{ "best_metric": null, "best_model_checkpoint": null, "epoch": 5.930971148695186, "global_step": 310000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.01, "learning_rate": 9.98e-07, "loss": 1.6418, "step": 500 }, { "epoch": 0.02, "learning_rate": 1.998e-06, "loss": 1.6387, "step": 1000 }, { "epoch": 0.03, "learning_rate": 2.9980000000000003e-06, "loss": 1.6386, "step": 1500 }, { "epoch": 0.04, "learning_rate": 3.9980000000000005e-06, "loss": 1.6357, "step": 2000 }, { "epoch": 0.05, "learning_rate": 4.998e-06, "loss": 1.633, "step": 2500 }, { "epoch": 0.06, "learning_rate": 5.996000000000001e-06, "loss": 1.6367, "step": 3000 }, { "epoch": 0.07, "learning_rate": 6.9960000000000004e-06, "loss": 1.6346, "step": 3500 }, { "epoch": 0.08, "learning_rate": 7.996000000000001e-06, "loss": 1.6368, "step": 4000 }, { "epoch": 0.09, "learning_rate": 8.996e-06, "loss": 1.6311, "step": 4500 }, { "epoch": 0.1, "learning_rate": 9.994000000000001e-06, "loss": 1.6388, "step": 5000 }, { "epoch": 0.11, "learning_rate": 9.990399474578892e-06, "loss": 1.6331, "step": 5500 }, { "epoch": 0.11, "learning_rate": 9.98074099830011e-06, "loss": 1.6372, "step": 6000 }, { "epoch": 0.12, "learning_rate": 9.971082522021327e-06, "loss": 1.6372, "step": 6500 }, { "epoch": 0.13, "learning_rate": 9.961424045742545e-06, "loss": 1.6315, "step": 7000 }, { "epoch": 0.14, "learning_rate": 9.95178488641632e-06, "loss": 1.6344, "step": 7500 }, { "epoch": 0.15, "learning_rate": 9.942126410137538e-06, "loss": 1.638, "step": 8000 }, { "epoch": 0.16, "learning_rate": 9.932467933858756e-06, "loss": 1.6345, "step": 8500 }, { "epoch": 0.17, "learning_rate": 9.922809457579972e-06, "loss": 1.6324, "step": 9000 }, { "epoch": 0.18, "learning_rate": 9.913170298253748e-06, "loss": 1.6336, "step": 9500 }, { "epoch": 0.19, "learning_rate": 9.903511821974966e-06, "loss": 1.6334, "step": 10000 }, { "epoch": 0.2, "learning_rate": 9.893853345696184e-06, "loss": 1.6336, "step": 10500 }, { "epoch": 0.21, "learning_rate": 9.884194869417402e-06, "loss": 1.6323, "step": 11000 }, { "epoch": 0.22, "learning_rate": 9.874555710091176e-06, "loss": 1.6342, "step": 11500 }, { "epoch": 0.23, "learning_rate": 9.864897233812394e-06, "loss": 1.6329, "step": 12000 }, { "epoch": 0.24, "learning_rate": 9.855238757533612e-06, "loss": 1.6333, "step": 12500 }, { "epoch": 0.25, "learning_rate": 9.84558028125483e-06, "loss": 1.634, "step": 13000 }, { "epoch": 0.26, "learning_rate": 9.835921804976048e-06, "loss": 1.6347, "step": 13500 }, { "epoch": 0.27, "learning_rate": 9.826263328697266e-06, "loss": 1.6276, "step": 14000 }, { "epoch": 0.28, "learning_rate": 9.81662416937104e-06, "loss": 1.632, "step": 14500 }, { "epoch": 0.29, "learning_rate": 9.806965693092258e-06, "loss": 1.6322, "step": 15000 }, { "epoch": 0.3, "learning_rate": 9.797307216813476e-06, "loss": 1.6318, "step": 15500 }, { "epoch": 0.31, "learning_rate": 9.787648740534694e-06, "loss": 1.6286, "step": 16000 }, { "epoch": 0.32, "learning_rate": 9.77800958120847e-06, "loss": 1.6347, "step": 16500 }, { "epoch": 0.33, "learning_rate": 9.768351104929688e-06, "loss": 1.6333, "step": 17000 }, { "epoch": 0.33, "learning_rate": 9.758692628650906e-06, "loss": 1.6345, "step": 17500 }, { "epoch": 0.34, "learning_rate": 9.749034152372122e-06, "loss": 1.6311, "step": 18000 }, { "epoch": 0.35, "learning_rate": 9.73937567609334e-06, "loss": 1.6308, "step": 18500 }, { "epoch": 0.36, "learning_rate": 9.729736516767115e-06, "loss": 1.6299, "step": 19000 }, { "epoch": 0.37, "learning_rate": 9.720078040488333e-06, "loss": 1.6284, "step": 19500 }, { "epoch": 0.38, "learning_rate": 9.71041956420955e-06, "loss": 1.6301, "step": 20000 }, { "epoch": 0.39, "learning_rate": 9.700761087930768e-06, "loss": 1.6305, "step": 20500 }, { "epoch": 0.4, "learning_rate": 9.691121928604545e-06, "loss": 1.6309, "step": 21000 }, { "epoch": 0.41, "learning_rate": 9.681463452325762e-06, "loss": 1.6328, "step": 21500 }, { "epoch": 0.42, "learning_rate": 9.67180497604698e-06, "loss": 1.6292, "step": 22000 }, { "epoch": 0.43, "learning_rate": 9.662146499768197e-06, "loss": 1.6292, "step": 22500 }, { "epoch": 0.44, "learning_rate": 9.652488023489415e-06, "loss": 1.6295, "step": 23000 }, { "epoch": 0.45, "learning_rate": 9.64284886416319e-06, "loss": 1.627, "step": 23500 }, { "epoch": 0.46, "learning_rate": 9.633190387884409e-06, "loss": 1.6281, "step": 24000 }, { "epoch": 0.47, "learning_rate": 9.623531911605627e-06, "loss": 1.6273, "step": 24500 }, { "epoch": 0.48, "learning_rate": 9.613873435326844e-06, "loss": 1.6305, "step": 25000 }, { "epoch": 0.49, "learning_rate": 9.604214959048062e-06, "loss": 1.6286, "step": 25500 }, { "epoch": 0.5, "learning_rate": 9.594575799721837e-06, "loss": 1.628, "step": 26000 }, { "epoch": 0.51, "learning_rate": 9.584917323443055e-06, "loss": 1.6279, "step": 26500 }, { "epoch": 0.52, "learning_rate": 9.575258847164273e-06, "loss": 1.6248, "step": 27000 }, { "epoch": 0.53, "learning_rate": 9.565600370885489e-06, "loss": 1.6244, "step": 27500 }, { "epoch": 0.54, "learning_rate": 9.555961211559265e-06, "loss": 1.6279, "step": 28000 }, { "epoch": 0.55, "learning_rate": 9.546302735280483e-06, "loss": 1.6269, "step": 28500 }, { "epoch": 0.55, "learning_rate": 9.536644259001701e-06, "loss": 1.6275, "step": 29000 }, { "epoch": 0.56, "learning_rate": 9.526985782722919e-06, "loss": 1.6281, "step": 29500 }, { "epoch": 0.57, "learning_rate": 9.517327306444137e-06, "loss": 1.6248, "step": 30000 }, { "epoch": 0.58, "learning_rate": 9.507668830165355e-06, "loss": 1.6251, "step": 30500 }, { "epoch": 0.59, "learning_rate": 9.498010353886573e-06, "loss": 1.6255, "step": 31000 }, { "epoch": 0.6, "learning_rate": 9.488371194560347e-06, "loss": 1.6276, "step": 31500 }, { "epoch": 0.61, "learning_rate": 9.478712718281565e-06, "loss": 1.6276, "step": 32000 }, { "epoch": 0.62, "learning_rate": 9.469054242002781e-06, "loss": 1.6269, "step": 32500 }, { "epoch": 0.63, "learning_rate": 9.459395765723999e-06, "loss": 1.6281, "step": 33000 }, { "epoch": 0.64, "learning_rate": 9.449756606397775e-06, "loss": 1.6231, "step": 33500 }, { "epoch": 0.65, "learning_rate": 9.440098130118993e-06, "loss": 1.625, "step": 34000 }, { "epoch": 0.66, "learning_rate": 9.430439653840211e-06, "loss": 1.6261, "step": 34500 }, { "epoch": 0.67, "learning_rate": 9.420781177561429e-06, "loss": 1.6254, "step": 35000 }, { "epoch": 0.68, "learning_rate": 9.411122701282647e-06, "loss": 1.6253, "step": 35500 }, { "epoch": 0.69, "learning_rate": 9.401483541956421e-06, "loss": 1.6272, "step": 36000 }, { "epoch": 0.7, "learning_rate": 9.39182506567764e-06, "loss": 1.6277, "step": 36500 }, { "epoch": 0.71, "learning_rate": 9.382166589398857e-06, "loss": 1.6239, "step": 37000 }, { "epoch": 0.72, "learning_rate": 9.372508113120075e-06, "loss": 1.6266, "step": 37500 }, { "epoch": 0.73, "learning_rate": 9.362849636841293e-06, "loss": 1.6233, "step": 38000 }, { "epoch": 0.74, "learning_rate": 9.353210477515067e-06, "loss": 1.6268, "step": 38500 }, { "epoch": 0.75, "learning_rate": 9.343552001236285e-06, "loss": 1.6226, "step": 39000 }, { "epoch": 0.76, "learning_rate": 9.333893524957503e-06, "loss": 1.6237, "step": 39500 }, { "epoch": 0.77, "learning_rate": 9.324235048678721e-06, "loss": 1.6261, "step": 40000 }, { "epoch": 0.77, "learning_rate": 9.31457657239994e-06, "loss": 1.6256, "step": 40500 }, { "epoch": 0.78, "learning_rate": 9.304937413073715e-06, "loss": 1.6223, "step": 41000 }, { "epoch": 0.79, "learning_rate": 9.295278936794933e-06, "loss": 1.6245, "step": 41500 }, { "epoch": 0.8, "learning_rate": 9.28562046051615e-06, "loss": 1.6265, "step": 42000 }, { "epoch": 0.81, "learning_rate": 9.275961984237367e-06, "loss": 1.6237, "step": 42500 }, { "epoch": 0.82, "learning_rate": 9.266322824911142e-06, "loss": 1.6212, "step": 43000 }, { "epoch": 0.83, "learning_rate": 9.25666434863236e-06, "loss": 1.6212, "step": 43500 }, { "epoch": 0.84, "learning_rate": 9.247005872353578e-06, "loss": 1.6216, "step": 44000 }, { "epoch": 0.85, "learning_rate": 9.237347396074796e-06, "loss": 1.6276, "step": 44500 }, { "epoch": 0.86, "learning_rate": 9.227688919796014e-06, "loss": 1.6307, "step": 45000 }, { "epoch": 0.87, "learning_rate": 9.21804976046979e-06, "loss": 1.6272, "step": 45500 }, { "epoch": 0.88, "learning_rate": 9.208391284191008e-06, "loss": 1.6252, "step": 46000 }, { "epoch": 0.89, "learning_rate": 9.198732807912224e-06, "loss": 1.6256, "step": 46500 }, { "epoch": 0.9, "learning_rate": 9.189074331633442e-06, "loss": 1.6262, "step": 47000 }, { "epoch": 0.91, "learning_rate": 9.17941585535466e-06, "loss": 1.6246, "step": 47500 }, { "epoch": 0.92, "learning_rate": 9.169776696028436e-06, "loss": 1.6261, "step": 48000 }, { "epoch": 0.93, "learning_rate": 9.160118219749654e-06, "loss": 1.6241, "step": 48500 }, { "epoch": 0.94, "learning_rate": 9.150459743470872e-06, "loss": 1.624, "step": 49000 }, { "epoch": 0.95, "learning_rate": 9.14080126719209e-06, "loss": 1.6239, "step": 49500 }, { "epoch": 0.96, "learning_rate": 9.131162107865864e-06, "loss": 1.6216, "step": 50000 }, { "epoch": 0.97, "learning_rate": 9.121503631587082e-06, "loss": 1.6294, "step": 50500 }, { "epoch": 0.98, "learning_rate": 9.1118451553083e-06, "loss": 1.6241, "step": 51000 }, { "epoch": 0.99, "learning_rate": 9.102186679029516e-06, "loss": 1.6193, "step": 51500 }, { "epoch": 0.99, "learning_rate": 9.092547519703292e-06, "loss": 1.6234, "step": 52000 }, { "epoch": 1.0, "learning_rate": 9.08288904342451e-06, "loss": 1.6231, "step": 52500 }, { "epoch": 1.01, "learning_rate": 9.073230567145728e-06, "loss": 1.6248, "step": 53000 }, { "epoch": 1.02, "learning_rate": 9.063572090866946e-06, "loss": 1.6273, "step": 53500 }, { "epoch": 1.03, "learning_rate": 9.053913614588164e-06, "loss": 1.6226, "step": 54000 }, { "epoch": 1.04, "learning_rate": 9.044274455261938e-06, "loss": 1.619, "step": 54500 }, { "epoch": 1.05, "learning_rate": 9.034615978983156e-06, "loss": 1.6266, "step": 55000 }, { "epoch": 1.06, "learning_rate": 9.024957502704374e-06, "loss": 1.6216, "step": 55500 }, { "epoch": 1.07, "learning_rate": 9.015299026425592e-06, "loss": 1.6224, "step": 56000 }, { "epoch": 1.08, "learning_rate": 9.005640550146808e-06, "loss": 1.6238, "step": 56500 }, { "epoch": 1.09, "learning_rate": 8.996001390820584e-06, "loss": 1.6201, "step": 57000 }, { "epoch": 1.1, "learning_rate": 8.986342914541802e-06, "loss": 1.6222, "step": 57500 }, { "epoch": 1.11, "learning_rate": 8.97668443826302e-06, "loss": 1.6233, "step": 58000 }, { "epoch": 1.12, "learning_rate": 8.967025961984238e-06, "loss": 1.6259, "step": 58500 }, { "epoch": 1.13, "learning_rate": 8.957386802658014e-06, "loss": 1.6218, "step": 59000 }, { "epoch": 1.14, "learning_rate": 8.94772832637923e-06, "loss": 1.6247, "step": 59500 }, { "epoch": 1.15, "learning_rate": 8.938069850100449e-06, "loss": 1.6239, "step": 60000 }, { "epoch": 1.16, "learning_rate": 8.928411373821666e-06, "loss": 1.621, "step": 60500 }, { "epoch": 1.17, "learning_rate": 8.918752897542884e-06, "loss": 1.6251, "step": 61000 }, { "epoch": 1.18, "learning_rate": 8.909094421264102e-06, "loss": 1.6237, "step": 61500 }, { "epoch": 1.19, "learning_rate": 8.899455261937877e-06, "loss": 1.6236, "step": 62000 }, { "epoch": 1.2, "learning_rate": 8.889796785659095e-06, "loss": 1.6223, "step": 62500 }, { "epoch": 1.21, "learning_rate": 8.880138309380313e-06, "loss": 1.6218, "step": 63000 }, { "epoch": 1.21, "learning_rate": 8.87047983310153e-06, "loss": 1.6249, "step": 63500 }, { "epoch": 1.22, "learning_rate": 8.860821356822748e-06, "loss": 1.6212, "step": 64000 }, { "epoch": 1.23, "learning_rate": 8.851162880543966e-06, "loss": 1.6219, "step": 64500 }, { "epoch": 1.24, "learning_rate": 8.841523721217742e-06, "loss": 1.6229, "step": 65000 }, { "epoch": 1.25, "learning_rate": 8.83186524493896e-06, "loss": 1.6214, "step": 65500 }, { "epoch": 1.26, "learning_rate": 8.822206768660177e-06, "loss": 1.6218, "step": 66000 }, { "epoch": 1.27, "learning_rate": 8.812548292381395e-06, "loss": 1.6271, "step": 66500 }, { "epoch": 1.28, "learning_rate": 8.802909133055169e-06, "loss": 1.6239, "step": 67000 }, { "epoch": 1.29, "learning_rate": 8.793250656776387e-06, "loss": 1.6238, "step": 67500 }, { "epoch": 1.3, "learning_rate": 8.783592180497605e-06, "loss": 1.62, "step": 68000 }, { "epoch": 1.31, "learning_rate": 8.773933704218823e-06, "loss": 1.6235, "step": 68500 }, { "epoch": 1.32, "learning_rate": 8.764294544892599e-06, "loss": 1.6227, "step": 69000 }, { "epoch": 1.33, "learning_rate": 8.754636068613817e-06, "loss": 1.6222, "step": 69500 }, { "epoch": 1.34, "learning_rate": 8.744977592335035e-06, "loss": 1.6215, "step": 70000 }, { "epoch": 1.35, "learning_rate": 8.735319116056251e-06, "loss": 1.624, "step": 70500 }, { "epoch": 1.36, "learning_rate": 8.725660639777469e-06, "loss": 1.6271, "step": 71000 }, { "epoch": 1.37, "learning_rate": 8.716002163498687e-06, "loss": 1.6218, "step": 71500 }, { "epoch": 1.38, "learning_rate": 8.706363004172463e-06, "loss": 1.6211, "step": 72000 }, { "epoch": 1.39, "learning_rate": 8.696704527893681e-06, "loss": 1.6274, "step": 72500 }, { "epoch": 1.4, "learning_rate": 8.687046051614899e-06, "loss": 1.624, "step": 73000 }, { "epoch": 1.41, "learning_rate": 8.677387575336117e-06, "loss": 1.624, "step": 73500 }, { "epoch": 1.42, "learning_rate": 8.667729099057333e-06, "loss": 1.6234, "step": 74000 }, { "epoch": 1.43, "learning_rate": 8.658089939731109e-06, "loss": 1.6189, "step": 74500 }, { "epoch": 1.43, "learning_rate": 8.648431463452327e-06, "loss": 1.6221, "step": 75000 }, { "epoch": 1.44, "learning_rate": 8.638772987173543e-06, "loss": 1.6224, "step": 75500 }, { "epoch": 1.45, "learning_rate": 8.629114510894761e-06, "loss": 1.6212, "step": 76000 }, { "epoch": 1.46, "learning_rate": 8.619456034615979e-06, "loss": 1.6189, "step": 76500 }, { "epoch": 1.47, "learning_rate": 8.609816875289755e-06, "loss": 1.6169, "step": 77000 }, { "epoch": 1.48, "learning_rate": 8.600158399010973e-06, "loss": 1.6216, "step": 77500 }, { "epoch": 1.49, "learning_rate": 8.590499922732191e-06, "loss": 1.6229, "step": 78000 }, { "epoch": 1.5, "learning_rate": 8.580841446453409e-06, "loss": 1.6219, "step": 78500 }, { "epoch": 1.51, "learning_rate": 8.571202287127183e-06, "loss": 1.6243, "step": 79000 }, { "epoch": 1.52, "learning_rate": 8.561543810848401e-06, "loss": 1.6197, "step": 79500 }, { "epoch": 1.53, "learning_rate": 8.55188533456962e-06, "loss": 1.6199, "step": 80000 }, { "epoch": 1.54, "learning_rate": 8.542226858290835e-06, "loss": 1.6204, "step": 80500 }, { "epoch": 1.55, "learning_rate": 8.532568382012053e-06, "loss": 1.6242, "step": 81000 }, { "epoch": 1.56, "learning_rate": 8.522909905733271e-06, "loss": 1.6201, "step": 81500 }, { "epoch": 1.57, "learning_rate": 8.513270746407047e-06, "loss": 1.6237, "step": 82000 }, { "epoch": 1.58, "learning_rate": 8.503612270128265e-06, "loss": 1.6191, "step": 82500 }, { "epoch": 1.59, "learning_rate": 8.493953793849483e-06, "loss": 1.6215, "step": 83000 }, { "epoch": 1.6, "learning_rate": 8.484295317570701e-06, "loss": 1.6233, "step": 83500 }, { "epoch": 1.61, "learning_rate": 8.474656158244476e-06, "loss": 1.6219, "step": 84000 }, { "epoch": 1.62, "learning_rate": 8.464997681965694e-06, "loss": 1.6202, "step": 84500 }, { "epoch": 1.63, "learning_rate": 8.455339205686912e-06, "loss": 1.6207, "step": 85000 }, { "epoch": 1.64, "learning_rate": 8.44568072940813e-06, "loss": 1.6219, "step": 85500 }, { "epoch": 1.65, "learning_rate": 8.436022253129347e-06, "loss": 1.6233, "step": 86000 }, { "epoch": 1.65, "learning_rate": 8.426383093803122e-06, "loss": 1.6221, "step": 86500 }, { "epoch": 1.66, "learning_rate": 8.41672461752434e-06, "loss": 1.6177, "step": 87000 }, { "epoch": 1.67, "learning_rate": 8.407066141245558e-06, "loss": 1.6232, "step": 87500 }, { "epoch": 1.68, "learning_rate": 8.397407664966776e-06, "loss": 1.6236, "step": 88000 }, { "epoch": 1.69, "learning_rate": 8.387749188687994e-06, "loss": 1.6204, "step": 88500 }, { "epoch": 1.7, "learning_rate": 8.378090712409211e-06, "loss": 1.6189, "step": 89000 }, { "epoch": 1.71, "learning_rate": 8.368451553082988e-06, "loss": 1.6251, "step": 89500 }, { "epoch": 1.72, "learning_rate": 8.358793076804204e-06, "loss": 1.6237, "step": 90000 }, { "epoch": 1.73, "learning_rate": 8.349134600525422e-06, "loss": 1.6214, "step": 90500 }, { "epoch": 1.74, "learning_rate": 8.33947612424664e-06, "loss": 1.6232, "step": 91000 }, { "epoch": 1.75, "learning_rate": 8.329817647967858e-06, "loss": 1.6204, "step": 91500 }, { "epoch": 1.76, "learning_rate": 8.320159171689075e-06, "loss": 1.6224, "step": 92000 }, { "epoch": 1.77, "learning_rate": 8.31052001236285e-06, "loss": 1.6234, "step": 92500 }, { "epoch": 1.78, "learning_rate": 8.300861536084068e-06, "loss": 1.6222, "step": 93000 }, { "epoch": 1.79, "learning_rate": 8.291203059805286e-06, "loss": 1.6242, "step": 93500 }, { "epoch": 1.8, "learning_rate": 8.281544583526504e-06, "loss": 1.6213, "step": 94000 }, { "epoch": 1.81, "learning_rate": 8.271905424200278e-06, "loss": 1.6185, "step": 94500 }, { "epoch": 1.82, "learning_rate": 8.262246947921496e-06, "loss": 1.6207, "step": 95000 }, { "epoch": 1.83, "learning_rate": 8.252588471642714e-06, "loss": 1.6218, "step": 95500 }, { "epoch": 1.84, "learning_rate": 8.242929995363932e-06, "loss": 1.6242, "step": 96000 }, { "epoch": 1.85, "learning_rate": 8.23327151908515e-06, "loss": 1.6207, "step": 96500 }, { "epoch": 1.86, "learning_rate": 8.223632359758926e-06, "loss": 1.6217, "step": 97000 }, { "epoch": 1.87, "learning_rate": 8.213973883480142e-06, "loss": 1.6192, "step": 97500 }, { "epoch": 1.87, "learning_rate": 8.20431540720136e-06, "loss": 1.6162, "step": 98000 }, { "epoch": 1.88, "learning_rate": 8.194656930922578e-06, "loss": 1.6201, "step": 98500 }, { "epoch": 1.89, "learning_rate": 8.185017771596354e-06, "loss": 1.621, "step": 99000 }, { "epoch": 1.9, "learning_rate": 8.17535929531757e-06, "loss": 1.6224, "step": 99500 }, { "epoch": 1.91, "learning_rate": 8.165700819038788e-06, "loss": 1.6233, "step": 100000 }, { "epoch": 1.92, "learning_rate": 8.156042342760006e-06, "loss": 1.6191, "step": 100500 }, { "epoch": 1.93, "learning_rate": 8.146383866481224e-06, "loss": 1.6213, "step": 101000 }, { "epoch": 1.94, "learning_rate": 8.136725390202442e-06, "loss": 1.6252, "step": 101500 }, { "epoch": 1.95, "learning_rate": 8.127086230876218e-06, "loss": 1.6195, "step": 102000 }, { "epoch": 1.96, "learning_rate": 8.117427754597436e-06, "loss": 1.6176, "step": 102500 }, { "epoch": 1.97, "learning_rate": 8.107769278318654e-06, "loss": 1.6205, "step": 103000 }, { "epoch": 1.98, "learning_rate": 8.098110802039872e-06, "loss": 1.6205, "step": 103500 }, { "epoch": 1.99, "learning_rate": 8.088452325761088e-06, "loss": 1.6224, "step": 104000 }, { "epoch": 2.0, "learning_rate": 8.078813166434863e-06, "loss": 1.6204, "step": 104500 }, { "epoch": 2.01, "learning_rate": 8.06915469015608e-06, "loss": 1.6198, "step": 105000 }, { "epoch": 2.02, "learning_rate": 8.059496213877298e-06, "loss": 1.6217, "step": 105500 }, { "epoch": 2.03, "learning_rate": 8.049837737598516e-06, "loss": 1.6196, "step": 106000 }, { "epoch": 2.04, "learning_rate": 8.040179261319734e-06, "loss": 1.6195, "step": 106500 }, { "epoch": 2.05, "learning_rate": 8.03054010199351e-06, "loss": 1.6219, "step": 107000 }, { "epoch": 2.06, "learning_rate": 8.020881625714728e-06, "loss": 1.6221, "step": 107500 }, { "epoch": 2.07, "learning_rate": 8.011223149435946e-06, "loss": 1.6193, "step": 108000 }, { "epoch": 2.08, "learning_rate": 8.001564673157164e-06, "loss": 1.6199, "step": 108500 }, { "epoch": 2.09, "learning_rate": 7.99190619687838e-06, "loss": 1.622, "step": 109000 }, { "epoch": 2.09, "learning_rate": 7.982267037552157e-06, "loss": 1.6222, "step": 109500 }, { "epoch": 2.1, "learning_rate": 7.972608561273375e-06, "loss": 1.6199, "step": 110000 }, { "epoch": 2.11, "learning_rate": 7.962950084994592e-06, "loss": 1.6212, "step": 110500 }, { "epoch": 2.12, "learning_rate": 7.95329160871581e-06, "loss": 1.6177, "step": 111000 }, { "epoch": 2.13, "learning_rate": 7.943633132437028e-06, "loss": 1.6214, "step": 111500 }, { "epoch": 2.14, "learning_rate": 7.933974656158245e-06, "loss": 1.6179, "step": 112000 }, { "epoch": 2.15, "learning_rate": 7.92433549683202e-06, "loss": 1.6199, "step": 112500 }, { "epoch": 2.16, "learning_rate": 7.914677020553239e-06, "loss": 1.6212, "step": 113000 }, { "epoch": 2.17, "learning_rate": 7.905018544274455e-06, "loss": 1.6208, "step": 113500 }, { "epoch": 2.18, "learning_rate": 7.895360067995673e-06, "loss": 1.6182, "step": 114000 }, { "epoch": 2.19, "learning_rate": 7.885720908669449e-06, "loss": 1.6159, "step": 114500 }, { "epoch": 2.2, "learning_rate": 7.876062432390667e-06, "loss": 1.6184, "step": 115000 }, { "epoch": 2.21, "learning_rate": 7.866403956111885e-06, "loss": 1.6225, "step": 115500 }, { "epoch": 2.22, "learning_rate": 7.856745479833103e-06, "loss": 1.6181, "step": 116000 }, { "epoch": 2.23, "learning_rate": 7.84708700355432e-06, "loss": 1.6179, "step": 116500 }, { "epoch": 2.24, "learning_rate": 7.837428527275539e-06, "loss": 1.6224, "step": 117000 }, { "epoch": 2.25, "learning_rate": 7.827789367949313e-06, "loss": 1.617, "step": 117500 }, { "epoch": 2.26, "learning_rate": 7.818130891670531e-06, "loss": 1.6193, "step": 118000 }, { "epoch": 2.27, "learning_rate": 7.808472415391749e-06, "loss": 1.617, "step": 118500 }, { "epoch": 2.28, "learning_rate": 7.798813939112965e-06, "loss": 1.618, "step": 119000 }, { "epoch": 2.29, "learning_rate": 7.789174779786741e-06, "loss": 1.6211, "step": 119500 }, { "epoch": 2.3, "learning_rate": 7.779516303507959e-06, "loss": 1.6171, "step": 120000 }, { "epoch": 2.31, "learning_rate": 7.769857827229177e-06, "loss": 1.619, "step": 120500 }, { "epoch": 2.31, "learning_rate": 7.760199350950395e-06, "loss": 1.6201, "step": 121000 }, { "epoch": 2.32, "learning_rate": 7.75056019162417e-06, "loss": 1.6184, "step": 121500 }, { "epoch": 2.33, "learning_rate": 7.740901715345387e-06, "loss": 1.6187, "step": 122000 }, { "epoch": 2.34, "learning_rate": 7.731243239066605e-06, "loss": 1.6174, "step": 122500 }, { "epoch": 2.35, "learning_rate": 7.721584762787823e-06, "loss": 1.6184, "step": 123000 }, { "epoch": 2.36, "learning_rate": 7.711926286509041e-06, "loss": 1.617, "step": 123500 }, { "epoch": 2.37, "learning_rate": 7.702287127182815e-06, "loss": 1.6202, "step": 124000 }, { "epoch": 2.38, "learning_rate": 7.692628650904033e-06, "loss": 1.6212, "step": 124500 }, { "epoch": 2.39, "learning_rate": 7.682970174625251e-06, "loss": 1.6175, "step": 125000 }, { "epoch": 2.4, "learning_rate": 7.67331169834647e-06, "loss": 1.6202, "step": 125500 }, { "epoch": 2.41, "learning_rate": 7.663653222067687e-06, "loss": 1.6202, "step": 126000 }, { "epoch": 2.42, "learning_rate": 7.654014062741463e-06, "loss": 1.6184, "step": 126500 }, { "epoch": 2.43, "learning_rate": 7.644355586462681e-06, "loss": 1.617, "step": 127000 }, { "epoch": 2.44, "learning_rate": 7.634697110183899e-06, "loss": 1.6186, "step": 127500 }, { "epoch": 2.45, "learning_rate": 7.625038633905115e-06, "loss": 1.6184, "step": 128000 }, { "epoch": 2.46, "learning_rate": 7.615399474578891e-06, "loss": 1.6176, "step": 128500 }, { "epoch": 2.47, "learning_rate": 7.605740998300109e-06, "loss": 1.6201, "step": 129000 }, { "epoch": 2.48, "learning_rate": 7.5960825220213265e-06, "loss": 1.619, "step": 129500 }, { "epoch": 2.49, "learning_rate": 7.5864240457425444e-06, "loss": 1.6237, "step": 130000 }, { "epoch": 2.5, "learning_rate": 7.576765569463762e-06, "loss": 1.6201, "step": 130500 }, { "epoch": 2.51, "learning_rate": 7.56710709318498e-06, "loss": 1.6199, "step": 131000 }, { "epoch": 2.52, "learning_rate": 7.5574679338587556e-06, "loss": 1.6186, "step": 131500 }, { "epoch": 2.53, "learning_rate": 7.5478094575799735e-06, "loss": 1.6196, "step": 132000 }, { "epoch": 2.54, "learning_rate": 7.538150981301191e-06, "loss": 1.6209, "step": 132500 }, { "epoch": 2.54, "learning_rate": 7.528492505022408e-06, "loss": 1.618, "step": 133000 }, { "epoch": 2.55, "learning_rate": 7.5188340287436256e-06, "loss": 1.616, "step": 133500 }, { "epoch": 2.56, "learning_rate": 7.5091755524648435e-06, "loss": 1.6182, "step": 134000 }, { "epoch": 2.57, "learning_rate": 7.499536393138619e-06, "loss": 1.6191, "step": 134500 }, { "epoch": 2.58, "learning_rate": 7.489877916859837e-06, "loss": 1.6191, "step": 135000 }, { "epoch": 2.59, "learning_rate": 7.480219440581055e-06, "loss": 1.6152, "step": 135500 }, { "epoch": 2.6, "learning_rate": 7.4705609643022726e-06, "loss": 1.6199, "step": 136000 }, { "epoch": 2.61, "learning_rate": 7.460921804976048e-06, "loss": 1.6196, "step": 136500 }, { "epoch": 2.62, "learning_rate": 7.451263328697266e-06, "loss": 1.619, "step": 137000 }, { "epoch": 2.63, "learning_rate": 7.441604852418483e-06, "loss": 1.6162, "step": 137500 }, { "epoch": 2.64, "learning_rate": 7.431946376139701e-06, "loss": 1.617, "step": 138000 }, { "epoch": 2.65, "learning_rate": 7.422287899860918e-06, "loss": 1.6192, "step": 138500 }, { "epoch": 2.66, "learning_rate": 7.412648740534694e-06, "loss": 1.6184, "step": 139000 }, { "epoch": 2.67, "learning_rate": 7.402990264255912e-06, "loss": 1.6161, "step": 139500 }, { "epoch": 2.68, "learning_rate": 7.393331787977129e-06, "loss": 1.6179, "step": 140000 }, { "epoch": 2.69, "learning_rate": 7.383673311698347e-06, "loss": 1.616, "step": 140500 }, { "epoch": 2.7, "learning_rate": 7.374034152372123e-06, "loss": 1.6199, "step": 141000 }, { "epoch": 2.71, "learning_rate": 7.36437567609334e-06, "loss": 1.621, "step": 141500 }, { "epoch": 2.72, "learning_rate": 7.354717199814558e-06, "loss": 1.6177, "step": 142000 }, { "epoch": 2.73, "learning_rate": 7.345058723535775e-06, "loss": 1.619, "step": 142500 }, { "epoch": 2.74, "learning_rate": 7.33541956420955e-06, "loss": 1.6187, "step": 143000 }, { "epoch": 2.75, "learning_rate": 7.325761087930768e-06, "loss": 1.6187, "step": 143500 }, { "epoch": 2.76, "learning_rate": 7.316102611651986e-06, "loss": 1.619, "step": 144000 }, { "epoch": 2.76, "learning_rate": 7.306444135373204e-06, "loss": 1.6207, "step": 144500 }, { "epoch": 2.77, "learning_rate": 7.296785659094422e-06, "loss": 1.6177, "step": 145000 }, { "epoch": 2.78, "learning_rate": 7.287146499768197e-06, "loss": 1.6186, "step": 145500 }, { "epoch": 2.79, "learning_rate": 7.277488023489415e-06, "loss": 1.6175, "step": 146000 }, { "epoch": 2.8, "learning_rate": 7.267829547210633e-06, "loss": 1.6151, "step": 146500 }, { "epoch": 2.81, "learning_rate": 7.258171070931851e-06, "loss": 1.6145, "step": 147000 }, { "epoch": 2.82, "learning_rate": 7.2485319116056256e-06, "loss": 1.6206, "step": 147500 }, { "epoch": 2.83, "learning_rate": 7.2388734353268435e-06, "loss": 1.6189, "step": 148000 }, { "epoch": 2.84, "learning_rate": 7.2292149590480606e-06, "loss": 1.6163, "step": 148500 }, { "epoch": 2.85, "learning_rate": 7.2195564827692785e-06, "loss": 1.617, "step": 149000 }, { "epoch": 2.86, "learning_rate": 7.209917323443054e-06, "loss": 1.6202, "step": 149500 }, { "epoch": 2.87, "learning_rate": 7.200258847164272e-06, "loss": 1.6177, "step": 150000 }, { "epoch": 2.88, "learning_rate": 7.19060037088549e-06, "loss": 1.6173, "step": 150500 }, { "epoch": 2.89, "learning_rate": 7.1809418946067075e-06, "loss": 1.6186, "step": 151000 }, { "epoch": 2.9, "learning_rate": 7.1712834183279255e-06, "loss": 1.619, "step": 151500 }, { "epoch": 2.91, "learning_rate": 7.161644259001701e-06, "loss": 1.6189, "step": 152000 }, { "epoch": 2.92, "learning_rate": 7.151985782722918e-06, "loss": 1.6184, "step": 152500 }, { "epoch": 2.93, "learning_rate": 7.142327306444136e-06, "loss": 1.6178, "step": 153000 }, { "epoch": 2.94, "learning_rate": 7.132668830165354e-06, "loss": 1.6174, "step": 153500 }, { "epoch": 2.95, "learning_rate": 7.123029670839129e-06, "loss": 1.6175, "step": 154000 }, { "epoch": 2.96, "learning_rate": 7.113371194560347e-06, "loss": 1.6187, "step": 154500 }, { "epoch": 2.97, "learning_rate": 7.103712718281565e-06, "loss": 1.6213, "step": 155000 }, { "epoch": 2.98, "learning_rate": 7.094054242002783e-06, "loss": 1.6151, "step": 155500 }, { "epoch": 2.98, "learning_rate": 7.084415082676558e-06, "loss": 1.6154, "step": 156000 }, { "epoch": 2.99, "learning_rate": 7.074756606397776e-06, "loss": 1.6157, "step": 156500 }, { "epoch": 3.0, "learning_rate": 7.065098130118992e-06, "loss": 1.6197, "step": 157000 }, { "epoch": 3.01, "learning_rate": 7.05543965384021e-06, "loss": 1.6173, "step": 157500 }, { "epoch": 3.02, "learning_rate": 7.045781177561428e-06, "loss": 1.6182, "step": 158000 }, { "epoch": 3.03, "learning_rate": 7.036122701282646e-06, "loss": 1.6156, "step": 158500 }, { "epoch": 3.04, "learning_rate": 7.026483541956421e-06, "loss": 1.6168, "step": 159000 }, { "epoch": 3.05, "learning_rate": 7.016825065677639e-06, "loss": 1.6189, "step": 159500 }, { "epoch": 3.06, "learning_rate": 7.007166589398857e-06, "loss": 1.6204, "step": 160000 }, { "epoch": 3.07, "learning_rate": 6.997508113120075e-06, "loss": 1.6139, "step": 160500 }, { "epoch": 3.08, "learning_rate": 6.98786895379385e-06, "loss": 1.6178, "step": 161000 }, { "epoch": 3.09, "learning_rate": 6.978210477515068e-06, "loss": 1.6171, "step": 161500 }, { "epoch": 3.1, "learning_rate": 6.968552001236285e-06, "loss": 1.6155, "step": 162000 }, { "epoch": 3.11, "learning_rate": 6.958893524957503e-06, "loss": 1.618, "step": 162500 }, { "epoch": 3.12, "learning_rate": 6.9492543656312785e-06, "loss": 1.6159, "step": 163000 }, { "epoch": 3.13, "learning_rate": 6.939595889352496e-06, "loss": 1.614, "step": 163500 }, { "epoch": 3.14, "learning_rate": 6.929937413073714e-06, "loss": 1.617, "step": 164000 }, { "epoch": 3.15, "learning_rate": 6.920278936794932e-06, "loss": 1.6172, "step": 164500 }, { "epoch": 3.16, "learning_rate": 6.9106397774687075e-06, "loss": 1.6129, "step": 165000 }, { "epoch": 3.17, "learning_rate": 6.9009813011899254e-06, "loss": 1.6201, "step": 165500 }, { "epoch": 3.18, "learning_rate": 6.891322824911143e-06, "loss": 1.6163, "step": 166000 }, { "epoch": 3.19, "learning_rate": 6.88166434863236e-06, "loss": 1.6196, "step": 166500 }, { "epoch": 3.2, "learning_rate": 6.8720058723535775e-06, "loss": 1.6174, "step": 167000 }, { "epoch": 3.2, "learning_rate": 6.862366713027353e-06, "loss": 1.6116, "step": 167500 }, { "epoch": 3.21, "learning_rate": 6.852708236748571e-06, "loss": 1.61, "step": 168000 }, { "epoch": 3.22, "learning_rate": 6.843049760469789e-06, "loss": 1.6156, "step": 168500 }, { "epoch": 3.23, "learning_rate": 6.833391284191007e-06, "loss": 1.6167, "step": 169000 }, { "epoch": 3.24, "learning_rate": 6.8237328079122245e-06, "loss": 1.6116, "step": 169500 }, { "epoch": 3.25, "learning_rate": 6.814093648586e-06, "loss": 1.617, "step": 170000 }, { "epoch": 3.26, "learning_rate": 6.804435172307218e-06, "loss": 1.6188, "step": 170500 }, { "epoch": 3.27, "learning_rate": 6.794776696028436e-06, "loss": 1.6116, "step": 171000 }, { "epoch": 3.28, "learning_rate": 6.785118219749653e-06, "loss": 1.6157, "step": 171500 }, { "epoch": 3.29, "learning_rate": 6.775479060423428e-06, "loss": 1.6157, "step": 172000 }, { "epoch": 3.3, "learning_rate": 6.765820584144646e-06, "loss": 1.616, "step": 172500 }, { "epoch": 3.31, "learning_rate": 6.756162107865864e-06, "loss": 1.6168, "step": 173000 }, { "epoch": 3.32, "learning_rate": 6.746503631587081e-06, "loss": 1.6169, "step": 173500 }, { "epoch": 3.33, "learning_rate": 6.736845155308299e-06, "loss": 1.6156, "step": 174000 }, { "epoch": 3.34, "learning_rate": 6.727186679029517e-06, "loss": 1.6176, "step": 174500 }, { "epoch": 3.35, "learning_rate": 6.717547519703292e-06, "loss": 1.6182, "step": 175000 }, { "epoch": 3.36, "learning_rate": 6.70788904342451e-06, "loss": 1.6145, "step": 175500 }, { "epoch": 3.37, "learning_rate": 6.698230567145728e-06, "loss": 1.6162, "step": 176000 }, { "epoch": 3.38, "learning_rate": 6.688572090866945e-06, "loss": 1.6162, "step": 176500 }, { "epoch": 3.39, "learning_rate": 6.678913614588163e-06, "loss": 1.6166, "step": 177000 }, { "epoch": 3.4, "learning_rate": 6.669274455261938e-06, "loss": 1.6141, "step": 177500 }, { "epoch": 3.41, "learning_rate": 6.659615978983156e-06, "loss": 1.6158, "step": 178000 }, { "epoch": 3.42, "learning_rate": 6.649957502704374e-06, "loss": 1.6186, "step": 178500 }, { "epoch": 3.42, "learning_rate": 6.640299026425592e-06, "loss": 1.6137, "step": 179000 }, { "epoch": 3.43, "learning_rate": 6.63064055014681e-06, "loss": 1.6178, "step": 179500 }, { "epoch": 3.44, "learning_rate": 6.620982073868028e-06, "loss": 1.62, "step": 180000 }, { "epoch": 3.45, "learning_rate": 6.611342914541803e-06, "loss": 1.6134, "step": 180500 }, { "epoch": 3.46, "learning_rate": 6.601684438263019e-06, "loss": 1.6139, "step": 181000 }, { "epoch": 3.47, "learning_rate": 6.592025961984237e-06, "loss": 1.6166, "step": 181500 }, { "epoch": 3.48, "learning_rate": 6.582367485705455e-06, "loss": 1.616, "step": 182000 }, { "epoch": 3.49, "learning_rate": 6.5727283263792304e-06, "loss": 1.6191, "step": 182500 }, { "epoch": 3.5, "learning_rate": 6.563069850100448e-06, "loss": 1.6187, "step": 183000 }, { "epoch": 3.51, "learning_rate": 6.553411373821666e-06, "loss": 1.6152, "step": 183500 }, { "epoch": 3.52, "learning_rate": 6.543752897542884e-06, "loss": 1.6162, "step": 184000 }, { "epoch": 3.53, "learning_rate": 6.5341137382166595e-06, "loss": 1.6145, "step": 184500 }, { "epoch": 3.54, "learning_rate": 6.524455261937877e-06, "loss": 1.6189, "step": 185000 }, { "epoch": 3.55, "learning_rate": 6.514796785659095e-06, "loss": 1.6163, "step": 185500 }, { "epoch": 3.56, "learning_rate": 6.5051383093803124e-06, "loss": 1.6172, "step": 186000 }, { "epoch": 3.57, "learning_rate": 6.49547983310153e-06, "loss": 1.617, "step": 186500 }, { "epoch": 3.58, "learning_rate": 6.485840673775306e-06, "loss": 1.6148, "step": 187000 }, { "epoch": 3.59, "learning_rate": 6.4761821974965236e-06, "loss": 1.617, "step": 187500 }, { "epoch": 3.6, "learning_rate": 6.4665237212177415e-06, "loss": 1.6119, "step": 188000 }, { "epoch": 3.61, "learning_rate": 6.456865244938959e-06, "loss": 1.6143, "step": 188500 }, { "epoch": 3.62, "learning_rate": 6.447226085612735e-06, "loss": 1.6165, "step": 189000 }, { "epoch": 3.63, "learning_rate": 6.43758692628651e-06, "loss": 1.6142, "step": 189500 }, { "epoch": 3.64, "learning_rate": 6.427928450007728e-06, "loss": 1.6163, "step": 190000 }, { "epoch": 3.64, "learning_rate": 6.418269973728946e-06, "loss": 1.6151, "step": 190500 }, { "epoch": 3.65, "learning_rate": 6.408611497450162e-06, "loss": 1.6152, "step": 191000 }, { "epoch": 3.66, "learning_rate": 6.39895302117138e-06, "loss": 1.6158, "step": 191500 }, { "epoch": 3.67, "learning_rate": 6.389294544892598e-06, "loss": 1.6178, "step": 192000 }, { "epoch": 3.68, "learning_rate": 6.379636068613816e-06, "loss": 1.6151, "step": 192500 }, { "epoch": 3.69, "learning_rate": 6.369977592335034e-06, "loss": 1.6157, "step": 193000 }, { "epoch": 3.7, "learning_rate": 6.360338433008809e-06, "loss": 1.6147, "step": 193500 }, { "epoch": 3.71, "learning_rate": 6.350679956730027e-06, "loss": 1.62, "step": 194000 }, { "epoch": 3.72, "learning_rate": 6.341021480451245e-06, "loss": 1.6171, "step": 194500 }, { "epoch": 3.73, "learning_rate": 6.331363004172463e-06, "loss": 1.613, "step": 195000 }, { "epoch": 3.74, "learning_rate": 6.321723844846237e-06, "loss": 1.6155, "step": 195500 }, { "epoch": 3.75, "learning_rate": 6.312065368567455e-06, "loss": 1.6122, "step": 196000 }, { "epoch": 3.76, "learning_rate": 6.302406892288673e-06, "loss": 1.6157, "step": 196500 }, { "epoch": 3.77, "learning_rate": 6.292748416009891e-06, "loss": 1.6173, "step": 197000 }, { "epoch": 3.78, "learning_rate": 6.283089939731108e-06, "loss": 1.6127, "step": 197500 }, { "epoch": 3.79, "learning_rate": 6.273450780404884e-06, "loss": 1.6149, "step": 198000 }, { "epoch": 3.8, "learning_rate": 6.263792304126102e-06, "loss": 1.6183, "step": 198500 }, { "epoch": 3.81, "learning_rate": 6.254133827847319e-06, "loss": 1.6152, "step": 199000 }, { "epoch": 3.82, "learning_rate": 6.244475351568537e-06, "loss": 1.6135, "step": 199500 }, { "epoch": 3.83, "learning_rate": 6.234816875289755e-06, "loss": 1.6153, "step": 200000 }, { "epoch": 3.84, "learning_rate": 6.225158399010972e-06, "loss": 1.6148, "step": 200500 }, { "epoch": 3.85, "learning_rate": 6.215519239684747e-06, "loss": 1.6136, "step": 201000 }, { "epoch": 3.86, "learning_rate": 6.205860763405965e-06, "loss": 1.616, "step": 201500 }, { "epoch": 3.86, "learning_rate": 6.196202287127183e-06, "loss": 1.6146, "step": 202000 }, { "epoch": 3.87, "learning_rate": 6.186543810848401e-06, "loss": 1.6133, "step": 202500 }, { "epoch": 3.88, "learning_rate": 6.176885334569619e-06, "loss": 1.6134, "step": 203000 }, { "epoch": 3.89, "learning_rate": 6.167246175243394e-06, "loss": 1.6138, "step": 203500 }, { "epoch": 3.9, "learning_rate": 6.157587698964612e-06, "loss": 1.6158, "step": 204000 }, { "epoch": 3.91, "learning_rate": 6.14792922268583e-06, "loss": 1.6148, "step": 204500 }, { "epoch": 3.92, "learning_rate": 6.1382707464070465e-06, "loss": 1.6155, "step": 205000 }, { "epoch": 3.93, "learning_rate": 6.128631587080822e-06, "loss": 1.6139, "step": 205500 }, { "epoch": 3.94, "learning_rate": 6.11897311080204e-06, "loss": 1.6175, "step": 206000 }, { "epoch": 3.95, "learning_rate": 6.109314634523258e-06, "loss": 1.6148, "step": 206500 }, { "epoch": 3.96, "learning_rate": 6.0996561582444755e-06, "loss": 1.6117, "step": 207000 }, { "epoch": 3.97, "learning_rate": 6.090016998918251e-06, "loss": 1.617, "step": 207500 }, { "epoch": 3.98, "learning_rate": 6.080358522639469e-06, "loss": 1.615, "step": 208000 }, { "epoch": 3.99, "learning_rate": 6.070700046360687e-06, "loss": 1.615, "step": 208500 }, { "epoch": 4.0, "learning_rate": 6.061041570081905e-06, "loss": 1.6126, "step": 209000 }, { "epoch": 4.01, "learning_rate": 6.0513830938031225e-06, "loss": 1.6145, "step": 209500 }, { "epoch": 4.02, "learning_rate": 6.04172461752434e-06, "loss": 1.6145, "step": 210000 }, { "epoch": 4.03, "learning_rate": 6.032085458198115e-06, "loss": 1.6169, "step": 210500 }, { "epoch": 4.04, "learning_rate": 6.022426981919333e-06, "loss": 1.6144, "step": 211000 }, { "epoch": 4.05, "learning_rate": 6.012768505640551e-06, "loss": 1.6122, "step": 211500 }, { "epoch": 4.06, "learning_rate": 6.003110029361769e-06, "loss": 1.6162, "step": 212000 }, { "epoch": 4.07, "learning_rate": 5.9934515530829866e-06, "loss": 1.6116, "step": 212500 }, { "epoch": 4.08, "learning_rate": 5.983812393756762e-06, "loss": 1.6115, "step": 213000 }, { "epoch": 4.08, "learning_rate": 5.97415391747798e-06, "loss": 1.6114, "step": 213500 }, { "epoch": 4.09, "learning_rate": 5.964495441199198e-06, "loss": 1.6111, "step": 214000 }, { "epoch": 4.1, "learning_rate": 5.954836964920415e-06, "loss": 1.6119, "step": 214500 }, { "epoch": 4.11, "learning_rate": 5.945197805594189e-06, "loss": 1.6192, "step": 215000 }, { "epoch": 4.12, "learning_rate": 5.935539329315407e-06, "loss": 1.6117, "step": 215500 }, { "epoch": 4.13, "learning_rate": 5.925880853036625e-06, "loss": 1.6133, "step": 216000 }, { "epoch": 4.14, "learning_rate": 5.916222376757843e-06, "loss": 1.6125, "step": 216500 }, { "epoch": 4.15, "learning_rate": 5.906583217431618e-06, "loss": 1.6106, "step": 217000 }, { "epoch": 4.16, "learning_rate": 5.896924741152836e-06, "loss": 1.6106, "step": 217500 }, { "epoch": 4.17, "learning_rate": 5.887266264874054e-06, "loss": 1.6127, "step": 218000 }, { "epoch": 4.18, "learning_rate": 5.877607788595272e-06, "loss": 1.6112, "step": 218500 }, { "epoch": 4.19, "learning_rate": 5.86794931231649e-06, "loss": 1.6127, "step": 219000 }, { "epoch": 4.2, "learning_rate": 5.858310152990265e-06, "loss": 1.6168, "step": 219500 }, { "epoch": 4.21, "learning_rate": 5.848651676711482e-06, "loss": 1.6077, "step": 220000 }, { "epoch": 4.22, "learning_rate": 5.8389932004327e-06, "loss": 1.6129, "step": 220500 }, { "epoch": 4.23, "learning_rate": 5.829334724153918e-06, "loss": 1.6152, "step": 221000 }, { "epoch": 4.24, "learning_rate": 5.8196955648276934e-06, "loss": 1.6168, "step": 221500 }, { "epoch": 4.25, "learning_rate": 5.810037088548911e-06, "loss": 1.6144, "step": 222000 }, { "epoch": 4.26, "learning_rate": 5.800378612270129e-06, "loss": 1.613, "step": 222500 }, { "epoch": 4.27, "learning_rate": 5.790720135991346e-06, "loss": 1.6119, "step": 223000 }, { "epoch": 4.28, "learning_rate": 5.781061659712564e-06, "loss": 1.614, "step": 223500 }, { "epoch": 4.29, "learning_rate": 5.77142250038634e-06, "loss": 1.6163, "step": 224000 }, { "epoch": 4.3, "learning_rate": 5.761764024107557e-06, "loss": 1.6141, "step": 224500 }, { "epoch": 4.3, "learning_rate": 5.752105547828775e-06, "loss": 1.6147, "step": 225000 }, { "epoch": 4.31, "learning_rate": 5.7424470715499925e-06, "loss": 1.6119, "step": 225500 }, { "epoch": 4.32, "learning_rate": 5.732807912223768e-06, "loss": 1.6122, "step": 226000 }, { "epoch": 4.33, "learning_rate": 5.723149435944986e-06, "loss": 1.6125, "step": 226500 }, { "epoch": 4.34, "learning_rate": 5.713490959666204e-06, "loss": 1.6177, "step": 227000 }, { "epoch": 4.35, "learning_rate": 5.7038324833874216e-06, "loss": 1.6126, "step": 227500 }, { "epoch": 4.36, "learning_rate": 5.6941740071086395e-06, "loss": 1.6151, "step": 228000 }, { "epoch": 4.37, "learning_rate": 5.684534847782415e-06, "loss": 1.613, "step": 228500 }, { "epoch": 4.38, "learning_rate": 5.674876371503633e-06, "loss": 1.6142, "step": 229000 }, { "epoch": 4.39, "learning_rate": 5.665217895224849e-06, "loss": 1.6158, "step": 229500 }, { "epoch": 4.4, "learning_rate": 5.655559418946067e-06, "loss": 1.6134, "step": 230000 }, { "epoch": 4.41, "learning_rate": 5.645900942667285e-06, "loss": 1.6123, "step": 230500 }, { "epoch": 4.42, "learning_rate": 5.63626178334106e-06, "loss": 1.6152, "step": 231000 }, { "epoch": 4.43, "learning_rate": 5.626603307062278e-06, "loss": 1.6142, "step": 231500 }, { "epoch": 4.44, "learning_rate": 5.616944830783496e-06, "loss": 1.6137, "step": 232000 }, { "epoch": 4.45, "learning_rate": 5.607286354504714e-06, "loss": 1.6111, "step": 232500 }, { "epoch": 4.46, "learning_rate": 5.597647195178489e-06, "loss": 1.6121, "step": 233000 }, { "epoch": 4.47, "learning_rate": 5.587988718899707e-06, "loss": 1.6171, "step": 233500 }, { "epoch": 4.48, "learning_rate": 5.578330242620924e-06, "loss": 1.6144, "step": 234000 }, { "epoch": 4.49, "learning_rate": 5.568671766342142e-06, "loss": 1.6173, "step": 234500 }, { "epoch": 4.5, "learning_rate": 5.55901329006336e-06, "loss": 1.6124, "step": 235000 }, { "epoch": 4.51, "learning_rate": 5.549374130737135e-06, "loss": 1.6104, "step": 235500 }, { "epoch": 4.52, "learning_rate": 5.539715654458353e-06, "loss": 1.6095, "step": 236000 }, { "epoch": 4.52, "learning_rate": 5.530057178179571e-06, "loss": 1.608, "step": 236500 }, { "epoch": 4.53, "learning_rate": 5.520398701900789e-06, "loss": 1.6132, "step": 237000 }, { "epoch": 4.54, "learning_rate": 5.510759542574564e-06, "loss": 1.6132, "step": 237500 }, { "epoch": 4.55, "learning_rate": 5.501101066295782e-06, "loss": 1.6147, "step": 238000 }, { "epoch": 4.56, "learning_rate": 5.491442590017e-06, "loss": 1.6135, "step": 238500 }, { "epoch": 4.57, "learning_rate": 5.481784113738216e-06, "loss": 1.6129, "step": 239000 }, { "epoch": 4.58, "learning_rate": 5.472125637459434e-06, "loss": 1.6141, "step": 239500 }, { "epoch": 4.59, "learning_rate": 5.4624864781332096e-06, "loss": 1.6139, "step": 240000 }, { "epoch": 4.6, "learning_rate": 5.4528280018544275e-06, "loss": 1.6125, "step": 240500 }, { "epoch": 4.61, "learning_rate": 5.443169525575645e-06, "loss": 1.613, "step": 241000 }, { "epoch": 4.62, "learning_rate": 5.433511049296863e-06, "loss": 1.6123, "step": 241500 }, { "epoch": 4.63, "learning_rate": 5.423871889970639e-06, "loss": 1.6149, "step": 242000 }, { "epoch": 4.64, "learning_rate": 5.4142134136918565e-06, "loss": 1.6131, "step": 242500 }, { "epoch": 4.65, "learning_rate": 5.4045549374130745e-06, "loss": 1.6118, "step": 243000 }, { "epoch": 4.66, "learning_rate": 5.394896461134292e-06, "loss": 1.6102, "step": 243500 }, { "epoch": 4.67, "learning_rate": 5.3852379848555095e-06, "loss": 1.6146, "step": 244000 }, { "epoch": 4.68, "learning_rate": 5.375579508576727e-06, "loss": 1.6123, "step": 244500 }, { "epoch": 4.69, "learning_rate": 5.365921032297945e-06, "loss": 1.6144, "step": 245000 }, { "epoch": 4.7, "learning_rate": 5.356281872971721e-06, "loss": 1.6161, "step": 245500 }, { "epoch": 4.71, "learning_rate": 5.3466233966929385e-06, "loss": 1.6171, "step": 246000 }, { "epoch": 4.72, "learning_rate": 5.3369649204141565e-06, "loss": 1.611, "step": 246500 }, { "epoch": 4.73, "learning_rate": 5.3273064441353735e-06, "loss": 1.6102, "step": 247000 }, { "epoch": 4.74, "learning_rate": 5.3176479678565915e-06, "loss": 1.6107, "step": 247500 }, { "epoch": 4.74, "learning_rate": 5.308008808530367e-06, "loss": 1.6101, "step": 248000 }, { "epoch": 4.75, "learning_rate": 5.298350332251584e-06, "loss": 1.6115, "step": 248500 }, { "epoch": 4.76, "learning_rate": 5.288691855972802e-06, "loss": 1.6109, "step": 249000 }, { "epoch": 4.77, "learning_rate": 5.27903337969402e-06, "loss": 1.6103, "step": 249500 }, { "epoch": 4.78, "learning_rate": 5.269374903415238e-06, "loss": 1.6111, "step": 250000 }, { "epoch": 4.79, "learning_rate": 5.259735744089013e-06, "loss": 1.6134, "step": 250500 }, { "epoch": 4.8, "learning_rate": 5.250077267810231e-06, "loss": 1.6124, "step": 251000 }, { "epoch": 4.81, "learning_rate": 5.240418791531449e-06, "loss": 1.6117, "step": 251500 }, { "epoch": 4.82, "learning_rate": 5.230760315252667e-06, "loss": 1.6151, "step": 252000 }, { "epoch": 4.83, "learning_rate": 5.2211018389738846e-06, "loss": 1.612, "step": 252500 }, { "epoch": 4.84, "learning_rate": 5.211443362695101e-06, "loss": 1.6137, "step": 253000 }, { "epoch": 4.85, "learning_rate": 5.201804203368876e-06, "loss": 1.6147, "step": 253500 }, { "epoch": 4.86, "learning_rate": 5.192145727090094e-06, "loss": 1.6138, "step": 254000 }, { "epoch": 4.87, "learning_rate": 5.182487250811312e-06, "loss": 1.6151, "step": 254500 }, { "epoch": 4.88, "learning_rate": 5.17282877453253e-06, "loss": 1.6146, "step": 255000 }, { "epoch": 4.89, "learning_rate": 5.163170298253748e-06, "loss": 1.6127, "step": 255500 }, { "epoch": 4.9, "learning_rate": 5.153531138927523e-06, "loss": 1.612, "step": 256000 }, { "epoch": 4.91, "learning_rate": 5.143872662648741e-06, "loss": 1.6127, "step": 256500 }, { "epoch": 4.92, "learning_rate": 5.134214186369959e-06, "loss": 1.6124, "step": 257000 }, { "epoch": 4.93, "learning_rate": 5.124555710091177e-06, "loss": 1.6141, "step": 257500 }, { "epoch": 4.94, "learning_rate": 5.114897233812394e-06, "loss": 1.6145, "step": 258000 }, { "epoch": 4.95, "learning_rate": 5.105258074486169e-06, "loss": 1.6109, "step": 258500 }, { "epoch": 4.96, "learning_rate": 5.095599598207387e-06, "loss": 1.6079, "step": 259000 }, { "epoch": 4.96, "learning_rate": 5.085941121928605e-06, "loss": 1.6126, "step": 259500 }, { "epoch": 4.97, "learning_rate": 5.076282645649823e-06, "loss": 1.6127, "step": 260000 }, { "epoch": 4.98, "learning_rate": 5.066624169371041e-06, "loss": 1.6095, "step": 260500 }, { "epoch": 4.99, "learning_rate": 5.056985010044816e-06, "loss": 1.613, "step": 261000 }, { "epoch": 5.0, "learning_rate": 5.047326533766034e-06, "loss": 1.6118, "step": 261500 }, { "epoch": 5.01, "learning_rate": 5.037668057487252e-06, "loss": 1.6124, "step": 262000 }, { "epoch": 5.02, "learning_rate": 5.028009581208469e-06, "loss": 1.6107, "step": 262500 }, { "epoch": 5.03, "learning_rate": 5.0183704218822435e-06, "loss": 1.6127, "step": 263000 }, { "epoch": 5.04, "learning_rate": 5.0087119456034615e-06, "loss": 1.6145, "step": 263500 }, { "epoch": 5.05, "learning_rate": 4.999053469324679e-06, "loss": 1.6085, "step": 264000 }, { "epoch": 5.06, "learning_rate": 4.989394993045897e-06, "loss": 1.6128, "step": 264500 }, { "epoch": 5.07, "learning_rate": 4.979755833719673e-06, "loss": 1.6102, "step": 265000 }, { "epoch": 5.08, "learning_rate": 4.9700973574408905e-06, "loss": 1.6097, "step": 265500 }, { "epoch": 5.09, "learning_rate": 4.9604388811621084e-06, "loss": 1.6089, "step": 266000 }, { "epoch": 5.1, "learning_rate": 4.9507804048833255e-06, "loss": 1.6106, "step": 266500 }, { "epoch": 5.11, "learning_rate": 4.9411219286045434e-06, "loss": 1.6085, "step": 267000 }, { "epoch": 5.12, "learning_rate": 4.931482769278319e-06, "loss": 1.6115, "step": 267500 }, { "epoch": 5.13, "learning_rate": 4.921824292999537e-06, "loss": 1.6096, "step": 268000 }, { "epoch": 5.14, "learning_rate": 4.9121658167207546e-06, "loss": 1.61, "step": 268500 }, { "epoch": 5.15, "learning_rate": 4.9025073404419725e-06, "loss": 1.6103, "step": 269000 }, { "epoch": 5.16, "learning_rate": 4.8928488641631896e-06, "loss": 1.6106, "step": 269500 }, { "epoch": 5.17, "learning_rate": 4.883209704836966e-06, "loss": 1.6139, "step": 270000 }, { "epoch": 5.18, "learning_rate": 4.873551228558183e-06, "loss": 1.6095, "step": 270500 }, { "epoch": 5.18, "learning_rate": 4.863892752279401e-06, "loss": 1.6117, "step": 271000 }, { "epoch": 5.19, "learning_rate": 4.854234276000619e-06, "loss": 1.6126, "step": 271500 }, { "epoch": 5.2, "learning_rate": 4.844595116674394e-06, "loss": 1.6104, "step": 272000 }, { "epoch": 5.21, "learning_rate": 4.834936640395612e-06, "loss": 1.6073, "step": 272500 }, { "epoch": 5.22, "learning_rate": 4.82527816411683e-06, "loss": 1.6107, "step": 273000 }, { "epoch": 5.23, "learning_rate": 4.815619687838048e-06, "loss": 1.6116, "step": 273500 }, { "epoch": 5.24, "learning_rate": 4.805961211559265e-06, "loss": 1.6106, "step": 274000 }, { "epoch": 5.25, "learning_rate": 4.79632205223304e-06, "loss": 1.61, "step": 274500 }, { "epoch": 5.26, "learning_rate": 4.786663575954258e-06, "loss": 1.608, "step": 275000 }, { "epoch": 5.27, "learning_rate": 4.777005099675476e-06, "loss": 1.6133, "step": 275500 }, { "epoch": 5.28, "learning_rate": 4.767346623396693e-06, "loss": 1.6104, "step": 276000 }, { "epoch": 5.29, "learning_rate": 4.757707464070468e-06, "loss": 1.6099, "step": 276500 }, { "epoch": 5.3, "learning_rate": 4.748048987791686e-06, "loss": 1.6101, "step": 277000 }, { "epoch": 5.31, "learning_rate": 4.738390511512904e-06, "loss": 1.6115, "step": 277500 }, { "epoch": 5.32, "learning_rate": 4.728732035234122e-06, "loss": 1.6121, "step": 278000 }, { "epoch": 5.33, "learning_rate": 4.719073558955339e-06, "loss": 1.6103, "step": 278500 }, { "epoch": 5.34, "learning_rate": 4.709434399629114e-06, "loss": 1.6097, "step": 279000 }, { "epoch": 5.35, "learning_rate": 4.699775923350332e-06, "loss": 1.6123, "step": 279500 }, { "epoch": 5.36, "learning_rate": 4.69011744707155e-06, "loss": 1.6097, "step": 280000 }, { "epoch": 5.37, "learning_rate": 4.680458970792768e-06, "loss": 1.6104, "step": 280500 }, { "epoch": 5.38, "learning_rate": 4.670800494513986e-06, "loss": 1.6107, "step": 281000 }, { "epoch": 5.39, "learning_rate": 4.661161335187761e-06, "loss": 1.6106, "step": 281500 }, { "epoch": 5.4, "learning_rate": 4.651502858908979e-06, "loss": 1.6077, "step": 282000 }, { "epoch": 5.4, "learning_rate": 4.641844382630196e-06, "loss": 1.6108, "step": 282500 }, { "epoch": 5.41, "learning_rate": 4.632185906351414e-06, "loss": 1.6107, "step": 283000 }, { "epoch": 5.42, "learning_rate": 4.6225467470251896e-06, "loss": 1.6144, "step": 283500 }, { "epoch": 5.43, "learning_rate": 4.6128882707464075e-06, "loss": 1.6144, "step": 284000 }, { "epoch": 5.44, "learning_rate": 4.603229794467625e-06, "loss": 1.6138, "step": 284500 }, { "epoch": 5.45, "learning_rate": 4.593571318188843e-06, "loss": 1.6136, "step": 285000 }, { "epoch": 5.46, "learning_rate": 4.583932158862619e-06, "loss": 1.6099, "step": 285500 }, { "epoch": 5.47, "learning_rate": 4.574273682583836e-06, "loss": 1.6091, "step": 286000 }, { "epoch": 5.48, "learning_rate": 4.564615206305054e-06, "loss": 1.6072, "step": 286500 }, { "epoch": 5.49, "learning_rate": 4.5549567300262715e-06, "loss": 1.6103, "step": 287000 }, { "epoch": 5.5, "learning_rate": 4.5452982537474895e-06, "loss": 1.613, "step": 287500 }, { "epoch": 5.51, "learning_rate": 4.535659094421265e-06, "loss": 1.6097, "step": 288000 }, { "epoch": 5.52, "learning_rate": 4.526000618142482e-06, "loss": 1.6116, "step": 288500 }, { "epoch": 5.53, "learning_rate": 4.5163421418637e-06, "loss": 1.6107, "step": 289000 }, { "epoch": 5.54, "learning_rate": 4.506683665584918e-06, "loss": 1.6067, "step": 289500 }, { "epoch": 5.55, "learning_rate": 4.497025189306136e-06, "loss": 1.6122, "step": 290000 }, { "epoch": 5.56, "learning_rate": 4.487386029979911e-06, "loss": 1.6097, "step": 290500 }, { "epoch": 5.57, "learning_rate": 4.477727553701128e-06, "loss": 1.6078, "step": 291000 }, { "epoch": 5.58, "learning_rate": 4.468069077422346e-06, "loss": 1.6107, "step": 291500 }, { "epoch": 5.59, "learning_rate": 4.458410601143564e-06, "loss": 1.6091, "step": 292000 }, { "epoch": 5.6, "learning_rate": 4.448752124864782e-06, "loss": 1.6121, "step": 292500 }, { "epoch": 5.61, "learning_rate": 4.439112965538557e-06, "loss": 1.6121, "step": 293000 }, { "epoch": 5.62, "learning_rate": 4.429454489259775e-06, "loss": 1.6126, "step": 293500 }, { "epoch": 5.62, "learning_rate": 4.419796012980993e-06, "loss": 1.6132, "step": 294000 }, { "epoch": 5.63, "learning_rate": 4.41013753670221e-06, "loss": 1.6111, "step": 294500 }, { "epoch": 5.64, "learning_rate": 4.400498377375986e-06, "loss": 1.6112, "step": 295000 }, { "epoch": 5.65, "learning_rate": 4.390839901097203e-06, "loss": 1.6102, "step": 295500 }, { "epoch": 5.66, "learning_rate": 4.381181424818421e-06, "loss": 1.6114, "step": 296000 }, { "epoch": 5.67, "learning_rate": 4.371522948539639e-06, "loss": 1.6093, "step": 296500 }, { "epoch": 5.68, "learning_rate": 4.361864472260857e-06, "loss": 1.6125, "step": 297000 }, { "epoch": 5.69, "learning_rate": 4.352225312934632e-06, "loss": 1.6101, "step": 297500 }, { "epoch": 5.7, "learning_rate": 4.342566836655849e-06, "loss": 1.611, "step": 298000 }, { "epoch": 5.71, "learning_rate": 4.332908360377067e-06, "loss": 1.6111, "step": 298500 }, { "epoch": 5.72, "learning_rate": 4.323249884098285e-06, "loss": 1.6117, "step": 299000 }, { "epoch": 5.73, "learning_rate": 4.31361072477206e-06, "loss": 1.6118, "step": 299500 }, { "epoch": 5.74, "learning_rate": 4.303952248493278e-06, "loss": 1.611, "step": 300000 }, { "epoch": 5.75, "learning_rate": 4.294293772214495e-06, "loss": 1.6147, "step": 300500 }, { "epoch": 5.76, "learning_rate": 4.284635295935713e-06, "loss": 1.6097, "step": 301000 }, { "epoch": 5.77, "learning_rate": 4.274976819656931e-06, "loss": 1.6082, "step": 301500 }, { "epoch": 5.78, "learning_rate": 4.265318343378149e-06, "loss": 1.6096, "step": 302000 }, { "epoch": 5.79, "learning_rate": 4.2556791840519244e-06, "loss": 1.6054, "step": 302500 }, { "epoch": 5.8, "learning_rate": 4.2460207077731415e-06, "loss": 1.6098, "step": 303000 }, { "epoch": 5.81, "learning_rate": 4.2363622314943595e-06, "loss": 1.6085, "step": 303500 }, { "epoch": 5.82, "learning_rate": 4.226703755215577e-06, "loss": 1.615, "step": 304000 }, { "epoch": 5.83, "learning_rate": 4.217064595889353e-06, "loss": 1.61, "step": 304500 }, { "epoch": 5.84, "learning_rate": 4.207406119610571e-06, "loss": 1.6108, "step": 305000 }, { "epoch": 5.84, "learning_rate": 4.1977476433317885e-06, "loss": 1.6102, "step": 305500 }, { "epoch": 5.85, "learning_rate": 4.1880891670530064e-06, "loss": 1.6127, "step": 306000 }, { "epoch": 5.86, "learning_rate": 4.1784306907742235e-06, "loss": 1.609, "step": 306500 }, { "epoch": 5.87, "learning_rate": 4.168791531448e-06, "loss": 1.6085, "step": 307000 }, { "epoch": 5.88, "learning_rate": 4.159133055169217e-06, "loss": 1.607, "step": 307500 }, { "epoch": 5.89, "learning_rate": 4.149474578890435e-06, "loss": 1.6097, "step": 308000 }, { "epoch": 5.9, "learning_rate": 4.1398161026116526e-06, "loss": 1.6093, "step": 308500 }, { "epoch": 5.91, "learning_rate": 4.130176943285428e-06, "loss": 1.6127, "step": 309000 }, { "epoch": 5.92, "learning_rate": 4.120518467006646e-06, "loss": 1.611, "step": 309500 }, { "epoch": 5.93, "learning_rate": 4.110859990727863e-06, "loss": 1.6101, "step": 310000 } ], "max_steps": 522680, "num_train_epochs": 10, "total_flos": 6.527459946085057e+18, "trial_name": null, "trial_params": null }