{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 10.0,
  "global_step": 84090,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.06,
      "learning_rate": 4.970448329171126e-05,
      "loss": 3.6877,
      "step": 500
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.9407777381377095e-05,
      "loss": 3.4145,
      "step": 1000
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.911047687002022e-05,
      "loss": 3.3234,
      "step": 1500
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.881317635866334e-05,
      "loss": 3.2698,
      "step": 2000
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.851587584730646e-05,
      "loss": 3.2274,
      "step": 2500
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.821857533594958e-05,
      "loss": 3.1993,
      "step": 3000
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.79212748245927e-05,
      "loss": 3.1753,
      "step": 3500
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.762397431323582e-05,
      "loss": 3.1476,
      "step": 4000
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.732667380187894e-05,
      "loss": 3.1343,
      "step": 4500
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.702937329052206e-05,
      "loss": 3.1173,
      "step": 5000
    },
    {
      "epoch": 0.65,
      "learning_rate": 4.673207277916518e-05,
      "loss": 3.0942,
      "step": 5500
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.6435366868831015e-05,
      "loss": 3.0873,
      "step": 6000
    },
    {
      "epoch": 0.77,
      "learning_rate": 4.613806635747413e-05,
      "loss": 3.0778,
      "step": 6500
    },
    {
      "epoch": 0.83,
      "learning_rate": 4.584076584611726e-05,
      "loss": 3.065,
      "step": 7000
    },
    {
      "epoch": 0.89,
      "learning_rate": 4.5543465334760376e-05,
      "loss": 3.0566,
      "step": 7500
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.5246164823403495e-05,
      "loss": 3.0428,
      "step": 8000
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.494945891306933e-05,
      "loss": 3.03,
      "step": 8500
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.4652158401712454e-05,
      "loss": 2.9815,
      "step": 9000
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.435485789035557e-05,
      "loss": 2.9757,
      "step": 9500
    },
    {
      "epoch": 1.19,
      "learning_rate": 4.405755737899869e-05,
      "loss": 2.9704,
      "step": 10000
    },
    {
      "epoch": 1.25,
      "learning_rate": 4.3760851468664525e-05,
      "loss": 2.9729,
      "step": 10500
    },
    {
      "epoch": 1.31,
      "learning_rate": 4.346355095730765e-05,
      "loss": 2.9675,
      "step": 11000
    },
    {
      "epoch": 1.37,
      "learning_rate": 4.316625044595077e-05,
      "loss": 2.9551,
      "step": 11500
    },
    {
      "epoch": 1.43,
      "learning_rate": 4.286894993459389e-05,
      "loss": 2.95,
      "step": 12000
    },
    {
      "epoch": 1.49,
      "learning_rate": 4.257164942323701e-05,
      "loss": 2.949,
      "step": 12500
    },
    {
      "epoch": 1.55,
      "learning_rate": 4.2274943512902846e-05,
      "loss": 2.9451,
      "step": 13000
    },
    {
      "epoch": 1.61,
      "learning_rate": 4.1977643001545964e-05,
      "loss": 2.9412,
      "step": 13500
    },
    {
      "epoch": 1.66,
      "learning_rate": 4.168034249018908e-05,
      "loss": 2.9367,
      "step": 14000
    },
    {
      "epoch": 1.72,
      "learning_rate": 4.13830419788322e-05,
      "loss": 2.9247,
      "step": 14500
    },
    {
      "epoch": 1.78,
      "learning_rate": 4.1085741467475326e-05,
      "loss": 2.9287,
      "step": 15000
    },
    {
      "epoch": 1.84,
      "learning_rate": 4.078903555714116e-05,
      "loss": 2.913,
      "step": 15500
    },
    {
      "epoch": 1.9,
      "learning_rate": 4.049173504578428e-05,
      "loss": 2.923,
      "step": 16000
    },
    {
      "epoch": 1.96,
      "learning_rate": 4.01944345344274e-05,
      "loss": 2.9127,
      "step": 16500
    },
    {
      "epoch": 2.02,
      "learning_rate": 3.989713402307052e-05,
      "loss": 2.8935,
      "step": 17000
    },
    {
      "epoch": 2.08,
      "learning_rate": 3.959983351171364e-05,
      "loss": 2.8674,
      "step": 17500
    },
    {
      "epoch": 2.14,
      "learning_rate": 3.930253300035676e-05,
      "loss": 2.8602,
      "step": 18000
    },
    {
      "epoch": 2.2,
      "learning_rate": 3.900582709002259e-05,
      "loss": 2.8658,
      "step": 18500
    },
    {
      "epoch": 2.26,
      "learning_rate": 3.870852657866572e-05,
      "loss": 2.8655,
      "step": 19000
    },
    {
      "epoch": 2.32,
      "learning_rate": 3.8411226067308836e-05,
      "loss": 2.8594,
      "step": 19500
    },
    {
      "epoch": 2.38,
      "learning_rate": 3.8113925555951955e-05,
      "loss": 2.8506,
      "step": 20000
    },
    {
      "epoch": 2.44,
      "learning_rate": 3.781662504459508e-05,
      "loss": 2.8645,
      "step": 20500
    },
    {
      "epoch": 2.5,
      "learning_rate": 3.75193245332382e-05,
      "loss": 2.8637,
      "step": 21000
    },
    {
      "epoch": 2.56,
      "learning_rate": 3.722202402188132e-05,
      "loss": 2.8506,
      "step": 21500
    },
    {
      "epoch": 2.62,
      "learning_rate": 3.692472351052444e-05,
      "loss": 2.8443,
      "step": 22000
    },
    {
      "epoch": 2.68,
      "learning_rate": 3.6628017600190276e-05,
      "loss": 2.8417,
      "step": 22500
    },
    {
      "epoch": 2.74,
      "learning_rate": 3.6330717088833394e-05,
      "loss": 2.8418,
      "step": 23000
    },
    {
      "epoch": 2.79,
      "learning_rate": 3.603341657747651e-05,
      "loss": 2.8462,
      "step": 23500
    },
    {
      "epoch": 2.85,
      "learning_rate": 3.573611606611964e-05,
      "loss": 2.8516,
      "step": 24000
    },
    {
      "epoch": 2.91,
      "learning_rate": 3.5438815554762756e-05,
      "loss": 2.8411,
      "step": 24500
    },
    {
      "epoch": 2.97,
      "learning_rate": 3.5141515043405875e-05,
      "loss": 2.8285,
      "step": 25000
    },
    {
      "epoch": 3.03,
      "learning_rate": 3.4844214532049e-05,
      "loss": 2.8064,
      "step": 25500
    },
    {
      "epoch": 3.09,
      "learning_rate": 3.454691402069212e-05,
      "loss": 2.7888,
      "step": 26000
    },
    {
      "epoch": 3.15,
      "learning_rate": 3.425020811035795e-05,
      "loss": 2.7971,
      "step": 26500
    },
    {
      "epoch": 3.21,
      "learning_rate": 3.395290759900107e-05,
      "loss": 2.793,
      "step": 27000
    },
    {
      "epoch": 3.27,
      "learning_rate": 3.365560708764419e-05,
      "loss": 2.7998,
      "step": 27500
    },
    {
      "epoch": 3.33,
      "learning_rate": 3.3358306576287314e-05,
      "loss": 2.7935,
      "step": 28000
    },
    {
      "epoch": 3.39,
      "learning_rate": 3.306160066595315e-05,
      "loss": 2.8,
      "step": 28500
    },
    {
      "epoch": 3.45,
      "learning_rate": 3.2764300154596267e-05,
      "loss": 2.7915,
      "step": 29000
    },
    {
      "epoch": 3.51,
      "learning_rate": 3.24675942442621e-05,
      "loss": 2.7924,
      "step": 29500
    },
    {
      "epoch": 3.57,
      "learning_rate": 3.217029373290522e-05,
      "loss": 2.7915,
      "step": 30000
    },
    {
      "epoch": 3.63,
      "learning_rate": 3.1872993221548344e-05,
      "loss": 2.7827,
      "step": 30500
    },
    {
      "epoch": 3.69,
      "learning_rate": 3.157569271019146e-05,
      "loss": 2.7958,
      "step": 31000
    },
    {
      "epoch": 3.75,
      "learning_rate": 3.127839219883458e-05,
      "loss": 2.7905,
      "step": 31500
    },
    {
      "epoch": 3.81,
      "learning_rate": 3.0981686288500415e-05,
      "loss": 2.778,
      "step": 32000
    },
    {
      "epoch": 3.86,
      "learning_rate": 3.068438577714354e-05,
      "loss": 2.7901,
      "step": 32500
    },
    {
      "epoch": 3.92,
      "learning_rate": 3.038708526578666e-05,
      "loss": 2.7851,
      "step": 33000
    },
    {
      "epoch": 3.98,
      "learning_rate": 3.0089784754429777e-05,
      "loss": 2.7936,
      "step": 33500
    },
    {
      "epoch": 4.04,
      "learning_rate": 2.97924842430729e-05,
      "loss": 2.765,
      "step": 34000
    },
    {
      "epoch": 4.1,
      "learning_rate": 2.9495778332738732e-05,
      "loss": 2.7393,
      "step": 34500
    },
    {
      "epoch": 4.16,
      "learning_rate": 2.9198477821381854e-05,
      "loss": 2.7426,
      "step": 35000
    },
    {
      "epoch": 4.22,
      "learning_rate": 2.8901177310024973e-05,
      "loss": 2.7467,
      "step": 35500
    },
    {
      "epoch": 4.28,
      "learning_rate": 2.8603876798668094e-05,
      "loss": 2.7458,
      "step": 36000
    },
    {
      "epoch": 4.34,
      "learning_rate": 2.8306576287311216e-05,
      "loss": 2.7524,
      "step": 36500
    },
    {
      "epoch": 4.4,
      "learning_rate": 2.8009275775954335e-05,
      "loss": 2.7428,
      "step": 37000
    },
    {
      "epoch": 4.46,
      "learning_rate": 2.7711975264597456e-05,
      "loss": 2.7437,
      "step": 37500
    },
    {
      "epoch": 4.52,
      "learning_rate": 2.7414674753240578e-05,
      "loss": 2.7496,
      "step": 38000
    },
    {
      "epoch": 4.58,
      "learning_rate": 2.7117968842906412e-05,
      "loss": 2.7503,
      "step": 38500
    },
    {
      "epoch": 4.64,
      "learning_rate": 2.682066833154953e-05,
      "loss": 2.7481,
      "step": 39000
    },
    {
      "epoch": 4.7,
      "learning_rate": 2.6523962421215365e-05,
      "loss": 2.7446,
      "step": 39500
    },
    {
      "epoch": 4.76,
      "learning_rate": 2.6226661909858486e-05,
      "loss": 2.7428,
      "step": 40000
    },
    {
      "epoch": 4.82,
      "learning_rate": 2.5929361398501605e-05,
      "loss": 2.7429,
      "step": 40500
    },
    {
      "epoch": 4.88,
      "learning_rate": 2.5632060887144726e-05,
      "loss": 2.7472,
      "step": 41000
    },
    {
      "epoch": 4.94,
      "learning_rate": 2.5334760375787848e-05,
      "loss": 2.7497,
      "step": 41500
    },
    {
      "epoch": 4.99,
      "learning_rate": 2.5037459864430967e-05,
      "loss": 2.7491,
      "step": 42000
    },
    {
      "epoch": 5.05,
      "learning_rate": 2.474015935307409e-05,
      "loss": 2.7089,
      "step": 42500
    },
    {
      "epoch": 5.11,
      "learning_rate": 2.444285884171721e-05,
      "loss": 2.7098,
      "step": 43000
    },
    {
      "epoch": 5.17,
      "learning_rate": 2.4146152931383044e-05,
      "loss": 2.71,
      "step": 43500
    },
    {
      "epoch": 5.23,
      "learning_rate": 2.3848852420026163e-05,
      "loss": 2.7139,
      "step": 44000
    },
    {
      "epoch": 5.29,
      "learning_rate": 2.3551551908669284e-05,
      "loss": 2.7187,
      "step": 44500
    },
    {
      "epoch": 5.35,
      "learning_rate": 2.3254251397312403e-05,
      "loss": 2.7163,
      "step": 45000
    },
    {
      "epoch": 5.41,
      "learning_rate": 2.2956950885955525e-05,
      "loss": 2.7097,
      "step": 45500
    },
    {
      "epoch": 5.47,
      "learning_rate": 2.266024497562136e-05,
      "loss": 2.7085,
      "step": 46000
    },
    {
      "epoch": 5.53,
      "learning_rate": 2.236294446426448e-05,
      "loss": 2.7001,
      "step": 46500
    },
    {
      "epoch": 5.59,
      "learning_rate": 2.20656439529076e-05,
      "loss": 2.7102,
      "step": 47000
    },
    {
      "epoch": 5.65,
      "learning_rate": 2.176834344155072e-05,
      "loss": 2.7191,
      "step": 47500
    },
    {
      "epoch": 5.71,
      "learning_rate": 2.1471637531216554e-05,
      "loss": 2.7124,
      "step": 48000
    },
    {
      "epoch": 5.77,
      "learning_rate": 2.1174337019859676e-05,
      "loss": 2.7145,
      "step": 48500
    },
    {
      "epoch": 5.83,
      "learning_rate": 2.0877631109525507e-05,
      "loss": 2.7162,
      "step": 49000
    },
    {
      "epoch": 5.89,
      "learning_rate": 2.058033059816863e-05,
      "loss": 2.7112,
      "step": 49500
    },
    {
      "epoch": 5.95,
      "learning_rate": 2.028303008681175e-05,
      "loss": 2.7122,
      "step": 50000
    },
    {
      "epoch": 6.01,
      "learning_rate": 1.998572957545487e-05,
      "loss": 2.7143,
      "step": 50500
    },
    {
      "epoch": 6.06,
      "learning_rate": 1.968842906409799e-05,
      "loss": 2.6759,
      "step": 51000
    },
    {
      "epoch": 6.12,
      "learning_rate": 1.9391128552741112e-05,
      "loss": 2.6826,
      "step": 51500
    },
    {
      "epoch": 6.18,
      "learning_rate": 1.909382804138423e-05,
      "loss": 2.6815,
      "step": 52000
    },
    {
      "epoch": 6.24,
      "learning_rate": 1.8796527530027352e-05,
      "loss": 2.6825,
      "step": 52500
    },
    {
      "epoch": 6.3,
      "learning_rate": 1.8499227018670474e-05,
      "loss": 2.678,
      "step": 53000
    },
    {
      "epoch": 6.36,
      "learning_rate": 1.8202521108336308e-05,
      "loss": 2.6829,
      "step": 53500
    },
    {
      "epoch": 6.42,
      "learning_rate": 1.7905220596979427e-05,
      "loss": 2.6912,
      "step": 54000
    },
    {
      "epoch": 6.48,
      "learning_rate": 1.760792008562255e-05,
      "loss": 2.6928,
      "step": 54500
    },
    {
      "epoch": 6.54,
      "learning_rate": 1.731061957426567e-05,
      "loss": 2.6916,
      "step": 55000
    },
    {
      "epoch": 6.6,
      "learning_rate": 1.70139136639315e-05,
      "loss": 2.6829,
      "step": 55500
    },
    {
      "epoch": 6.66,
      "learning_rate": 1.6716613152574623e-05,
      "loss": 2.6827,
      "step": 56000
    },
    {
      "epoch": 6.72,
      "learning_rate": 1.6419312641217744e-05,
      "loss": 2.684,
      "step": 56500
    },
    {
      "epoch": 6.78,
      "learning_rate": 1.6122012129860863e-05,
      "loss": 2.6825,
      "step": 57000
    },
    {
      "epoch": 6.84,
      "learning_rate": 1.5824711618503984e-05,
      "loss": 2.6761,
      "step": 57500
    },
    {
      "epoch": 6.9,
      "learning_rate": 1.552800570816982e-05,
      "loss": 2.6913,
      "step": 58000
    },
    {
      "epoch": 6.96,
      "learning_rate": 1.5230705196812939e-05,
      "loss": 2.6902,
      "step": 58500
    },
    {
      "epoch": 7.02,
      "learning_rate": 1.4933404685456059e-05,
      "loss": 2.6832,
      "step": 59000
    },
    {
      "epoch": 7.08,
      "learning_rate": 1.463610417409918e-05,
      "loss": 2.6619,
      "step": 59500
    },
    {
      "epoch": 7.14,
      "learning_rate": 1.4339398263765014e-05,
      "loss": 2.6594,
      "step": 60000
    },
    {
      "epoch": 7.19,
      "learning_rate": 1.4042097752408134e-05,
      "loss": 2.6659,
      "step": 60500
    },
    {
      "epoch": 7.25,
      "learning_rate": 1.3745391842073968e-05,
      "loss": 2.6654,
      "step": 61000
    },
    {
      "epoch": 7.31,
      "learning_rate": 1.3448091330717088e-05,
      "loss": 2.6676,
      "step": 61500
    },
    {
      "epoch": 7.37,
      "learning_rate": 1.3150790819360209e-05,
      "loss": 2.6582,
      "step": 62000
    },
    {
      "epoch": 7.43,
      "learning_rate": 1.285349030800333e-05,
      "loss": 2.6585,
      "step": 62500
    },
    {
      "epoch": 7.49,
      "learning_rate": 1.255618979664645e-05,
      "loss": 2.6636,
      "step": 63000
    },
    {
      "epoch": 7.55,
      "learning_rate": 1.225888928528957e-05,
      "loss": 2.6627,
      "step": 63500
    },
    {
      "epoch": 7.61,
      "learning_rate": 1.1961588773932692e-05,
      "loss": 2.656,
      "step": 64000
    },
    {
      "epoch": 7.67,
      "learning_rate": 1.1664288262575812e-05,
      "loss": 2.6634,
      "step": 64500
    },
    {
      "epoch": 7.73,
      "learning_rate": 1.1366987751218933e-05,
      "loss": 2.6616,
      "step": 65000
    },
    {
      "epoch": 7.79,
      "learning_rate": 1.1070281840884766e-05,
      "loss": 2.6614,
      "step": 65500
    },
    {
      "epoch": 7.85,
      "learning_rate": 1.0772981329527887e-05,
      "loss": 2.6709,
      "step": 66000
    },
    {
      "epoch": 7.91,
      "learning_rate": 1.0475680818171008e-05,
      "loss": 2.6681,
      "step": 66500
    },
    {
      "epoch": 7.97,
      "learning_rate": 1.0178380306814128e-05,
      "loss": 2.6731,
      "step": 67000
    },
    {
      "epoch": 8.03,
      "learning_rate": 9.881674396479962e-06,
      "loss": 2.6451,
      "step": 67500
    },
    {
      "epoch": 8.09,
      "learning_rate": 9.584968486145796e-06,
      "loss": 2.6538,
      "step": 68000
    },
    {
      "epoch": 8.15,
      "learning_rate": 9.287667974788918e-06,
      "loss": 2.6463,
      "step": 68500
    },
    {
      "epoch": 8.21,
      "learning_rate": 8.990367463432038e-06,
      "loss": 2.6519,
      "step": 69000
    },
    {
      "epoch": 8.26,
      "learning_rate": 8.693066952075157e-06,
      "loss": 2.6401,
      "step": 69500
    },
    {
      "epoch": 8.32,
      "learning_rate": 8.395766440718278e-06,
      "loss": 2.6448,
      "step": 70000
    },
    {
      "epoch": 8.38,
      "learning_rate": 8.098465929361398e-06,
      "loss": 2.6482,
      "step": 70500
    },
    {
      "epoch": 8.44,
      "learning_rate": 7.801165418004519e-06,
      "loss": 2.6444,
      "step": 71000
    },
    {
      "epoch": 8.5,
      "learning_rate": 7.5038649066476395e-06,
      "loss": 2.6438,
      "step": 71500
    },
    {
      "epoch": 8.56,
      "learning_rate": 7.2065643952907604e-06,
      "loss": 2.6448,
      "step": 72000
    },
    {
      "epoch": 8.62,
      "learning_rate": 6.9098584849565936e-06,
      "loss": 2.6443,
      "step": 72500
    },
    {
      "epoch": 8.68,
      "learning_rate": 6.6125579735997145e-06,
      "loss": 2.6521,
      "step": 73000
    },
    {
      "epoch": 8.74,
      "learning_rate": 6.3152574622428354e-06,
      "loss": 2.6491,
      "step": 73500
    },
    {
      "epoch": 8.8,
      "learning_rate": 6.018551551908669e-06,
      "loss": 2.6341,
      "step": 74000
    },
    {
      "epoch": 8.86,
      "learning_rate": 5.72125104055179e-06,
      "loss": 2.6593,
      "step": 74500
    },
    {
      "epoch": 8.92,
      "learning_rate": 5.42395052919491e-06,
      "loss": 2.6501,
      "step": 75000
    },
    {
      "epoch": 8.98,
      "learning_rate": 5.126650017838031e-06,
      "loss": 2.6496,
      "step": 75500
    },
    {
      "epoch": 9.04,
      "learning_rate": 4.8293495064811514e-06,
      "loss": 2.6418,
      "step": 76000
    },
    {
      "epoch": 9.1,
      "learning_rate": 4.5320489951242715e-06,
      "loss": 2.6389,
      "step": 76500
    },
    {
      "epoch": 9.16,
      "learning_rate": 4.2347484837673925e-06,
      "loss": 2.6309,
      "step": 77000
    },
    {
      "epoch": 9.22,
      "learning_rate": 3.9374479724105126e-06,
      "loss": 2.6315,
      "step": 77500
    },
    {
      "epoch": 9.28,
      "learning_rate": 3.6407420620763465e-06,
      "loss": 2.6371,
      "step": 78000
    },
    {
      "epoch": 9.34,
      "learning_rate": 3.343441550719467e-06,
      "loss": 2.6414,
      "step": 78500
    },
    {
      "epoch": 9.39,
      "learning_rate": 3.046141039362588e-06,
      "loss": 2.6361,
      "step": 79000
    },
    {
      "epoch": 9.45,
      "learning_rate": 2.7488405280057085e-06,
      "loss": 2.632,
      "step": 79500
    },
    {
      "epoch": 9.51,
      "learning_rate": 2.4521346176715424e-06,
      "loss": 2.6387,
      "step": 80000
    },
    {
      "epoch": 9.57,
      "learning_rate": 2.154834106314663e-06,
      "loss": 2.6371,
      "step": 80500
    },
    {
      "epoch": 9.63,
      "learning_rate": 1.8575335949577837e-06,
      "loss": 2.6383,
      "step": 81000
    },
    {
      "epoch": 9.69,
      "learning_rate": 1.5602330836009038e-06,
      "loss": 2.6379,
      "step": 81500
    },
    {
      "epoch": 9.75,
      "learning_rate": 1.2629325722440245e-06,
      "loss": 2.6272,
      "step": 82000
    },
    {
      "epoch": 9.81,
      "learning_rate": 9.668212629325723e-07,
      "loss": 2.6353,
      "step": 82500
    },
    {
      "epoch": 9.87,
      "learning_rate": 6.695207515756928e-07,
      "loss": 2.6404,
      "step": 83000
    },
    {
      "epoch": 9.93,
      "learning_rate": 3.722202402188132e-07,
      "loss": 2.6373,
      "step": 83500
    },
    {
      "epoch": 9.99,
      "learning_rate": 7.491972886193364e-08,
      "loss": 2.6304,
      "step": 84000
    },
    {
      "epoch": 10.0,
      "step": 84090,
      "total_flos": 2.5714385279778816e+17,
      "train_runtime": 27131.1773,
      "train_samples_per_second": 3.099
    }
  ],
  "max_steps": 84090,
  "num_train_epochs": 10,
  "total_flos": 2.5714385279778816e+17,
  "trial_name": null,
  "trial_params": null
}