "loss": 1.0581,
      "step": 23000
    },
    {
      "epoch": 1.2,
      "learning_rate": 3.005059508650402e-05,
      "loss": 1.0561,
      "step": 23500
    },
    {
      "epoch": 1.22,
      "learning_rate": 2.9626139662812613e-05,
      "loss": 1.0298,
      "step": 24000
    },
    {
      "epoch": 1.25,
      "learning_rate": 2.9201684239121207e-05,
      "loss": 1.0451,
      "step": 24500
    },
    {
      "epoch": 1.27,
      "learning_rate": 2.877722881542981e-05,
      "loss": 1.0611,
      "step": 25000
    },
    {
      "epoch": 1.3,
      "learning_rate": 2.8352773391738403e-05,
      "loss": 1.0445,
      "step": 25500
    },
    {
      "epoch": 1.32,
      "learning_rate": 2.7928317968046997e-05,
      "loss": 1.0592,
      "step": 26000
    },
    {
      "epoch": 1.35,
      "learning_rate": 2.7503862544355592e-05,
      "loss": 1.0445,
      "step": 26500
    },
    {
      "epoch": 1.38,
      "learning_rate": 2.707940712066419e-05,
      "loss": 1.0338,
      "step": 27000
    },
    {
      "epoch": 1.4,
      "learning_rate": 2.6654951696972784e-05,
      "loss": 1.0331,
      "step": 27500
    },
    {
      "epoch": 1.43,
      "learning_rate": 2.623049627328138e-05,
      "loss": 1.0308,
      "step": 28000
    },
    {
      "epoch": 1.45,
      "learning_rate": 2.580604084958998e-05,
      "loss": 1.0329,
      "step": 28500
    },
    {
      "epoch": 1.48,
      "learning_rate": 2.5381585425898574e-05,
      "loss": 1.0178,
      "step": 29000
    },
    {
      "epoch": 1.5,
      "learning_rate": 2.495713000220717e-05,
      "loss": 1.0197,
      "step": 29500
    },
    {
      "epoch": 1.53,
      "learning_rate": 2.4532674578515766e-05,
      "loss": 1.0168,
      "step": 30000
    },
    {
      "epoch": 1.55,
      "learning_rate": 2.410821915482436e-05,
      "loss": 1.007,
      "step": 30500
    },
    {
      "epoch": 1.58,
      "learning_rate": 2.368376373113296e-05,
      "loss": 1.0249,
      "step": 31000
    },
    {
      "epoch": 1.6,
      "learning_rate": 2.3259308307441553e-05,
      "loss": 1.0178,
      "step": 31500
    },
    {
      "epoch": 1.63,
      "learning_rate": 2.2834852883750148e-05,
      "loss": 1.0199,
      "step": 32000
    },
    {
      "epoch": 1.66,
      "learning_rate": 2.2410397460058745e-05,
      "loss": 1.0089,
      "step": 32500
    },
    {
      "epoch": 1.68,
      "learning_rate": 2.198594203636734e-05,
      "loss": 1.0016,
      "step": 33000
    },
    {
      "epoch": 1.71,
      "learning_rate": 2.1561486612675938e-05,
      "loss": 1.0011,
      "step": 33500
    },
    {
      "epoch": 1.73,
      "learning_rate": 2.1137031188984535e-05,
      "loss": 0.9975,
      "step": 34000
    },
    {
      "epoch": 1.76,
      "learning_rate": 2.071257576529313e-05,
      "loss": 0.9856,
      "step": 34500
    },
    {
      "epoch": 1.78,
      "learning_rate": 2.0288120341601728e-05,
      "loss": 0.9977,
      "step": 35000
    },
    {
      "epoch": 1.81,
      "learning_rate": 1.9863664917910322e-05,
      "loss": 0.9933,
      "step": 35500
    },
    {
      "epoch": 1.83,
      "learning_rate": 1.943920949421892e-05,
      "loss": 1.0059,
      "step": 36000
    },
    {
      "epoch": 1.86,
      "learning_rate": 1.9014754070527514e-05,
      "loss": 1.0033,
      "step": 36500
    },
    {
      "epoch": 1.88,
      "learning_rate": 1.859029864683611e-05,
      "loss": 0.9744,
      "step": 37000
    },
    {
      "epoch": 1.91,
      "learning_rate": 1.8165843223144707e-05,
      "loss": 0.986,
      "step": 37500
    },
    {
      "epoch": 1.94,
      "learning_rate": 1.77413877994533e-05,
      "loss": 0.9853,
      "step": 38000
    },
    {
      "epoch": 1.96,
      "learning_rate": 1.73169323757619e-05,
      "loss": 0.9758,
      "step": 38500
    },
    {
      "epoch": 1.99,
      "learning_rate": 1.6892476952070493e-05,
      "loss": 0.9772,
      "step": 39000
    },
    {
      "epoch": 2.01,
      "learning_rate": 1.646802152837909e-05,
      "loss": 0.946,
      "step": 39500
    },
    {
      "epoch": 2.04,
      "learning_rate": 1.604356610468769e-05,
      "loss": 0.8919,
      "step": 40000
    },
    {
      "epoch": 2.06,
      "learning_rate": 1.5619110680996283e-05,
      "loss": 0.8903,
      "step": 40500
    },
    {
      "epoch": 2.09,
      "learning_rate": 1.519465525730488e-05,
      "loss": 0.8867,
      "step": 41000
    },
    {
      "epoch": 2.11,
      "learning_rate": 1.4770199833613474e-05,
      "loss": 0.8749,
      "step": 41500
    },
    {
      "epoch": 2.14,
      "learning_rate": 1.4345744409922072e-05,
      "loss": 0.8733,
      "step": 42000
    },
    {
      "epoch": 2.16,
      "learning_rate": 1.3921288986230666e-05,
      "loss": 0.8909,
      "step": 42500
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.3496833562539262e-05,
      "loss": 0.871,
      "step": 43000
    },
    {
      "epoch": 2.22,
      "learning_rate": 1.307237813884786e-05,
      "loss": 0.8741,
      "step": 43500
    },
    {
      "epoch": 2.24,
      "learning_rate": 1.2647922715156455e-05,
      "loss": 0.8856,
      "step": 44000
    },
    {
      "epoch": 2.27,
      "learning_rate": 1.222346729146505e-05,
      "loss": 0.8833,
      "step": 44500
    },
    {
      "epoch": 2.29,
      "learning_rate": 1.1799011867773647e-05,
      "loss": 0.8824,
      "step": 45000
    },
    {
      "epoch": 2.32,
      "learning_rate": 1.1374556444082243e-05,
      "loss": 0.8711,
      "step": 45500
    },
    {
      "epoch": 2.34,
      "learning_rate": 1.0950101020390839e-05,
      "loss": 0.8789,
      "step": 46000
    },
    {
      "epoch": 2.37,
      "learning_rate": 1.0525645596699435e-05,
      "loss": 0.8823,
      "step": 46500
    },
    {
      "epoch": 2.39,
      "learning_rate": 1.0101190173008031e-05,
      "loss": 0.8755,
      "step": 47000
    },
    {
      "epoch": 2.42,
      "learning_rate": 9.676734749316626e-06,
      "loss": 0.8715,
      "step": 47500
    },
    {
      "epoch": 2.44,
      "learning_rate": 9.252279325625224e-06,
      "loss": 0.8726,
      "step": 48000
    },
    {
      "epoch": 2.47,
      "learning_rate": 8.82782390193382e-06,
      "loss": 0.8811,
      "step": 48500
    },
    {
      "epoch": 2.5,
      "learning_rate": 8.403368478242416e-06,
      "loss": 0.8676,
      "step": 49000
    },
    {
      "epoch": 2.52,
      "learning_rate": 7.978913054551012e-06,
      "loss": 0.8727,
      "step": 49500
    },
    {
      "epoch": 2.55,
      "learning_rate": 7.554457630859607e-06,
      "loss": 0.8572,
      "step": 50000
    },
    {
      "epoch": 2.57,
      "learning_rate": 7.1300022071682025e-06,
      "loss": 0.86,
      "step": 50500
    },
    {
      "epoch": 2.6,
      "learning_rate": 6.7055467834768e-06,
      "loss": 0.8571,
      "step": 51000
    },
    {
      "epoch": 2.62,
      "learning_rate": 6.281091359785396e-06,
      "loss": 0.8488,
      "step": 51500
    },
    {
      "epoch": 2.65,
      "learning_rate": 5.856635936093992e-06,
      "loss": 0.857,
      "step": 52000
    },
    {
      "epoch": 2.67,
      "learning_rate": 5.432180512402588e-06,
      "loss": 0.8544,
      "step": 52500
    },
    {
      "epoch": 2.7,
      "learning_rate": 5.007725088711183e-06,
      "loss": 0.8611,
      "step": 53000
    },
    {
      "epoch": 2.73,
      "learning_rate": 4.58326966501978e-06,
      "loss": 0.8542,
      "step": 53500
    },
    {
      "epoch": 2.75,
      "learning_rate": 4.158814241328375e-06,
      "loss": 0.867,
      "step": 54000
    },
    {
      "epoch": 2.78,
      "learning_rate": 3.7343588176369715e-06,
      "loss": 0.8534,
      "step": 54500
    },
    {
      "epoch": 2.8,
      "learning_rate": 3.309903393945568e-06,
      "loss": 0.8754,
      "step": 55000
    },
    {
      "epoch": 2.83,
      "learning_rate": 2.8854479702541638e-06,
      "loss": 0.8451,
      "step": 55500
    },
    {
      "epoch": 2.85,
      "learning_rate": 2.4609925465627603e-06,
      "loss": 0.8531,
      "step": 56000
    },
    {
      "epoch": 2.88,
      "learning_rate": 2.0365371228713564e-06,
      "loss": 0.8538,
      "step": 56500
    },
    {
      "epoch": 2.9,
      "learning_rate": 1.6120816991799521e-06,
      "loss": 0.8474,
      "step": 57000
    },
    {
      "epoch": 2.93,
      "learning_rate": 1.1876262754885483e-06,
      "loss": 0.8585,
      "step": 57500
    },
    {
      "epoch": 2.95,
      "learning_rate": 7.631708517971443e-07,
      "loss": 0.8511,
      "step": 58000
    },
    {
      "epoch": 2.98,
      "learning_rate": 3.3871542810574035e-07,
      "loss": 0.8521,
      "step": 58500
    },
    {
      "epoch": 3.0,
      "step": 58899,
      "total_flos": 8.70325170291671e+16,
      "train_loss": 1.0935829183452552,
      "train_runtime": 19576.7665,
      "train_samples_per_second": 30.085,
      "train_steps_per_second": 3.009
    }
  ],
  "max_steps": 58899,
  "num_train_epochs": 3,
  "total_flos": 8.70325170291671e+16,
  "trial_name": null,
  "trial_params": null
}