{ "best_metric": null, "best_model_checkpoint": null, "epoch": 0.9994610049580405, "global_step": 280000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 4.9982152482054324e-05, "loss": 3.2655, "step": 100 }, { "epoch": 0.0, "learning_rate": 4.9964304964108645e-05, "loss": 3.2142, "step": 200 }, { "epoch": 0.0, "learning_rate": 4.9946457446162966e-05, "loss": 3.1805, "step": 300 }, { "epoch": 0.0, "learning_rate": 4.992860992821729e-05, "loss": 3.1781, "step": 400 }, { "epoch": 0.0, "learning_rate": 4.991076241027161e-05, "loss": 3.1664, "step": 500 }, { "epoch": 0.0, "learning_rate": 4.989291489232592e-05, "loss": 3.158, "step": 600 }, { "epoch": 0.0, "learning_rate": 4.9875067374380244e-05, "loss": 3.1607, "step": 700 }, { "epoch": 0.0, "learning_rate": 4.9857219856434565e-05, "loss": 3.1491, "step": 800 }, { "epoch": 0.0, "learning_rate": 4.9839372338488887e-05, "loss": 3.141, "step": 900 }, { "epoch": 0.0, "learning_rate": 4.982152482054321e-05, "loss": 3.1409, "step": 1000 }, { "epoch": 0.0, "learning_rate": 4.980367730259753e-05, "loss": 3.1489, "step": 1100 }, { "epoch": 0.0, "learning_rate": 4.978582978465185e-05, "loss": 3.1458, "step": 1200 }, { "epoch": 0.0, "learning_rate": 4.976798226670617e-05, "loss": 3.1312, "step": 1300 }, { "epoch": 0.0, "learning_rate": 4.975013474876049e-05, "loss": 3.1286, "step": 1400 }, { "epoch": 0.01, "learning_rate": 4.9732287230814814e-05, "loss": 3.1277, "step": 1500 }, { "epoch": 0.01, "learning_rate": 4.9714439712869135e-05, "loss": 3.1141, "step": 1600 }, { "epoch": 0.01, "learning_rate": 4.969659219492345e-05, "loss": 3.1065, "step": 1700 }, { "epoch": 0.01, "learning_rate": 4.967874467697777e-05, "loss": 3.1102, "step": 1800 }, { "epoch": 0.01, "learning_rate": 4.966089715903209e-05, "loss": 3.1169, "step": 1900 }, { "epoch": 0.01, "learning_rate": 4.964304964108641e-05, "loss": 3.115, "step": 2000 }, { "epoch": 0.01, "learning_rate": 4.9625202123140734e-05, "loss": 3.1067, "step": 2100 }, { "epoch": 0.01, "learning_rate": 4.9607354605195056e-05, "loss": 3.0965, "step": 2200 }, { "epoch": 0.01, "learning_rate": 4.958950708724938e-05, "loss": 3.0902, "step": 2300 }, { "epoch": 0.01, "learning_rate": 4.95716595693037e-05, "loss": 3.0859, "step": 2400 }, { "epoch": 0.01, "learning_rate": 4.955381205135802e-05, "loss": 3.0999, "step": 2500 }, { "epoch": 0.01, "learning_rate": 4.953596453341234e-05, "loss": 3.0828, "step": 2600 }, { "epoch": 0.01, "learning_rate": 4.951811701546666e-05, "loss": 3.0801, "step": 2700 }, { "epoch": 0.01, "learning_rate": 4.950026949752098e-05, "loss": 3.0769, "step": 2800 }, { "epoch": 0.01, "learning_rate": 4.9482421979575304e-05, "loss": 3.0826, "step": 2900 }, { "epoch": 0.01, "learning_rate": 4.9464574461629625e-05, "loss": 3.0805, "step": 3000 }, { "epoch": 0.01, "learning_rate": 4.9446726943683947e-05, "loss": 3.0767, "step": 3100 }, { "epoch": 0.01, "learning_rate": 4.942887942573827e-05, "loss": 3.0758, "step": 3200 }, { "epoch": 0.01, "learning_rate": 4.941103190779259e-05, "loss": 3.0669, "step": 3300 }, { "epoch": 0.01, "learning_rate": 4.939318438984691e-05, "loss": 3.0613, "step": 3400 }, { "epoch": 0.01, "learning_rate": 4.937533687190123e-05, "loss": 3.0743, "step": 3500 }, { "epoch": 0.01, "learning_rate": 4.935748935395555e-05, "loss": 3.0717, "step": 3600 }, { "epoch": 0.01, "learning_rate": 4.933964183600987e-05, "loss": 3.0649, "step": 3700 }, { "epoch": 0.01, "learning_rate": 4.932179431806419e-05, "loss": 3.0618, "step": 3800 }, { "epoch": 0.01, "learning_rate": 4.930394680011851e-05, "loss": 3.0586, "step": 3900 }, { "epoch": 0.01, "learning_rate": 4.928609928217283e-05, "loss": 3.0575, "step": 4000 }, { "epoch": 0.01, "learning_rate": 4.926825176422715e-05, "loss": 3.0582, "step": 4100 }, { "epoch": 0.01, "learning_rate": 4.925040424628147e-05, "loss": 3.0567, "step": 4200 }, { "epoch": 0.02, "learning_rate": 4.9232556728335794e-05, "loss": 3.0404, "step": 4300 }, { "epoch": 0.02, "learning_rate": 4.9214709210390116e-05, "loss": 3.0449, "step": 4400 }, { "epoch": 0.02, "learning_rate": 4.919686169244444e-05, "loss": 3.0541, "step": 4500 }, { "epoch": 0.02, "learning_rate": 4.917901417449876e-05, "loss": 3.0451, "step": 4600 }, { "epoch": 0.02, "learning_rate": 4.916116665655308e-05, "loss": 3.0483, "step": 4700 }, { "epoch": 0.02, "learning_rate": 4.9143319138607394e-05, "loss": 3.0485, "step": 4800 }, { "epoch": 0.02, "learning_rate": 4.9125471620661715e-05, "loss": 3.0481, "step": 4900 }, { "epoch": 0.02, "learning_rate": 4.9107624102716036e-05, "loss": 3.0357, "step": 5000 }, { "epoch": 0.02, "learning_rate": 4.908977658477036e-05, "loss": 3.0463, "step": 5100 }, { "epoch": 0.02, "learning_rate": 4.907192906682468e-05, "loss": 3.0369, "step": 5200 }, { "epoch": 0.02, "learning_rate": 4.9054081548879e-05, "loss": 3.0465, "step": 5300 }, { "epoch": 0.02, "learning_rate": 4.903623403093332e-05, "loss": 3.0229, "step": 5400 }, { "epoch": 0.02, "learning_rate": 4.901838651298764e-05, "loss": 3.0339, "step": 5500 }, { "epoch": 0.02, "learning_rate": 4.900053899504196e-05, "loss": 3.0449, "step": 5600 }, { "epoch": 0.02, "learning_rate": 4.8982691477096285e-05, "loss": 3.0402, "step": 5700 }, { "epoch": 0.02, "learning_rate": 4.8964843959150606e-05, "loss": 3.0287, "step": 5800 }, { "epoch": 0.02, "learning_rate": 4.894699644120492e-05, "loss": 3.0269, "step": 5900 }, { "epoch": 0.02, "learning_rate": 4.892914892325924e-05, "loss": 3.0284, "step": 6000 }, { "epoch": 0.02, "learning_rate": 4.891130140531356e-05, "loss": 3.0195, "step": 6100 }, { "epoch": 0.02, "learning_rate": 4.8893453887367884e-05, "loss": 3.0309, "step": 6200 }, { "epoch": 0.02, "learning_rate": 4.8875606369422205e-05, "loss": 3.0327, "step": 6300 }, { "epoch": 0.02, "learning_rate": 4.8857758851476526e-05, "loss": 3.0303, "step": 6400 }, { "epoch": 0.02, "learning_rate": 4.883991133353085e-05, "loss": 3.0224, "step": 6500 }, { "epoch": 0.02, "learning_rate": 4.882206381558517e-05, "loss": 3.0304, "step": 6600 }, { "epoch": 0.02, "learning_rate": 4.880421629763949e-05, "loss": 3.0275, "step": 6700 }, { "epoch": 0.02, "learning_rate": 4.878636877969381e-05, "loss": 3.0202, "step": 6800 }, { "epoch": 0.02, "learning_rate": 4.876852126174813e-05, "loss": 3.0301, "step": 6900 }, { "epoch": 0.02, "learning_rate": 4.875067374380245e-05, "loss": 3.0177, "step": 7000 }, { "epoch": 0.03, "learning_rate": 4.873282622585677e-05, "loss": 3.0258, "step": 7100 }, { "epoch": 0.03, "learning_rate": 4.871497870791109e-05, "loss": 3.0146, "step": 7200 }, { "epoch": 0.03, "learning_rate": 4.869713118996541e-05, "loss": 3.0197, "step": 7300 }, { "epoch": 0.03, "learning_rate": 4.867928367201973e-05, "loss": 3.0226, "step": 7400 }, { "epoch": 0.03, "learning_rate": 4.866143615407405e-05, "loss": 3.0154, "step": 7500 }, { "epoch": 0.03, "learning_rate": 4.8643588636128374e-05, "loss": 3.0075, "step": 7600 }, { "epoch": 0.03, "learning_rate": 4.86257411181827e-05, "loss": 3.0174, "step": 7700 }, { "epoch": 0.03, "learning_rate": 4.860789360023702e-05, "loss": 3.0117, "step": 7800 }, { "epoch": 0.03, "learning_rate": 4.859004608229134e-05, "loss": 3.0067, "step": 7900 }, { "epoch": 0.03, "learning_rate": 4.857219856434566e-05, "loss": 3.0154, "step": 8000 }, { "epoch": 0.03, "learning_rate": 4.855435104639998e-05, "loss": 3.016, "step": 8100 }, { "epoch": 0.03, "learning_rate": 4.85365035284543e-05, "loss": 3.0083, "step": 8200 }, { "epoch": 0.03, "learning_rate": 4.851865601050862e-05, "loss": 3.0094, "step": 8300 }, { "epoch": 0.03, "learning_rate": 4.8500808492562944e-05, "loss": 3.0071, "step": 8400 }, { "epoch": 0.03, "learning_rate": 4.8482960974617265e-05, "loss": 3.0037, "step": 8500 }, { "epoch": 0.03, "learning_rate": 4.8465113456671586e-05, "loss": 3.0166, "step": 8600 }, { "epoch": 0.03, "learning_rate": 4.844726593872591e-05, "loss": 3.0083, "step": 8700 }, { "epoch": 0.03, "learning_rate": 4.842941842078023e-05, "loss": 3.0019, "step": 8800 }, { "epoch": 0.03, "learning_rate": 4.841157090283455e-05, "loss": 3.0104, "step": 8900 }, { "epoch": 0.03, "learning_rate": 4.8393723384888864e-05, "loss": 2.9979, "step": 9000 }, { "epoch": 0.03, "learning_rate": 4.8375875866943186e-05, "loss": 3.0066, "step": 9100 }, { "epoch": 0.03, "learning_rate": 4.835802834899751e-05, "loss": 3.0047, "step": 9200 }, { "epoch": 0.03, "learning_rate": 4.834018083105183e-05, "loss": 3.0059, "step": 9300 }, { "epoch": 0.03, "learning_rate": 4.832233331310615e-05, "loss": 2.9923, "step": 9400 }, { "epoch": 0.03, "learning_rate": 4.830448579516047e-05, "loss": 2.9994, "step": 9500 }, { "epoch": 0.03, "learning_rate": 4.828663827721479e-05, "loss": 2.9912, "step": 9600 }, { "epoch": 0.03, "learning_rate": 4.826879075926911e-05, "loss": 2.993, "step": 9700 }, { "epoch": 0.03, "learning_rate": 4.8250943241323434e-05, "loss": 2.985, "step": 9800 }, { "epoch": 0.04, "learning_rate": 4.8233095723377755e-05, "loss": 2.9926, "step": 9900 }, { "epoch": 0.04, "learning_rate": 4.8215248205432076e-05, "loss": 2.9905, "step": 10000 }, { "epoch": 0.04, "learning_rate": 4.819740068748639e-05, "loss": 2.9819, "step": 10100 }, { "epoch": 0.04, "learning_rate": 4.817955316954071e-05, "loss": 2.9921, "step": 10200 }, { "epoch": 0.04, "learning_rate": 4.816170565159503e-05, "loss": 2.9812, "step": 10300 }, { "epoch": 0.04, "learning_rate": 4.8143858133649355e-05, "loss": 2.9954, "step": 10400 }, { "epoch": 0.04, "learning_rate": 4.8126010615703676e-05, "loss": 2.9938, "step": 10500 }, { "epoch": 0.04, "learning_rate": 4.8108163097758e-05, "loss": 2.983, "step": 10600 }, { "epoch": 0.04, "learning_rate": 4.809031557981232e-05, "loss": 2.9933, "step": 10700 }, { "epoch": 0.04, "learning_rate": 4.807246806186664e-05, "loss": 2.9954, "step": 10800 }, { "epoch": 0.04, "learning_rate": 4.805462054392096e-05, "loss": 2.9773, "step": 10900 }, { "epoch": 0.04, "learning_rate": 4.803677302597528e-05, "loss": 2.9905, "step": 11000 }, { "epoch": 0.04, "learning_rate": 4.80189255080296e-05, "loss": 2.9884, "step": 11100 }, { "epoch": 0.04, "learning_rate": 4.800107799008392e-05, "loss": 2.9752, "step": 11200 }, { "epoch": 0.04, "learning_rate": 4.798323047213824e-05, "loss": 2.9862, "step": 11300 }, { "epoch": 0.04, "learning_rate": 4.796538295419256e-05, "loss": 2.983, "step": 11400 }, { "epoch": 0.04, "learning_rate": 4.794753543624688e-05, "loss": 2.9796, "step": 11500 }, { "epoch": 0.04, "learning_rate": 4.79296879183012e-05, "loss": 2.9725, "step": 11600 }, { "epoch": 0.04, "learning_rate": 4.7911840400355524e-05, "loss": 2.9862, "step": 11700 }, { "epoch": 0.04, "learning_rate": 4.7893992882409845e-05, "loss": 2.9857, "step": 11800 }, { "epoch": 0.04, "learning_rate": 4.7876145364464166e-05, "loss": 2.9709, "step": 11900 }, { "epoch": 0.04, "learning_rate": 4.785829784651849e-05, "loss": 2.9848, "step": 12000 }, { "epoch": 0.04, "learning_rate": 4.784045032857281e-05, "loss": 2.9783, "step": 12100 }, { "epoch": 0.04, "learning_rate": 4.782260281062712e-05, "loss": 2.9718, "step": 12200 }, { "epoch": 0.04, "learning_rate": 4.7804755292681444e-05, "loss": 2.9787, "step": 12300 }, { "epoch": 0.04, "learning_rate": 4.7786907774735765e-05, "loss": 2.9676, "step": 12400 }, { "epoch": 0.04, "learning_rate": 4.7769060256790086e-05, "loss": 2.9666, "step": 12500 }, { "epoch": 0.04, "learning_rate": 4.775121273884441e-05, "loss": 2.9778, "step": 12600 }, { "epoch": 0.05, "learning_rate": 4.773336522089873e-05, "loss": 2.9741, "step": 12700 }, { "epoch": 0.05, "learning_rate": 4.771551770295306e-05, "loss": 2.9708, "step": 12800 }, { "epoch": 0.05, "learning_rate": 4.769767018500738e-05, "loss": 2.9686, "step": 12900 }, { "epoch": 0.05, "learning_rate": 4.76798226670617e-05, "loss": 2.9707, "step": 13000 }, { "epoch": 0.05, "learning_rate": 4.7661975149116014e-05, "loss": 2.9682, "step": 13100 }, { "epoch": 0.05, "learning_rate": 4.7644127631170335e-05, "loss": 2.966, "step": 13200 }, { "epoch": 0.05, "learning_rate": 4.7626280113224656e-05, "loss": 2.9716, "step": 13300 }, { "epoch": 0.05, "learning_rate": 4.760843259527898e-05, "loss": 2.9772, "step": 13400 }, { "epoch": 0.05, "learning_rate": 4.75905850773333e-05, "loss": 2.9659, "step": 13500 }, { "epoch": 0.05, "learning_rate": 4.757273755938762e-05, "loss": 2.9753, "step": 13600 }, { "epoch": 0.05, "learning_rate": 4.755489004144194e-05, "loss": 2.9617, "step": 13700 }, { "epoch": 0.05, "learning_rate": 4.753704252349626e-05, "loss": 2.9614, "step": 13800 }, { "epoch": 0.05, "learning_rate": 4.7519195005550584e-05, "loss": 2.9713, "step": 13900 }, { "epoch": 0.05, "learning_rate": 4.7501347487604905e-05, "loss": 2.9694, "step": 14000 }, { "epoch": 0.05, "learning_rate": 4.7483499969659226e-05, "loss": 2.9663, "step": 14100 }, { "epoch": 0.05, "learning_rate": 4.746565245171354e-05, "loss": 2.9633, "step": 14200 }, { "epoch": 0.05, "learning_rate": 4.744780493376786e-05, "loss": 2.9668, "step": 14300 }, { "epoch": 0.05, "learning_rate": 4.742995741582218e-05, "loss": 2.956, "step": 14400 }, { "epoch": 0.05, "learning_rate": 4.7412109897876504e-05, "loss": 2.9591, "step": 14500 }, { "epoch": 0.05, "learning_rate": 4.7394262379930825e-05, "loss": 2.9598, "step": 14600 }, { "epoch": 0.05, "learning_rate": 4.7376414861985146e-05, "loss": 2.95, "step": 14700 }, { "epoch": 0.05, "learning_rate": 4.735856734403947e-05, "loss": 2.9574, "step": 14800 }, { "epoch": 0.05, "learning_rate": 4.734071982609379e-05, "loss": 2.9655, "step": 14900 }, { "epoch": 0.05, "learning_rate": 4.732287230814811e-05, "loss": 2.9585, "step": 15000 }, { "epoch": 0.05, "learning_rate": 4.730502479020243e-05, "loss": 2.9571, "step": 15100 }, { "epoch": 0.05, "learning_rate": 4.728717727225675e-05, "loss": 2.9675, "step": 15200 }, { "epoch": 0.05, "learning_rate": 4.726932975431107e-05, "loss": 2.9577, "step": 15300 }, { "epoch": 0.05, "learning_rate": 4.725148223636539e-05, "loss": 2.9613, "step": 15400 }, { "epoch": 0.06, "learning_rate": 4.723363471841971e-05, "loss": 2.9628, "step": 15500 }, { "epoch": 0.06, "learning_rate": 4.721578720047403e-05, "loss": 2.9584, "step": 15600 }, { "epoch": 0.06, "learning_rate": 4.719793968252835e-05, "loss": 2.9501, "step": 15700 }, { "epoch": 0.06, "learning_rate": 4.718009216458267e-05, "loss": 2.9566, "step": 15800 }, { "epoch": 0.06, "learning_rate": 4.7162244646636994e-05, "loss": 2.9547, "step": 15900 }, { "epoch": 0.06, "learning_rate": 4.7144397128691315e-05, "loss": 2.9563, "step": 16000 }, { "epoch": 0.06, "learning_rate": 4.712654961074564e-05, "loss": 2.9514, "step": 16100 }, { "epoch": 0.06, "learning_rate": 4.710870209279996e-05, "loss": 2.953, "step": 16200 }, { "epoch": 0.06, "learning_rate": 4.709085457485428e-05, "loss": 2.9614, "step": 16300 }, { "epoch": 0.06, "learning_rate": 4.7073007056908594e-05, "loss": 2.9524, "step": 16400 }, { "epoch": 0.06, "learning_rate": 4.7055159538962915e-05, "loss": 2.9636, "step": 16500 }, { "epoch": 0.06, "learning_rate": 4.7037312021017236e-05, "loss": 2.9469, "step": 16600 }, { "epoch": 0.06, "learning_rate": 4.701946450307156e-05, "loss": 2.9505, "step": 16700 }, { "epoch": 0.06, "learning_rate": 4.700161698512588e-05, "loss": 2.9413, "step": 16800 }, { "epoch": 0.06, "learning_rate": 4.69837694671802e-05, "loss": 2.9437, "step": 16900 }, { "epoch": 0.06, "learning_rate": 4.696592194923452e-05, "loss": 2.9568, "step": 17000 }, { "epoch": 0.06, "learning_rate": 4.694807443128884e-05, "loss": 2.9489, "step": 17100 }, { "epoch": 0.06, "learning_rate": 4.693022691334316e-05, "loss": 2.9461, "step": 17200 }, { "epoch": 0.06, "learning_rate": 4.6912379395397484e-05, "loss": 2.9527, "step": 17300 }, { "epoch": 0.06, "learning_rate": 4.6894531877451806e-05, "loss": 2.9453, "step": 17400 }, { "epoch": 0.06, "learning_rate": 4.687668435950612e-05, "loss": 2.9504, "step": 17500 }, { "epoch": 0.06, "learning_rate": 4.685883684156044e-05, "loss": 2.9493, "step": 17600 }, { "epoch": 0.06, "learning_rate": 4.684098932361476e-05, "loss": 2.9494, "step": 17700 }, { "epoch": 0.06, "learning_rate": 4.6823141805669084e-05, "loss": 2.9507, "step": 17800 }, { "epoch": 0.06, "learning_rate": 4.680529428772341e-05, "loss": 2.9527, "step": 17900 }, { "epoch": 0.06, "learning_rate": 4.678744676977773e-05, "loss": 2.9302, "step": 18000 }, { "epoch": 0.06, "learning_rate": 4.6769599251832054e-05, "loss": 2.9517, "step": 18100 }, { "epoch": 0.06, "learning_rate": 4.6751751733886375e-05, "loss": 2.927, "step": 18200 }, { "epoch": 0.07, "learning_rate": 4.67339042159407e-05, "loss": 2.9401, "step": 18300 }, { "epoch": 0.07, "learning_rate": 4.671605669799501e-05, "loss": 2.9564, "step": 18400 }, { "epoch": 0.07, "learning_rate": 4.669820918004933e-05, "loss": 2.943, "step": 18500 }, { "epoch": 0.07, "learning_rate": 4.6680361662103654e-05, "loss": 2.9414, "step": 18600 }, { "epoch": 0.07, "learning_rate": 4.6662514144157975e-05, "loss": 2.9446, "step": 18700 }, { "epoch": 0.07, "learning_rate": 4.6644666626212296e-05, "loss": 2.9462, "step": 18800 }, { "epoch": 0.07, "learning_rate": 4.662681910826662e-05, "loss": 2.9353, "step": 18900 }, { "epoch": 0.07, "learning_rate": 4.660897159032094e-05, "loss": 2.9467, "step": 19000 }, { "epoch": 0.07, "learning_rate": 4.659112407237526e-05, "loss": 2.9301, "step": 19100 }, { "epoch": 0.07, "learning_rate": 4.657327655442958e-05, "loss": 2.9428, "step": 19200 }, { "epoch": 0.07, "learning_rate": 4.65554290364839e-05, "loss": 2.9403, "step": 19300 }, { "epoch": 0.07, "learning_rate": 4.653758151853822e-05, "loss": 2.9301, "step": 19400 }, { "epoch": 0.07, "learning_rate": 4.651973400059254e-05, "loss": 2.9338, "step": 19500 }, { "epoch": 0.07, "learning_rate": 4.650188648264686e-05, "loss": 2.9398, "step": 19600 }, { "epoch": 0.07, "learning_rate": 4.648403896470118e-05, "loss": 2.9341, "step": 19700 }, { "epoch": 0.07, "learning_rate": 4.64661914467555e-05, "loss": 2.9378, "step": 19800 }, { "epoch": 0.07, "learning_rate": 4.644834392880982e-05, "loss": 2.9242, "step": 19900 }, { "epoch": 0.07, "learning_rate": 4.6430496410864144e-05, "loss": 2.9369, "step": 20000 }, { "epoch": 0.07, "learning_rate": 4.6412648892918465e-05, "loss": 2.9442, "step": 20100 }, { "epoch": 0.07, "learning_rate": 4.6394801374972786e-05, "loss": 2.934, "step": 20200 }, { "epoch": 0.07, "learning_rate": 4.637695385702711e-05, "loss": 2.9385, "step": 20300 }, { "epoch": 0.07, "learning_rate": 4.635910633908143e-05, "loss": 2.9304, "step": 20400 }, { "epoch": 0.07, "learning_rate": 4.634125882113575e-05, "loss": 2.9351, "step": 20500 }, { "epoch": 0.07, "learning_rate": 4.6323411303190064e-05, "loss": 2.9297, "step": 20600 }, { "epoch": 0.07, "learning_rate": 4.6305563785244385e-05, "loss": 2.9271, "step": 20700 }, { "epoch": 0.07, "learning_rate": 4.628771626729871e-05, "loss": 2.9317, "step": 20800 }, { "epoch": 0.07, "learning_rate": 4.626986874935303e-05, "loss": 2.9295, "step": 20900 }, { "epoch": 0.07, "learning_rate": 4.625202123140735e-05, "loss": 2.9402, "step": 21000 }, { "epoch": 0.08, "learning_rate": 4.623417371346167e-05, "loss": 2.9259, "step": 21100 }, { "epoch": 0.08, "learning_rate": 4.621632619551599e-05, "loss": 2.933, "step": 21200 }, { "epoch": 0.08, "learning_rate": 4.619847867757031e-05, "loss": 2.9216, "step": 21300 }, { "epoch": 0.08, "learning_rate": 4.6180631159624634e-05, "loss": 2.933, "step": 21400 }, { "epoch": 0.08, "learning_rate": 4.6162783641678955e-05, "loss": 2.9376, "step": 21500 }, { "epoch": 0.08, "learning_rate": 4.6144936123733276e-05, "loss": 2.9255, "step": 21600 }, { "epoch": 0.08, "learning_rate": 4.612708860578759e-05, "loss": 2.9342, "step": 21700 }, { "epoch": 0.08, "learning_rate": 4.610924108784191e-05, "loss": 2.9278, "step": 21800 }, { "epoch": 0.08, "learning_rate": 4.609139356989623e-05, "loss": 2.9181, "step": 21900 }, { "epoch": 0.08, "learning_rate": 4.6073546051950554e-05, "loss": 2.927, "step": 22000 }, { "epoch": 0.08, "learning_rate": 4.6055698534004876e-05, "loss": 2.9297, "step": 22100 }, { "epoch": 0.08, "learning_rate": 4.60378510160592e-05, "loss": 2.9301, "step": 22200 }, { "epoch": 0.08, "learning_rate": 4.602000349811352e-05, "loss": 2.9257, "step": 22300 }, { "epoch": 0.08, "learning_rate": 4.600215598016784e-05, "loss": 2.9263, "step": 22400 }, { "epoch": 0.08, "learning_rate": 4.598430846222216e-05, "loss": 2.9152, "step": 22500 }, { "epoch": 0.08, "learning_rate": 4.596646094427648e-05, "loss": 2.9272, "step": 22600 }, { "epoch": 0.08, "learning_rate": 4.59486134263308e-05, "loss": 2.9196, "step": 22700 }, { "epoch": 0.08, "learning_rate": 4.593076590838512e-05, "loss": 2.9285, "step": 22800 }, { "epoch": 0.08, "learning_rate": 4.591291839043944e-05, "loss": 2.9176, "step": 22900 }, { "epoch": 0.08, "learning_rate": 4.589507087249377e-05, "loss": 2.9267, "step": 23000 }, { "epoch": 0.08, "learning_rate": 4.587722335454809e-05, "loss": 2.9226, "step": 23100 }, { "epoch": 0.08, "learning_rate": 4.585937583660241e-05, "loss": 2.9278, "step": 23200 }, { "epoch": 0.08, "learning_rate": 4.584152831865673e-05, "loss": 2.9186, "step": 23300 }, { "epoch": 0.08, "learning_rate": 4.582368080071105e-05, "loss": 2.9262, "step": 23400 }, { "epoch": 0.08, "learning_rate": 4.580583328276537e-05, "loss": 2.9204, "step": 23500 }, { "epoch": 0.08, "learning_rate": 4.5787985764819694e-05, "loss": 2.9287, "step": 23600 }, { "epoch": 0.08, "learning_rate": 4.577013824687401e-05, "loss": 2.9274, "step": 23700 }, { "epoch": 0.08, "learning_rate": 4.575229072892833e-05, "loss": 2.904, "step": 23800 }, { "epoch": 0.09, "learning_rate": 4.573444321098265e-05, "loss": 2.9017, "step": 23900 }, { "epoch": 0.09, "learning_rate": 4.571659569303697e-05, "loss": 2.9142, "step": 24000 }, { "epoch": 0.09, "learning_rate": 4.569874817509129e-05, "loss": 2.9195, "step": 24100 }, { "epoch": 0.09, "learning_rate": 4.5680900657145614e-05, "loss": 2.9171, "step": 24200 }, { "epoch": 0.09, "learning_rate": 4.5663053139199936e-05, "loss": 2.921, "step": 24300 }, { "epoch": 0.09, "learning_rate": 4.564520562125426e-05, "loss": 2.9243, "step": 24400 }, { "epoch": 0.09, "learning_rate": 4.562735810330858e-05, "loss": 2.9245, "step": 24500 }, { "epoch": 0.09, "learning_rate": 4.56095105853629e-05, "loss": 2.9135, "step": 24600 }, { "epoch": 0.09, "learning_rate": 4.559166306741722e-05, "loss": 2.9224, "step": 24700 }, { "epoch": 0.09, "learning_rate": 4.5573815549471535e-05, "loss": 2.9328, "step": 24800 }, { "epoch": 0.09, "learning_rate": 4.5555968031525856e-05, "loss": 2.9171, "step": 24900 }, { "epoch": 0.09, "learning_rate": 4.553812051358018e-05, "loss": 2.9026, "step": 25000 }, { "epoch": 0.09, "learning_rate": 4.55202729956345e-05, "loss": 2.9174, "step": 25100 }, { "epoch": 0.09, "learning_rate": 4.550242547768882e-05, "loss": 2.915, "step": 25200 }, { "epoch": 0.09, "learning_rate": 4.548457795974314e-05, "loss": 2.9166, "step": 25300 }, { "epoch": 0.09, "learning_rate": 4.546673044179746e-05, "loss": 2.92, "step": 25400 }, { "epoch": 0.09, "learning_rate": 4.5448882923851783e-05, "loss": 2.9208, "step": 25500 }, { "epoch": 0.09, "learning_rate": 4.5431035405906105e-05, "loss": 2.9125, "step": 25600 }, { "epoch": 0.09, "learning_rate": 4.5413187887960426e-05, "loss": 2.9141, "step": 25700 }, { "epoch": 0.09, "learning_rate": 4.539534037001475e-05, "loss": 2.9063, "step": 25800 }, { "epoch": 0.09, "learning_rate": 4.537749285206906e-05, "loss": 2.9108, "step": 25900 }, { "epoch": 0.09, "learning_rate": 4.535964533412338e-05, "loss": 2.8963, "step": 26000 }, { "epoch": 0.09, "learning_rate": 4.5341797816177704e-05, "loss": 2.9079, "step": 26100 }, { "epoch": 0.09, "learning_rate": 4.5323950298232025e-05, "loss": 2.9122, "step": 26200 }, { "epoch": 0.09, "learning_rate": 4.5306102780286346e-05, "loss": 2.9094, "step": 26300 }, { "epoch": 0.09, "learning_rate": 4.528825526234067e-05, "loss": 2.9195, "step": 26400 }, { "epoch": 0.09, "learning_rate": 4.527040774439499e-05, "loss": 2.9079, "step": 26500 }, { "epoch": 0.09, "learning_rate": 4.525256022644931e-05, "loss": 2.9167, "step": 26600 }, { "epoch": 0.1, "learning_rate": 4.523471270850363e-05, "loss": 2.9091, "step": 26700 }, { "epoch": 0.1, "learning_rate": 4.521686519055795e-05, "loss": 2.9022, "step": 26800 }, { "epoch": 0.1, "learning_rate": 4.5199017672612274e-05, "loss": 2.9099, "step": 26900 }, { "epoch": 0.1, "learning_rate": 4.518117015466659e-05, "loss": 2.9264, "step": 27000 }, { "epoch": 0.1, "learning_rate": 4.516332263672091e-05, "loss": 2.911, "step": 27100 }, { "epoch": 0.1, "learning_rate": 4.514547511877523e-05, "loss": 2.9119, "step": 27200 }, { "epoch": 0.1, "learning_rate": 4.512762760082955e-05, "loss": 2.9108, "step": 27300 }, { "epoch": 0.1, "learning_rate": 4.510978008288387e-05, "loss": 2.9135, "step": 27400 }, { "epoch": 0.1, "learning_rate": 4.5091932564938194e-05, "loss": 2.9107, "step": 27500 }, { "epoch": 0.1, "learning_rate": 4.5074085046992515e-05, "loss": 2.8936, "step": 27600 }, { "epoch": 0.1, "learning_rate": 4.505623752904684e-05, "loss": 2.9127, "step": 27700 }, { "epoch": 0.1, "learning_rate": 4.503839001110116e-05, "loss": 2.9002, "step": 27800 }, { "epoch": 0.1, "learning_rate": 4.502054249315548e-05, "loss": 2.9018, "step": 27900 }, { "epoch": 0.1, "learning_rate": 4.50026949752098e-05, "loss": 2.8991, "step": 28000 }, { "epoch": 0.1, "learning_rate": 4.498484745726412e-05, "loss": 2.9029, "step": 28100 }, { "epoch": 0.1, "learning_rate": 4.496699993931844e-05, "loss": 2.9135, "step": 28200 }, { "epoch": 0.1, "learning_rate": 4.4949152421372764e-05, "loss": 2.9018, "step": 28300 }, { "epoch": 0.1, "learning_rate": 4.4931304903427085e-05, "loss": 2.9132, "step": 28400 }, { "epoch": 0.1, "learning_rate": 4.4913457385481406e-05, "loss": 2.915, "step": 28500 }, { "epoch": 0.1, "learning_rate": 4.489560986753573e-05, "loss": 2.8909, "step": 28600 }, { "epoch": 0.1, "learning_rate": 4.487776234959005e-05, "loss": 2.9024, "step": 28700 }, { "epoch": 0.1, "learning_rate": 4.485991483164437e-05, "loss": 2.8926, "step": 28800 }, { "epoch": 0.1, "learning_rate": 4.484206731369869e-05, "loss": 2.8982, "step": 28900 }, { "epoch": 0.1, "learning_rate": 4.4824219795753006e-05, "loss": 2.9044, "step": 29000 }, { "epoch": 0.1, "learning_rate": 4.480637227780733e-05, "loss": 2.9007, "step": 29100 }, { "epoch": 0.1, "learning_rate": 4.478852475986165e-05, "loss": 2.8998, "step": 29200 }, { "epoch": 0.1, "learning_rate": 4.477067724191597e-05, "loss": 2.9021, "step": 29300 }, { "epoch": 0.1, "learning_rate": 4.475282972397029e-05, "loss": 2.9067, "step": 29400 }, { "epoch": 0.11, "learning_rate": 4.473498220602461e-05, "loss": 2.9029, "step": 29500 }, { "epoch": 0.11, "learning_rate": 4.471713468807893e-05, "loss": 2.8996, "step": 29600 }, { "epoch": 0.11, "learning_rate": 4.4699287170133254e-05, "loss": 2.8947, "step": 29700 }, { "epoch": 0.11, "learning_rate": 4.4681439652187575e-05, "loss": 2.8999, "step": 29800 }, { "epoch": 0.11, "learning_rate": 4.4663592134241897e-05, "loss": 2.8911, "step": 29900 }, { "epoch": 0.11, "learning_rate": 4.464574461629622e-05, "loss": 2.8956, "step": 30000 }, { "epoch": 0.11, "learning_rate": 4.462789709835053e-05, "loss": 2.9037, "step": 30100 }, { "epoch": 0.11, "learning_rate": 4.4610049580404853e-05, "loss": 2.888, "step": 30200 }, { "epoch": 0.11, "learning_rate": 4.4592202062459175e-05, "loss": 2.8946, "step": 30300 }, { "epoch": 0.11, "learning_rate": 4.4574354544513496e-05, "loss": 2.9023, "step": 30400 }, { "epoch": 0.11, "learning_rate": 4.455650702656782e-05, "loss": 2.9015, "step": 30500 }, { "epoch": 0.11, "learning_rate": 4.453865950862214e-05, "loss": 2.8981, "step": 30600 }, { "epoch": 0.11, "learning_rate": 4.452081199067646e-05, "loss": 2.9012, "step": 30700 }, { "epoch": 0.11, "learning_rate": 4.450296447273078e-05, "loss": 2.9027, "step": 30800 }, { "epoch": 0.11, "learning_rate": 4.44851169547851e-05, "loss": 2.8998, "step": 30900 }, { "epoch": 0.11, "learning_rate": 4.446726943683942e-05, "loss": 2.896, "step": 31000 }, { "epoch": 0.11, "learning_rate": 4.4449421918893744e-05, "loss": 2.8943, "step": 31100 }, { "epoch": 0.11, "learning_rate": 4.443157440094806e-05, "loss": 2.8959, "step": 31200 }, { "epoch": 0.11, "learning_rate": 4.441372688300238e-05, "loss": 2.8864, "step": 31300 }, { "epoch": 0.11, "learning_rate": 4.43958793650567e-05, "loss": 2.8969, "step": 31400 }, { "epoch": 0.11, "learning_rate": 4.437803184711102e-05, "loss": 2.8877, "step": 31500 }, { "epoch": 0.11, "learning_rate": 4.4360184329165344e-05, "loss": 2.888, "step": 31600 }, { "epoch": 0.11, "learning_rate": 4.4342336811219665e-05, "loss": 2.8799, "step": 31700 }, { "epoch": 0.11, "learning_rate": 4.4324489293273986e-05, "loss": 2.8901, "step": 31800 }, { "epoch": 0.11, "learning_rate": 4.430664177532831e-05, "loss": 2.8944, "step": 31900 }, { "epoch": 0.11, "learning_rate": 4.428879425738263e-05, "loss": 2.8914, "step": 32000 }, { "epoch": 0.11, "learning_rate": 4.427094673943695e-05, "loss": 2.8816, "step": 32100 }, { "epoch": 0.11, "learning_rate": 4.425309922149127e-05, "loss": 2.8752, "step": 32200 }, { "epoch": 0.12, "learning_rate": 4.4235251703545585e-05, "loss": 2.8947, "step": 32300 }, { "epoch": 0.12, "learning_rate": 4.421740418559991e-05, "loss": 2.8882, "step": 32400 }, { "epoch": 0.12, "learning_rate": 4.419955666765423e-05, "loss": 2.8851, "step": 32500 }, { "epoch": 0.12, "learning_rate": 4.418170914970855e-05, "loss": 2.8871, "step": 32600 }, { "epoch": 0.12, "learning_rate": 4.416386163176287e-05, "loss": 2.8999, "step": 32700 }, { "epoch": 0.12, "learning_rate": 4.414601411381719e-05, "loss": 2.8968, "step": 32800 }, { "epoch": 0.12, "learning_rate": 4.412816659587151e-05, "loss": 2.8961, "step": 32900 }, { "epoch": 0.12, "learning_rate": 4.4110319077925834e-05, "loss": 2.8843, "step": 33000 }, { "epoch": 0.12, "learning_rate": 4.4092471559980155e-05, "loss": 2.8937, "step": 33100 }, { "epoch": 0.12, "learning_rate": 4.4074624042034476e-05, "loss": 2.8943, "step": 33200 }, { "epoch": 0.12, "learning_rate": 4.40567765240888e-05, "loss": 2.8934, "step": 33300 }, { "epoch": 0.12, "learning_rate": 4.403892900614312e-05, "loss": 2.8827, "step": 33400 }, { "epoch": 0.12, "learning_rate": 4.402108148819744e-05, "loss": 2.8842, "step": 33500 }, { "epoch": 0.12, "learning_rate": 4.400323397025176e-05, "loss": 2.897, "step": 33600 }, { "epoch": 0.12, "learning_rate": 4.398538645230608e-05, "loss": 2.8934, "step": 33700 }, { "epoch": 0.12, "learning_rate": 4.3967538934360404e-05, "loss": 2.8872, "step": 33800 }, { "epoch": 0.12, "learning_rate": 4.3949691416414725e-05, "loss": 2.8913, "step": 33900 }, { "epoch": 0.12, "learning_rate": 4.3931843898469046e-05, "loss": 2.8967, "step": 34000 }, { "epoch": 0.12, "learning_rate": 4.391399638052337e-05, "loss": 2.8891, "step": 34100 }, { "epoch": 0.12, "learning_rate": 4.389614886257769e-05, "loss": 2.8905, "step": 34200 }, { "epoch": 0.12, "learning_rate": 4.3878301344632e-05, "loss": 2.8873, "step": 34300 }, { "epoch": 0.12, "learning_rate": 4.3860453826686324e-05, "loss": 2.8666, "step": 34400 }, { "epoch": 0.12, "learning_rate": 4.3842606308740645e-05, "loss": 2.8838, "step": 34500 }, { "epoch": 0.12, "learning_rate": 4.3824758790794967e-05, "loss": 2.8856, "step": 34600 }, { "epoch": 0.12, "learning_rate": 4.380691127284929e-05, "loss": 2.8823, "step": 34700 }, { "epoch": 0.12, "learning_rate": 4.378906375490361e-05, "loss": 2.8862, "step": 34800 }, { "epoch": 0.12, "learning_rate": 4.377121623695793e-05, "loss": 2.8845, "step": 34900 }, { "epoch": 0.12, "learning_rate": 4.375336871901225e-05, "loss": 2.8771, "step": 35000 }, { "epoch": 0.13, "learning_rate": 4.373552120106657e-05, "loss": 2.8801, "step": 35100 }, { "epoch": 0.13, "learning_rate": 4.3717673683120894e-05, "loss": 2.8853, "step": 35200 }, { "epoch": 0.13, "learning_rate": 4.369982616517521e-05, "loss": 2.8846, "step": 35300 }, { "epoch": 0.13, "learning_rate": 4.368197864722953e-05, "loss": 2.8738, "step": 35400 }, { "epoch": 0.13, "learning_rate": 4.366413112928385e-05, "loss": 2.8921, "step": 35500 }, { "epoch": 0.13, "learning_rate": 4.364628361133817e-05, "loss": 2.8795, "step": 35600 }, { "epoch": 0.13, "learning_rate": 4.362843609339249e-05, "loss": 2.8811, "step": 35700 }, { "epoch": 0.13, "learning_rate": 4.3610588575446814e-05, "loss": 2.8788, "step": 35800 }, { "epoch": 0.13, "learning_rate": 4.3592741057501136e-05, "loss": 2.8811, "step": 35900 }, { "epoch": 0.13, "learning_rate": 4.357489353955546e-05, "loss": 2.8831, "step": 36000 }, { "epoch": 0.13, "learning_rate": 4.355704602160978e-05, "loss": 2.8804, "step": 36100 }, { "epoch": 0.13, "learning_rate": 4.35391985036641e-05, "loss": 2.8745, "step": 36200 }, { "epoch": 0.13, "learning_rate": 4.352135098571842e-05, "loss": 2.8806, "step": 36300 }, { "epoch": 0.13, "learning_rate": 4.3503503467772735e-05, "loss": 2.889, "step": 36400 }, { "epoch": 0.13, "learning_rate": 4.3485655949827056e-05, "loss": 2.8757, "step": 36500 }, { "epoch": 0.13, "learning_rate": 4.346780843188138e-05, "loss": 2.8937, "step": 36600 }, { "epoch": 0.13, "learning_rate": 4.34499609139357e-05, "loss": 2.8649, "step": 36700 }, { "epoch": 0.13, "learning_rate": 4.343211339599002e-05, "loss": 2.8782, "step": 36800 }, { "epoch": 0.13, "learning_rate": 4.341426587804434e-05, "loss": 2.8843, "step": 36900 }, { "epoch": 0.13, "learning_rate": 4.339641836009866e-05, "loss": 2.8823, "step": 37000 }, { "epoch": 0.13, "learning_rate": 4.3378570842152983e-05, "loss": 2.8677, "step": 37100 }, { "epoch": 0.13, "learning_rate": 4.3360723324207305e-05, "loss": 2.8788, "step": 37200 }, { "epoch": 0.13, "learning_rate": 4.3342875806261626e-05, "loss": 2.8773, "step": 37300 }, { "epoch": 0.13, "learning_rate": 4.332502828831595e-05, "loss": 2.8774, "step": 37400 }, { "epoch": 0.13, "learning_rate": 4.330718077037026e-05, "loss": 2.8756, "step": 37500 }, { "epoch": 0.13, "learning_rate": 4.328933325242458e-05, "loss": 2.8975, "step": 37600 }, { "epoch": 0.13, "learning_rate": 4.3271485734478904e-05, "loss": 2.8696, "step": 37700 }, { "epoch": 0.13, "learning_rate": 4.3253638216533225e-05, "loss": 2.8816, "step": 37800 }, { "epoch": 0.14, "learning_rate": 4.3235790698587546e-05, "loss": 2.8705, "step": 37900 }, { "epoch": 0.14, "learning_rate": 4.321794318064187e-05, "loss": 2.8702, "step": 38000 }, { "epoch": 0.14, "learning_rate": 4.320009566269619e-05, "loss": 2.8791, "step": 38100 }, { "epoch": 0.14, "learning_rate": 4.318224814475051e-05, "loss": 2.8812, "step": 38200 }, { "epoch": 0.14, "learning_rate": 4.316440062680484e-05, "loss": 2.868, "step": 38300 }, { "epoch": 0.14, "learning_rate": 4.314655310885915e-05, "loss": 2.881, "step": 38400 }, { "epoch": 0.14, "learning_rate": 4.3128705590913474e-05, "loss": 2.8783, "step": 38500 }, { "epoch": 0.14, "learning_rate": 4.3110858072967795e-05, "loss": 2.8848, "step": 38600 }, { "epoch": 0.14, "learning_rate": 4.3093010555022116e-05, "loss": 2.8771, "step": 38700 }, { "epoch": 0.14, "learning_rate": 4.307516303707644e-05, "loss": 2.8584, "step": 38800 }, { "epoch": 0.14, "learning_rate": 4.305731551913076e-05, "loss": 2.8824, "step": 38900 }, { "epoch": 0.14, "learning_rate": 4.303946800118508e-05, "loss": 2.879, "step": 39000 }, { "epoch": 0.14, "learning_rate": 4.30216204832394e-05, "loss": 2.8753, "step": 39100 }, { "epoch": 0.14, "learning_rate": 4.300377296529372e-05, "loss": 2.8864, "step": 39200 }, { "epoch": 0.14, "learning_rate": 4.298592544734804e-05, "loss": 2.8741, "step": 39300 }, { "epoch": 0.14, "learning_rate": 4.2968077929402365e-05, "loss": 2.8719, "step": 39400 }, { "epoch": 0.14, "learning_rate": 4.295023041145668e-05, "loss": 2.8661, "step": 39500 }, { "epoch": 0.14, "learning_rate": 4.2932382893511e-05, "loss": 2.8768, "step": 39600 }, { "epoch": 0.14, "learning_rate": 4.291453537556532e-05, "loss": 2.858, "step": 39700 }, { "epoch": 0.14, "learning_rate": 4.289668785761964e-05, "loss": 2.8816, "step": 39800 }, { "epoch": 0.14, "learning_rate": 4.2878840339673964e-05, "loss": 2.8814, "step": 39900 }, { "epoch": 0.14, "learning_rate": 4.2860992821728285e-05, "loss": 2.8632, "step": 40000 }, { "epoch": 0.14, "learning_rate": 4.2843145303782606e-05, "loss": 2.8693, "step": 40100 }, { "epoch": 0.14, "learning_rate": 4.282529778583693e-05, "loss": 2.8715, "step": 40200 }, { "epoch": 0.14, "learning_rate": 4.280745026789125e-05, "loss": 2.8737, "step": 40300 }, { "epoch": 0.14, "learning_rate": 4.278960274994557e-05, "loss": 2.8663, "step": 40400 }, { "epoch": 0.14, "learning_rate": 4.277175523199989e-05, "loss": 2.8713, "step": 40500 }, { "epoch": 0.14, "learning_rate": 4.2753907714054206e-05, "loss": 2.869, "step": 40600 }, { "epoch": 0.15, "learning_rate": 4.273606019610853e-05, "loss": 2.8638, "step": 40700 }, { "epoch": 0.15, "learning_rate": 4.271821267816285e-05, "loss": 2.8783, "step": 40800 }, { "epoch": 0.15, "learning_rate": 4.270036516021717e-05, "loss": 2.8704, "step": 40900 }, { "epoch": 0.15, "learning_rate": 4.268251764227149e-05, "loss": 2.8665, "step": 41000 }, { "epoch": 0.15, "learning_rate": 4.266467012432581e-05, "loss": 2.8649, "step": 41100 }, { "epoch": 0.15, "learning_rate": 4.264682260638013e-05, "loss": 2.8722, "step": 41200 }, { "epoch": 0.15, "learning_rate": 4.2628975088434454e-05, "loss": 2.8696, "step": 41300 }, { "epoch": 0.15, "learning_rate": 4.2611127570488775e-05, "loss": 2.8669, "step": 41400 }, { "epoch": 0.15, "learning_rate": 4.2593280052543097e-05, "loss": 2.8631, "step": 41500 }, { "epoch": 0.15, "learning_rate": 4.257543253459742e-05, "loss": 2.8778, "step": 41600 }, { "epoch": 0.15, "learning_rate": 4.255758501665173e-05, "loss": 2.8643, "step": 41700 }, { "epoch": 0.15, "learning_rate": 4.2539737498706053e-05, "loss": 2.854, "step": 41800 }, { "epoch": 0.15, "learning_rate": 4.2521889980760375e-05, "loss": 2.8628, "step": 41900 }, { "epoch": 0.15, "learning_rate": 4.2504042462814696e-05, "loss": 2.8594, "step": 42000 }, { "epoch": 0.15, "learning_rate": 4.248619494486902e-05, "loss": 2.8626, "step": 42100 }, { "epoch": 0.15, "learning_rate": 4.246834742692334e-05, "loss": 2.8617, "step": 42200 }, { "epoch": 0.15, "learning_rate": 4.245049990897766e-05, "loss": 2.8608, "step": 42300 }, { "epoch": 0.15, "learning_rate": 4.243265239103198e-05, "loss": 2.8686, "step": 42400 }, { "epoch": 0.15, "learning_rate": 4.24148048730863e-05, "loss": 2.8715, "step": 42500 }, { "epoch": 0.15, "learning_rate": 4.239695735514062e-05, "loss": 2.8743, "step": 42600 }, { "epoch": 0.15, "learning_rate": 4.2379109837194944e-05, "loss": 2.8715, "step": 42700 }, { "epoch": 0.15, "learning_rate": 4.236126231924926e-05, "loss": 2.8652, "step": 42800 }, { "epoch": 0.15, "learning_rate": 4.234341480130358e-05, "loss": 2.8489, "step": 42900 }, { "epoch": 0.15, "learning_rate": 4.23255672833579e-05, "loss": 2.8605, "step": 43000 }, { "epoch": 0.15, "learning_rate": 4.230771976541222e-05, "loss": 2.8606, "step": 43100 }, { "epoch": 0.15, "learning_rate": 4.2289872247466544e-05, "loss": 2.8656, "step": 43200 }, { "epoch": 0.15, "learning_rate": 4.2272024729520865e-05, "loss": 2.8607, "step": 43300 }, { "epoch": 0.15, "learning_rate": 4.2254177211575186e-05, "loss": 2.8614, "step": 43400 }, { "epoch": 0.16, "learning_rate": 4.2236329693629514e-05, "loss": 2.858, "step": 43500 }, { "epoch": 0.16, "learning_rate": 4.2218482175683835e-05, "loss": 2.8582, "step": 43600 }, { "epoch": 0.16, "learning_rate": 4.220063465773815e-05, "loss": 2.8664, "step": 43700 }, { "epoch": 0.16, "learning_rate": 4.218278713979247e-05, "loss": 2.867, "step": 43800 }, { "epoch": 0.16, "learning_rate": 4.216493962184679e-05, "loss": 2.8731, "step": 43900 }, { "epoch": 0.16, "learning_rate": 4.214709210390111e-05, "loss": 2.8612, "step": 44000 }, { "epoch": 0.16, "learning_rate": 4.2129244585955435e-05, "loss": 2.8553, "step": 44100 }, { "epoch": 0.16, "learning_rate": 4.2111397068009756e-05, "loss": 2.8642, "step": 44200 }, { "epoch": 0.16, "learning_rate": 4.209354955006408e-05, "loss": 2.8693, "step": 44300 }, { "epoch": 0.16, "learning_rate": 4.20757020321184e-05, "loss": 2.8573, "step": 44400 }, { "epoch": 0.16, "learning_rate": 4.205785451417272e-05, "loss": 2.8646, "step": 44500 }, { "epoch": 0.16, "learning_rate": 4.204000699622704e-05, "loss": 2.863, "step": 44600 }, { "epoch": 0.16, "learning_rate": 4.202215947828136e-05, "loss": 2.8674, "step": 44700 }, { "epoch": 0.16, "learning_rate": 4.2004311960335676e-05, "loss": 2.8609, "step": 44800 }, { "epoch": 0.16, "learning_rate": 4.198646444239e-05, "loss": 2.8604, "step": 44900 }, { "epoch": 0.16, "learning_rate": 4.196861692444432e-05, "loss": 2.8587, "step": 45000 }, { "epoch": 0.16, "learning_rate": 4.195076940649864e-05, "loss": 2.8611, "step": 45100 }, { "epoch": 0.16, "learning_rate": 4.193292188855296e-05, "loss": 2.8605, "step": 45200 }, { "epoch": 0.16, "learning_rate": 4.191507437060728e-05, "loss": 2.8599, "step": 45300 }, { "epoch": 0.16, "learning_rate": 4.1897226852661604e-05, "loss": 2.8597, "step": 45400 }, { "epoch": 0.16, "learning_rate": 4.1879379334715925e-05, "loss": 2.8613, "step": 45500 }, { "epoch": 0.16, "learning_rate": 4.1861531816770246e-05, "loss": 2.8523, "step": 45600 }, { "epoch": 0.16, "learning_rate": 4.184368429882457e-05, "loss": 2.8463, "step": 45700 }, { "epoch": 0.16, "learning_rate": 4.182583678087889e-05, "loss": 2.8558, "step": 45800 }, { "epoch": 0.16, "learning_rate": 4.18079892629332e-05, "loss": 2.854, "step": 45900 }, { "epoch": 0.16, "learning_rate": 4.1790141744987524e-05, "loss": 2.8539, "step": 46000 }, { "epoch": 0.16, "learning_rate": 4.1772294227041845e-05, "loss": 2.8612, "step": 46100 }, { "epoch": 0.16, "learning_rate": 4.1754446709096167e-05, "loss": 2.8571, "step": 46200 }, { "epoch": 0.17, "learning_rate": 4.173659919115049e-05, "loss": 2.8656, "step": 46300 }, { "epoch": 0.17, "learning_rate": 4.171875167320481e-05, "loss": 2.8571, "step": 46400 }, { "epoch": 0.17, "learning_rate": 4.170090415525913e-05, "loss": 2.8538, "step": 46500 }, { "epoch": 0.17, "learning_rate": 4.168305663731345e-05, "loss": 2.8643, "step": 46600 }, { "epoch": 0.17, "learning_rate": 4.166520911936777e-05, "loss": 2.8593, "step": 46700 }, { "epoch": 0.17, "learning_rate": 4.1647361601422094e-05, "loss": 2.8608, "step": 46800 }, { "epoch": 0.17, "learning_rate": 4.1629514083476415e-05, "loss": 2.8656, "step": 46900 }, { "epoch": 0.17, "learning_rate": 4.161166656553073e-05, "loss": 2.8556, "step": 47000 }, { "epoch": 0.17, "learning_rate": 4.159381904758505e-05, "loss": 2.8533, "step": 47100 }, { "epoch": 0.17, "learning_rate": 4.157597152963937e-05, "loss": 2.8573, "step": 47200 }, { "epoch": 0.17, "learning_rate": 4.155812401169369e-05, "loss": 2.8553, "step": 47300 }, { "epoch": 0.17, "learning_rate": 4.1540276493748014e-05, "loss": 2.8608, "step": 47400 }, { "epoch": 0.17, "learning_rate": 4.1522428975802336e-05, "loss": 2.8571, "step": 47500 }, { "epoch": 0.17, "learning_rate": 4.150458145785666e-05, "loss": 2.8527, "step": 47600 }, { "epoch": 0.17, "learning_rate": 4.148673393991098e-05, "loss": 2.8564, "step": 47700 }, { "epoch": 0.17, "learning_rate": 4.14688864219653e-05, "loss": 2.8488, "step": 47800 }, { "epoch": 0.17, "learning_rate": 4.145103890401962e-05, "loss": 2.8546, "step": 47900 }, { "epoch": 0.17, "learning_rate": 4.143319138607394e-05, "loss": 2.8542, "step": 48000 }, { "epoch": 0.17, "learning_rate": 4.1415343868128256e-05, "loss": 2.8474, "step": 48100 }, { "epoch": 0.17, "learning_rate": 4.139749635018258e-05, "loss": 2.8504, "step": 48200 }, { "epoch": 0.17, "learning_rate": 4.13796488322369e-05, "loss": 2.858, "step": 48300 }, { "epoch": 0.17, "learning_rate": 4.136180131429122e-05, "loss": 2.847, "step": 48400 }, { "epoch": 0.17, "learning_rate": 4.134395379634554e-05, "loss": 2.8431, "step": 48500 }, { "epoch": 0.17, "learning_rate": 4.132610627839987e-05, "loss": 2.8534, "step": 48600 }, { "epoch": 0.17, "learning_rate": 4.130825876045419e-05, "loss": 2.8457, "step": 48700 }, { "epoch": 0.17, "learning_rate": 4.129041124250851e-05, "loss": 2.8611, "step": 48800 }, { "epoch": 0.17, "learning_rate": 4.127256372456283e-05, "loss": 2.8424, "step": 48900 }, { "epoch": 0.17, "learning_rate": 4.125471620661715e-05, "loss": 2.8518, "step": 49000 }, { "epoch": 0.18, "learning_rate": 4.123686868867147e-05, "loss": 2.8601, "step": 49100 }, { "epoch": 0.18, "learning_rate": 4.121902117072579e-05, "loss": 2.8475, "step": 49200 }, { "epoch": 0.18, "learning_rate": 4.120117365278011e-05, "loss": 2.8561, "step": 49300 }, { "epoch": 0.18, "learning_rate": 4.118332613483443e-05, "loss": 2.8666, "step": 49400 }, { "epoch": 0.18, "learning_rate": 4.116547861688875e-05, "loss": 2.8522, "step": 49500 }, { "epoch": 0.18, "learning_rate": 4.1147631098943074e-05, "loss": 2.8481, "step": 49600 }, { "epoch": 0.18, "learning_rate": 4.1129783580997395e-05, "loss": 2.8584, "step": 49700 }, { "epoch": 0.18, "learning_rate": 4.111193606305172e-05, "loss": 2.8557, "step": 49800 }, { "epoch": 0.18, "learning_rate": 4.109408854510604e-05, "loss": 2.8509, "step": 49900 }, { "epoch": 0.18, "learning_rate": 4.107624102716036e-05, "loss": 2.8536, "step": 50000 }, { "epoch": 0.18, "learning_rate": 4.1058393509214674e-05, "loss": 2.843, "step": 50100 }, { "epoch": 0.18, "learning_rate": 4.1040545991268995e-05, "loss": 2.848, "step": 50200 }, { "epoch": 0.18, "learning_rate": 4.1022698473323316e-05, "loss": 2.8521, "step": 50300 }, { "epoch": 0.18, "learning_rate": 4.100485095537764e-05, "loss": 2.8368, "step": 50400 }, { "epoch": 0.18, "learning_rate": 4.098700343743196e-05, "loss": 2.8653, "step": 50500 }, { "epoch": 0.18, "learning_rate": 4.096915591948628e-05, "loss": 2.8587, "step": 50600 }, { "epoch": 0.18, "learning_rate": 4.09513084015406e-05, "loss": 2.8547, "step": 50700 }, { "epoch": 0.18, "learning_rate": 4.093346088359492e-05, "loss": 2.847, "step": 50800 }, { "epoch": 0.18, "learning_rate": 4.091561336564924e-05, "loss": 2.8457, "step": 50900 }, { "epoch": 0.18, "learning_rate": 4.0897765847703565e-05, "loss": 2.8517, "step": 51000 }, { "epoch": 0.18, "learning_rate": 4.0879918329757886e-05, "loss": 2.8362, "step": 51100 }, { "epoch": 0.18, "learning_rate": 4.08620708118122e-05, "loss": 2.8562, "step": 51200 }, { "epoch": 0.18, "learning_rate": 4.084422329386652e-05, "loss": 2.8518, "step": 51300 }, { "epoch": 0.18, "learning_rate": 4.082637577592084e-05, "loss": 2.8371, "step": 51400 }, { "epoch": 0.18, "learning_rate": 4.0808528257975164e-05, "loss": 2.8484, "step": 51500 }, { "epoch": 0.18, "learning_rate": 4.0790680740029485e-05, "loss": 2.8457, "step": 51600 }, { "epoch": 0.18, "learning_rate": 4.0772833222083806e-05, "loss": 2.843, "step": 51700 }, { "epoch": 0.18, "learning_rate": 4.075498570413813e-05, "loss": 2.8378, "step": 51800 }, { "epoch": 0.19, "learning_rate": 4.073713818619245e-05, "loss": 2.8577, "step": 51900 }, { "epoch": 0.19, "learning_rate": 4.071929066824677e-05, "loss": 2.8458, "step": 52000 }, { "epoch": 0.19, "learning_rate": 4.070144315030109e-05, "loss": 2.8408, "step": 52100 }, { "epoch": 0.19, "learning_rate": 4.068359563235541e-05, "loss": 2.8462, "step": 52200 }, { "epoch": 0.19, "learning_rate": 4.066574811440973e-05, "loss": 2.828, "step": 52300 }, { "epoch": 0.19, "learning_rate": 4.064790059646405e-05, "loss": 2.8387, "step": 52400 }, { "epoch": 0.19, "learning_rate": 4.063005307851837e-05, "loss": 2.8412, "step": 52500 }, { "epoch": 0.19, "learning_rate": 4.061220556057269e-05, "loss": 2.8409, "step": 52600 }, { "epoch": 0.19, "learning_rate": 4.059435804262701e-05, "loss": 2.8331, "step": 52700 }, { "epoch": 0.19, "learning_rate": 4.057651052468133e-05, "loss": 2.8533, "step": 52800 }, { "epoch": 0.19, "learning_rate": 4.0558663006735654e-05, "loss": 2.8419, "step": 52900 }, { "epoch": 0.19, "learning_rate": 4.0540815488789975e-05, "loss": 2.844, "step": 53000 }, { "epoch": 0.19, "learning_rate": 4.0522967970844296e-05, "loss": 2.8462, "step": 53100 }, { "epoch": 0.19, "learning_rate": 4.050512045289862e-05, "loss": 2.8403, "step": 53200 }, { "epoch": 0.19, "learning_rate": 4.048727293495294e-05, "loss": 2.8335, "step": 53300 }, { "epoch": 0.19, "learning_rate": 4.046942541700725e-05, "loss": 2.8462, "step": 53400 }, { "epoch": 0.19, "learning_rate": 4.0451577899061575e-05, "loss": 2.838, "step": 53500 }, { "epoch": 0.19, "learning_rate": 4.0433730381115896e-05, "loss": 2.8352, "step": 53600 }, { "epoch": 0.19, "learning_rate": 4.0415882863170224e-05, "loss": 2.8374, "step": 53700 }, { "epoch": 0.19, "learning_rate": 4.0398035345224545e-05, "loss": 2.8347, "step": 53800 }, { "epoch": 0.19, "learning_rate": 4.0380187827278866e-05, "loss": 2.8214, "step": 53900 }, { "epoch": 0.19, "learning_rate": 4.036234030933319e-05, "loss": 2.849, "step": 54000 }, { "epoch": 0.19, "learning_rate": 4.034449279138751e-05, "loss": 2.8423, "step": 54100 }, { "epoch": 0.19, "learning_rate": 4.032664527344183e-05, "loss": 2.8396, "step": 54200 }, { "epoch": 0.19, "learning_rate": 4.0308797755496144e-05, "loss": 2.8333, "step": 54300 }, { "epoch": 0.19, "learning_rate": 4.0290950237550465e-05, "loss": 2.8334, "step": 54400 }, { "epoch": 0.19, "learning_rate": 4.027310271960479e-05, "loss": 2.8289, "step": 54500 }, { "epoch": 0.19, "learning_rate": 4.025525520165911e-05, "loss": 2.8505, "step": 54600 }, { "epoch": 0.2, "learning_rate": 4.023740768371343e-05, "loss": 2.8387, "step": 54700 }, { "epoch": 0.2, "learning_rate": 4.021956016576775e-05, "loss": 2.8407, "step": 54800 }, { "epoch": 0.2, "learning_rate": 4.020171264782207e-05, "loss": 2.8396, "step": 54900 }, { "epoch": 0.2, "learning_rate": 4.018386512987639e-05, "loss": 2.8391, "step": 55000 }, { "epoch": 0.2, "learning_rate": 4.0166017611930714e-05, "loss": 2.8314, "step": 55100 }, { "epoch": 0.2, "learning_rate": 4.0148170093985035e-05, "loss": 2.8403, "step": 55200 }, { "epoch": 0.2, "learning_rate": 4.0130322576039356e-05, "loss": 2.842, "step": 55300 }, { "epoch": 0.2, "learning_rate": 4.011247505809367e-05, "loss": 2.8382, "step": 55400 }, { "epoch": 0.2, "learning_rate": 4.009462754014799e-05, "loss": 2.8445, "step": 55500 }, { "epoch": 0.2, "learning_rate": 4.007678002220231e-05, "loss": 2.8377, "step": 55600 }, { "epoch": 0.2, "learning_rate": 4.0058932504256634e-05, "loss": 2.8388, "step": 55700 }, { "epoch": 0.2, "learning_rate": 4.0041084986310956e-05, "loss": 2.8286, "step": 55800 }, { "epoch": 0.2, "learning_rate": 4.002323746836528e-05, "loss": 2.8204, "step": 55900 }, { "epoch": 0.2, "learning_rate": 4.00053899504196e-05, "loss": 2.8434, "step": 56000 }, { "epoch": 0.2, "learning_rate": 3.998754243247392e-05, "loss": 2.8375, "step": 56100 }, { "epoch": 0.2, "learning_rate": 3.996969491452824e-05, "loss": 2.8283, "step": 56200 }, { "epoch": 0.2, "learning_rate": 3.995184739658256e-05, "loss": 2.8268, "step": 56300 }, { "epoch": 0.2, "learning_rate": 3.993399987863688e-05, "loss": 2.8338, "step": 56400 }, { "epoch": 0.2, "learning_rate": 3.99161523606912e-05, "loss": 2.8414, "step": 56500 }, { "epoch": 0.2, "learning_rate": 3.989830484274552e-05, "loss": 2.8335, "step": 56600 }, { "epoch": 0.2, "learning_rate": 3.988045732479984e-05, "loss": 2.8301, "step": 56700 }, { "epoch": 0.2, "learning_rate": 3.986260980685416e-05, "loss": 2.8353, "step": 56800 }, { "epoch": 0.2, "learning_rate": 3.984476228890848e-05, "loss": 2.8319, "step": 56900 }, { "epoch": 0.2, "learning_rate": 3.9826914770962804e-05, "loss": 2.8368, "step": 57000 }, { "epoch": 0.2, "learning_rate": 3.9809067253017125e-05, "loss": 2.8305, "step": 57100 }, { "epoch": 0.2, "learning_rate": 3.9791219735071446e-05, "loss": 2.8269, "step": 57200 }, { "epoch": 0.2, "learning_rate": 3.977337221712577e-05, "loss": 2.8326, "step": 57300 }, { "epoch": 0.2, "learning_rate": 3.975552469918009e-05, "loss": 2.8347, "step": 57400 }, { "epoch": 0.21, "learning_rate": 3.97376771812344e-05, "loss": 2.8247, "step": 57500 }, { "epoch": 0.21, "learning_rate": 3.9719829663288724e-05, "loss": 2.8293, "step": 57600 }, { "epoch": 0.21, "learning_rate": 3.9701982145343045e-05, "loss": 2.8288, "step": 57700 }, { "epoch": 0.21, "learning_rate": 3.9684134627397366e-05, "loss": 2.8195, "step": 57800 }, { "epoch": 0.21, "learning_rate": 3.966628710945169e-05, "loss": 2.8307, "step": 57900 }, { "epoch": 0.21, "learning_rate": 3.964843959150601e-05, "loss": 2.8316, "step": 58000 }, { "epoch": 0.21, "learning_rate": 3.963059207356033e-05, "loss": 2.835, "step": 58100 }, { "epoch": 0.21, "learning_rate": 3.961274455561465e-05, "loss": 2.8219, "step": 58200 }, { "epoch": 0.21, "learning_rate": 3.959489703766897e-05, "loss": 2.8291, "step": 58300 }, { "epoch": 0.21, "learning_rate": 3.9577049519723294e-05, "loss": 2.8261, "step": 58400 }, { "epoch": 0.21, "learning_rate": 3.9559202001777615e-05, "loss": 2.8232, "step": 58500 }, { "epoch": 0.21, "learning_rate": 3.954135448383193e-05, "loss": 2.8356, "step": 58600 }, { "epoch": 0.21, "learning_rate": 3.952350696588625e-05, "loss": 2.8329, "step": 58700 }, { "epoch": 0.21, "learning_rate": 3.950565944794058e-05, "loss": 2.8351, "step": 58800 }, { "epoch": 0.21, "learning_rate": 3.94878119299949e-05, "loss": 2.8262, "step": 58900 }, { "epoch": 0.21, "learning_rate": 3.946996441204922e-05, "loss": 2.8217, "step": 59000 }, { "epoch": 0.21, "learning_rate": 3.945211689410354e-05, "loss": 2.8293, "step": 59100 }, { "epoch": 0.21, "learning_rate": 3.9434269376157863e-05, "loss": 2.8246, "step": 59200 }, { "epoch": 0.21, "learning_rate": 3.9416421858212185e-05, "loss": 2.8263, "step": 59300 }, { "epoch": 0.21, "learning_rate": 3.9398574340266506e-05, "loss": 2.8327, "step": 59400 }, { "epoch": 0.21, "learning_rate": 3.938072682232082e-05, "loss": 2.8302, "step": 59500 }, { "epoch": 0.21, "learning_rate": 3.936287930437514e-05, "loss": 2.8261, "step": 59600 }, { "epoch": 0.21, "learning_rate": 3.934503178642946e-05, "loss": 2.8255, "step": 59700 }, { "epoch": 0.21, "learning_rate": 3.9327184268483784e-05, "loss": 2.8291, "step": 59800 }, { "epoch": 0.21, "learning_rate": 3.9309336750538105e-05, "loss": 2.8335, "step": 59900 }, { "epoch": 0.21, "learning_rate": 3.9291489232592426e-05, "loss": 2.8345, "step": 60000 }, { "epoch": 0.21, "learning_rate": 3.927364171464675e-05, "loss": 2.8267, "step": 60100 }, { "epoch": 0.21, "learning_rate": 3.925579419670107e-05, "loss": 2.8294, "step": 60200 }, { "epoch": 0.22, "learning_rate": 3.923794667875539e-05, "loss": 2.8296, "step": 60300 }, { "epoch": 0.22, "learning_rate": 3.922009916080971e-05, "loss": 2.825, "step": 60400 }, { "epoch": 0.22, "learning_rate": 3.920225164286403e-05, "loss": 2.8255, "step": 60500 }, { "epoch": 0.22, "learning_rate": 3.918440412491835e-05, "loss": 2.8265, "step": 60600 }, { "epoch": 0.22, "learning_rate": 3.916655660697267e-05, "loss": 2.824, "step": 60700 }, { "epoch": 0.22, "learning_rate": 3.914870908902699e-05, "loss": 2.8239, "step": 60800 }, { "epoch": 0.22, "learning_rate": 3.913086157108131e-05, "loss": 2.8292, "step": 60900 }, { "epoch": 0.22, "learning_rate": 3.911301405313563e-05, "loss": 2.8268, "step": 61000 }, { "epoch": 0.22, "learning_rate": 3.909516653518995e-05, "loss": 2.8234, "step": 61100 }, { "epoch": 0.22, "learning_rate": 3.9077319017244274e-05, "loss": 2.8276, "step": 61200 }, { "epoch": 0.22, "learning_rate": 3.9059471499298595e-05, "loss": 2.8208, "step": 61300 }, { "epoch": 0.22, "learning_rate": 3.904162398135292e-05, "loss": 2.8301, "step": 61400 }, { "epoch": 0.22, "learning_rate": 3.902377646340724e-05, "loss": 2.8209, "step": 61500 }, { "epoch": 0.22, "learning_rate": 3.900592894546156e-05, "loss": 2.8219, "step": 61600 }, { "epoch": 0.22, "learning_rate": 3.8988081427515874e-05, "loss": 2.8197, "step": 61700 }, { "epoch": 0.22, "learning_rate": 3.8970233909570195e-05, "loss": 2.828, "step": 61800 }, { "epoch": 0.22, "learning_rate": 3.8952386391624516e-05, "loss": 2.8193, "step": 61900 }, { "epoch": 0.22, "learning_rate": 3.893453887367884e-05, "loss": 2.824, "step": 62000 }, { "epoch": 0.22, "learning_rate": 3.891669135573316e-05, "loss": 2.8243, "step": 62100 }, { "epoch": 0.22, "learning_rate": 3.889884383778748e-05, "loss": 2.8251, "step": 62200 }, { "epoch": 0.22, "learning_rate": 3.88809963198418e-05, "loss": 2.8178, "step": 62300 }, { "epoch": 0.22, "learning_rate": 3.886314880189612e-05, "loss": 2.8137, "step": 62400 }, { "epoch": 0.22, "learning_rate": 3.884530128395044e-05, "loss": 2.8303, "step": 62500 }, { "epoch": 0.22, "learning_rate": 3.8827453766004764e-05, "loss": 2.8176, "step": 62600 }, { "epoch": 0.22, "learning_rate": 3.8809606248059086e-05, "loss": 2.8195, "step": 62700 }, { "epoch": 0.22, "learning_rate": 3.87917587301134e-05, "loss": 2.8333, "step": 62800 }, { "epoch": 0.22, "learning_rate": 3.877391121216772e-05, "loss": 2.8282, "step": 62900 }, { "epoch": 0.22, "learning_rate": 3.875606369422204e-05, "loss": 2.8149, "step": 63000 }, { "epoch": 0.23, "learning_rate": 3.8738216176276364e-05, "loss": 2.8329, "step": 63100 }, { "epoch": 0.23, "learning_rate": 3.8720368658330685e-05, "loss": 2.8254, "step": 63200 }, { "epoch": 0.23, "learning_rate": 3.8702521140385006e-05, "loss": 2.8217, "step": 63300 }, { "epoch": 0.23, "learning_rate": 3.868467362243933e-05, "loss": 2.8281, "step": 63400 }, { "epoch": 0.23, "learning_rate": 3.866682610449365e-05, "loss": 2.8274, "step": 63500 }, { "epoch": 0.23, "learning_rate": 3.864897858654797e-05, "loss": 2.8173, "step": 63600 }, { "epoch": 0.23, "learning_rate": 3.863113106860229e-05, "loss": 2.8199, "step": 63700 }, { "epoch": 0.23, "learning_rate": 3.861328355065661e-05, "loss": 2.8304, "step": 63800 }, { "epoch": 0.23, "learning_rate": 3.8595436032710933e-05, "loss": 2.8193, "step": 63900 }, { "epoch": 0.23, "learning_rate": 3.8577588514765255e-05, "loss": 2.8193, "step": 64000 }, { "epoch": 0.23, "learning_rate": 3.8559740996819576e-05, "loss": 2.8161, "step": 64100 }, { "epoch": 0.23, "learning_rate": 3.85418934788739e-05, "loss": 2.822, "step": 64200 }, { "epoch": 0.23, "learning_rate": 3.852404596092822e-05, "loss": 2.8195, "step": 64300 }, { "epoch": 0.23, "learning_rate": 3.850619844298254e-05, "loss": 2.836, "step": 64400 }, { "epoch": 0.23, "learning_rate": 3.848835092503686e-05, "loss": 2.8261, "step": 64500 }, { "epoch": 0.23, "learning_rate": 3.847050340709118e-05, "loss": 2.8232, "step": 64600 }, { "epoch": 0.23, "learning_rate": 3.84526558891455e-05, "loss": 2.8256, "step": 64700 }, { "epoch": 0.23, "learning_rate": 3.843480837119982e-05, "loss": 2.8224, "step": 64800 }, { "epoch": 0.23, "learning_rate": 3.841696085325414e-05, "loss": 2.8168, "step": 64900 }, { "epoch": 0.23, "learning_rate": 3.839911333530846e-05, "loss": 2.8229, "step": 65000 }, { "epoch": 0.23, "learning_rate": 3.838126581736278e-05, "loss": 2.8128, "step": 65100 }, { "epoch": 0.23, "learning_rate": 3.83634182994171e-05, "loss": 2.8177, "step": 65200 }, { "epoch": 0.23, "learning_rate": 3.8345570781471424e-05, "loss": 2.8243, "step": 65300 }, { "epoch": 0.23, "learning_rate": 3.8327723263525745e-05, "loss": 2.8253, "step": 65400 }, { "epoch": 0.23, "learning_rate": 3.8309875745580066e-05, "loss": 2.8213, "step": 65500 }, { "epoch": 0.23, "learning_rate": 3.829202822763439e-05, "loss": 2.8196, "step": 65600 }, { "epoch": 0.23, "learning_rate": 3.827418070968871e-05, "loss": 2.8226, "step": 65700 }, { "epoch": 0.23, "learning_rate": 3.825633319174303e-05, "loss": 2.8216, "step": 65800 }, { "epoch": 0.24, "learning_rate": 3.8238485673797344e-05, "loss": 2.8176, "step": 65900 }, { "epoch": 0.24, "learning_rate": 3.8220638155851665e-05, "loss": 2.8155, "step": 66000 }, { "epoch": 0.24, "learning_rate": 3.820279063790599e-05, "loss": 2.8205, "step": 66100 }, { "epoch": 0.24, "learning_rate": 3.818494311996031e-05, "loss": 2.8173, "step": 66200 }, { "epoch": 0.24, "learning_rate": 3.816709560201463e-05, "loss": 2.8177, "step": 66300 }, { "epoch": 0.24, "learning_rate": 3.814924808406895e-05, "loss": 2.8228, "step": 66400 }, { "epoch": 0.24, "learning_rate": 3.813140056612327e-05, "loss": 2.8189, "step": 66500 }, { "epoch": 0.24, "learning_rate": 3.811355304817759e-05, "loss": 2.8131, "step": 66600 }, { "epoch": 0.24, "learning_rate": 3.8095705530231914e-05, "loss": 2.8143, "step": 66700 }, { "epoch": 0.24, "learning_rate": 3.8077858012286235e-05, "loss": 2.8232, "step": 66800 }, { "epoch": 0.24, "learning_rate": 3.8060010494340556e-05, "loss": 2.8274, "step": 66900 }, { "epoch": 0.24, "learning_rate": 3.804216297639487e-05, "loss": 2.8239, "step": 67000 }, { "epoch": 0.24, "learning_rate": 3.802431545844919e-05, "loss": 2.8201, "step": 67100 }, { "epoch": 0.24, "learning_rate": 3.800646794050351e-05, "loss": 2.8165, "step": 67200 }, { "epoch": 0.24, "learning_rate": 3.7988620422557834e-05, "loss": 2.8003, "step": 67300 }, { "epoch": 0.24, "learning_rate": 3.7970772904612156e-05, "loss": 2.8125, "step": 67400 }, { "epoch": 0.24, "learning_rate": 3.795292538666648e-05, "loss": 2.8208, "step": 67500 }, { "epoch": 0.24, "learning_rate": 3.79350778687208e-05, "loss": 2.8136, "step": 67600 }, { "epoch": 0.24, "learning_rate": 3.791723035077512e-05, "loss": 2.815, "step": 67700 }, { "epoch": 0.24, "learning_rate": 3.789938283282944e-05, "loss": 2.8129, "step": 67800 }, { "epoch": 0.24, "learning_rate": 3.788153531488376e-05, "loss": 2.8263, "step": 67900 }, { "epoch": 0.24, "learning_rate": 3.786368779693808e-05, "loss": 2.8188, "step": 68000 }, { "epoch": 0.24, "learning_rate": 3.78458402789924e-05, "loss": 2.8157, "step": 68100 }, { "epoch": 0.24, "learning_rate": 3.782799276104672e-05, "loss": 2.8161, "step": 68200 }, { "epoch": 0.24, "learning_rate": 3.781014524310104e-05, "loss": 2.8165, "step": 68300 }, { "epoch": 0.24, "learning_rate": 3.779229772515536e-05, "loss": 2.8123, "step": 68400 }, { "epoch": 0.24, "learning_rate": 3.777445020720968e-05, "loss": 2.8163, "step": 68500 }, { "epoch": 0.24, "learning_rate": 3.7756602689264003e-05, "loss": 2.8181, "step": 68600 }, { "epoch": 0.25, "learning_rate": 3.7738755171318325e-05, "loss": 2.8145, "step": 68700 }, { "epoch": 0.25, "learning_rate": 3.7720907653372646e-05, "loss": 2.8149, "step": 68800 }, { "epoch": 0.25, "learning_rate": 3.770306013542697e-05, "loss": 2.811, "step": 68900 }, { "epoch": 0.25, "learning_rate": 3.768521261748129e-05, "loss": 2.8121, "step": 69000 }, { "epoch": 0.25, "learning_rate": 3.766736509953561e-05, "loss": 2.8145, "step": 69100 }, { "epoch": 0.25, "learning_rate": 3.764951758158993e-05, "loss": 2.8179, "step": 69200 }, { "epoch": 0.25, "learning_rate": 3.763167006364425e-05, "loss": 2.8092, "step": 69300 }, { "epoch": 0.25, "learning_rate": 3.761382254569857e-05, "loss": 2.8153, "step": 69400 }, { "epoch": 0.25, "learning_rate": 3.7595975027752894e-05, "loss": 2.8191, "step": 69500 }, { "epoch": 0.25, "learning_rate": 3.7578127509807216e-05, "loss": 2.8241, "step": 69600 }, { "epoch": 0.25, "learning_rate": 3.756027999186154e-05, "loss": 2.812, "step": 69700 }, { "epoch": 0.25, "learning_rate": 3.754243247391586e-05, "loss": 2.8149, "step": 69800 }, { "epoch": 0.25, "learning_rate": 3.752458495597018e-05, "loss": 2.8231, "step": 69900 }, { "epoch": 0.25, "learning_rate": 3.75067374380245e-05, "loss": 2.8172, "step": 70000 }, { "epoch": 0.25, "learning_rate": 3.7488889920078815e-05, "loss": 2.8039, "step": 70100 }, { "epoch": 0.25, "learning_rate": 3.7471042402133136e-05, "loss": 2.8142, "step": 70200 }, { "epoch": 0.25, "learning_rate": 3.745319488418746e-05, "loss": 2.8196, "step": 70300 }, { "epoch": 0.25, "learning_rate": 3.743534736624178e-05, "loss": 2.8097, "step": 70400 }, { "epoch": 0.25, "learning_rate": 3.74174998482961e-05, "loss": 2.8031, "step": 70500 }, { "epoch": 0.25, "learning_rate": 3.739965233035042e-05, "loss": 2.8111, "step": 70600 }, { "epoch": 0.25, "learning_rate": 3.738180481240474e-05, "loss": 2.8112, "step": 70700 }, { "epoch": 0.25, "learning_rate": 3.7363957294459063e-05, "loss": 2.818, "step": 70800 }, { "epoch": 0.25, "learning_rate": 3.7346109776513385e-05, "loss": 2.8037, "step": 70900 }, { "epoch": 0.25, "learning_rate": 3.7328262258567706e-05, "loss": 2.8099, "step": 71000 }, { "epoch": 0.25, "learning_rate": 3.731041474062203e-05, "loss": 2.8041, "step": 71100 }, { "epoch": 0.25, "learning_rate": 3.729256722267634e-05, "loss": 2.8205, "step": 71200 }, { "epoch": 0.25, "learning_rate": 3.727471970473066e-05, "loss": 2.8189, "step": 71300 }, { "epoch": 0.25, "learning_rate": 3.7256872186784984e-05, "loss": 2.8132, "step": 71400 }, { "epoch": 0.26, "learning_rate": 3.7239024668839305e-05, "loss": 2.8115, "step": 71500 }, { "epoch": 0.26, "learning_rate": 3.7221177150893626e-05, "loss": 2.8013, "step": 71600 }, { "epoch": 0.26, "learning_rate": 3.720332963294795e-05, "loss": 2.8026, "step": 71700 }, { "epoch": 0.26, "learning_rate": 3.718548211500227e-05, "loss": 2.8047, "step": 71800 }, { "epoch": 0.26, "learning_rate": 3.716763459705659e-05, "loss": 2.8157, "step": 71900 }, { "epoch": 0.26, "learning_rate": 3.714978707911091e-05, "loss": 2.814, "step": 72000 }, { "epoch": 0.26, "learning_rate": 3.713193956116523e-05, "loss": 2.8161, "step": 72100 }, { "epoch": 0.26, "learning_rate": 3.7114092043219554e-05, "loss": 2.8115, "step": 72200 }, { "epoch": 0.26, "learning_rate": 3.709624452527387e-05, "loss": 2.8193, "step": 72300 }, { "epoch": 0.26, "learning_rate": 3.707839700732819e-05, "loss": 2.8088, "step": 72400 }, { "epoch": 0.26, "learning_rate": 3.706054948938251e-05, "loss": 2.804, "step": 72500 }, { "epoch": 0.26, "learning_rate": 3.704270197143683e-05, "loss": 2.798, "step": 72600 }, { "epoch": 0.26, "learning_rate": 3.702485445349115e-05, "loss": 2.8055, "step": 72700 }, { "epoch": 0.26, "learning_rate": 3.7007006935545474e-05, "loss": 2.8116, "step": 72800 }, { "epoch": 0.26, "learning_rate": 3.6989159417599795e-05, "loss": 2.8163, "step": 72900 }, { "epoch": 0.26, "learning_rate": 3.6971311899654117e-05, "loss": 2.8096, "step": 73000 }, { "epoch": 0.26, "learning_rate": 3.695346438170844e-05, "loss": 2.8119, "step": 73100 }, { "epoch": 0.26, "learning_rate": 3.693561686376276e-05, "loss": 2.8193, "step": 73200 }, { "epoch": 0.26, "learning_rate": 3.691776934581708e-05, "loss": 2.8066, "step": 73300 }, { "epoch": 0.26, "learning_rate": 3.6899921827871395e-05, "loss": 2.8117, "step": 73400 }, { "epoch": 0.26, "learning_rate": 3.6882074309925716e-05, "loss": 2.8091, "step": 73500 }, { "epoch": 0.26, "learning_rate": 3.686422679198004e-05, "loss": 2.8075, "step": 73600 }, { "epoch": 0.26, "learning_rate": 3.684637927403436e-05, "loss": 2.8055, "step": 73700 }, { "epoch": 0.26, "learning_rate": 3.682853175608868e-05, "loss": 2.8076, "step": 73800 }, { "epoch": 0.26, "learning_rate": 3.6810684238143e-05, "loss": 2.8159, "step": 73900 }, { "epoch": 0.26, "learning_rate": 3.679283672019732e-05, "loss": 2.8066, "step": 74000 }, { "epoch": 0.26, "learning_rate": 3.677498920225165e-05, "loss": 2.8005, "step": 74100 }, { "epoch": 0.26, "learning_rate": 3.675714168430597e-05, "loss": 2.809, "step": 74200 }, { "epoch": 0.27, "learning_rate": 3.6739294166360286e-05, "loss": 2.81, "step": 74300 }, { "epoch": 0.27, "learning_rate": 3.672144664841461e-05, "loss": 2.7959, "step": 74400 }, { "epoch": 0.27, "learning_rate": 3.670359913046893e-05, "loss": 2.7977, "step": 74500 }, { "epoch": 0.27, "learning_rate": 3.668575161252325e-05, "loss": 2.8063, "step": 74600 }, { "epoch": 0.27, "learning_rate": 3.666790409457757e-05, "loss": 2.811, "step": 74700 }, { "epoch": 0.27, "learning_rate": 3.665005657663189e-05, "loss": 2.8095, "step": 74800 }, { "epoch": 0.27, "learning_rate": 3.663220905868621e-05, "loss": 2.81, "step": 74900 }, { "epoch": 0.27, "learning_rate": 3.6614361540740534e-05, "loss": 2.8034, "step": 75000 }, { "epoch": 0.27, "learning_rate": 3.6596514022794855e-05, "loss": 2.8054, "step": 75100 }, { "epoch": 0.27, "learning_rate": 3.6578666504849177e-05, "loss": 2.8026, "step": 75200 }, { "epoch": 0.27, "learning_rate": 3.65608189869035e-05, "loss": 2.8075, "step": 75300 }, { "epoch": 0.27, "learning_rate": 3.654297146895781e-05, "loss": 2.8087, "step": 75400 }, { "epoch": 0.27, "learning_rate": 3.6525123951012133e-05, "loss": 2.7983, "step": 75500 }, { "epoch": 0.27, "learning_rate": 3.6507276433066455e-05, "loss": 2.8036, "step": 75600 }, { "epoch": 0.27, "learning_rate": 3.6489428915120776e-05, "loss": 2.8184, "step": 75700 }, { "epoch": 0.27, "learning_rate": 3.64715813971751e-05, "loss": 2.7996, "step": 75800 }, { "epoch": 0.27, "learning_rate": 3.645373387922942e-05, "loss": 2.8028, "step": 75900 }, { "epoch": 0.27, "learning_rate": 3.643588636128374e-05, "loss": 2.8051, "step": 76000 }, { "epoch": 0.27, "learning_rate": 3.641803884333806e-05, "loss": 2.7947, "step": 76100 }, { "epoch": 0.27, "learning_rate": 3.640019132539238e-05, "loss": 2.8027, "step": 76200 }, { "epoch": 0.27, "learning_rate": 3.63823438074467e-05, "loss": 2.7995, "step": 76300 }, { "epoch": 0.27, "learning_rate": 3.6364496289501024e-05, "loss": 2.8027, "step": 76400 }, { "epoch": 0.27, "learning_rate": 3.634664877155534e-05, "loss": 2.813, "step": 76500 }, { "epoch": 0.27, "learning_rate": 3.632880125360966e-05, "loss": 2.7952, "step": 76600 }, { "epoch": 0.27, "learning_rate": 3.631095373566398e-05, "loss": 2.8047, "step": 76700 }, { "epoch": 0.27, "learning_rate": 3.62931062177183e-05, "loss": 2.8081, "step": 76800 }, { "epoch": 0.27, "learning_rate": 3.6275258699772624e-05, "loss": 2.8098, "step": 76900 }, { "epoch": 0.27, "learning_rate": 3.6257411181826945e-05, "loss": 2.8081, "step": 77000 }, { "epoch": 0.28, "learning_rate": 3.6239563663881266e-05, "loss": 2.8029, "step": 77100 }, { "epoch": 0.28, "learning_rate": 3.622171614593559e-05, "loss": 2.8009, "step": 77200 }, { "epoch": 0.28, "learning_rate": 3.620386862798991e-05, "loss": 2.8057, "step": 77300 }, { "epoch": 0.28, "learning_rate": 3.618602111004423e-05, "loss": 2.8088, "step": 77400 }, { "epoch": 0.28, "learning_rate": 3.616817359209855e-05, "loss": 2.8007, "step": 77500 }, { "epoch": 0.28, "learning_rate": 3.6150326074152865e-05, "loss": 2.8024, "step": 77600 }, { "epoch": 0.28, "learning_rate": 3.6132478556207187e-05, "loss": 2.7984, "step": 77700 }, { "epoch": 0.28, "learning_rate": 3.611463103826151e-05, "loss": 2.811, "step": 77800 }, { "epoch": 0.28, "learning_rate": 3.609678352031583e-05, "loss": 2.7933, "step": 77900 }, { "epoch": 0.28, "learning_rate": 3.607893600237015e-05, "loss": 2.7904, "step": 78000 }, { "epoch": 0.28, "learning_rate": 3.606108848442447e-05, "loss": 2.8006, "step": 78100 }, { "epoch": 0.28, "learning_rate": 3.604324096647879e-05, "loss": 2.8043, "step": 78200 }, { "epoch": 0.28, "learning_rate": 3.6025393448533114e-05, "loss": 2.7954, "step": 78300 }, { "epoch": 0.28, "learning_rate": 3.6007545930587435e-05, "loss": 2.7991, "step": 78400 }, { "epoch": 0.28, "learning_rate": 3.5989698412641756e-05, "loss": 2.8027, "step": 78500 }, { "epoch": 0.28, "learning_rate": 3.597185089469608e-05, "loss": 2.8098, "step": 78600 }, { "epoch": 0.28, "learning_rate": 3.595400337675039e-05, "loss": 2.8082, "step": 78700 }, { "epoch": 0.28, "learning_rate": 3.593615585880471e-05, "loss": 2.7968, "step": 78800 }, { "epoch": 0.28, "learning_rate": 3.5918308340859034e-05, "loss": 2.8025, "step": 78900 }, { "epoch": 0.28, "learning_rate": 3.5900460822913356e-05, "loss": 2.7976, "step": 79000 }, { "epoch": 0.28, "learning_rate": 3.588261330496768e-05, "loss": 2.8049, "step": 79100 }, { "epoch": 0.28, "learning_rate": 3.5864765787022005e-05, "loss": 2.8057, "step": 79200 }, { "epoch": 0.28, "learning_rate": 3.5846918269076326e-05, "loss": 2.8104, "step": 79300 }, { "epoch": 0.28, "learning_rate": 3.582907075113065e-05, "loss": 2.796, "step": 79400 }, { "epoch": 0.28, "learning_rate": 3.581122323318497e-05, "loss": 2.8066, "step": 79500 }, { "epoch": 0.28, "learning_rate": 3.579337571523928e-05, "loss": 2.8091, "step": 79600 }, { "epoch": 0.28, "learning_rate": 3.5775528197293604e-05, "loss": 2.8055, "step": 79700 }, { "epoch": 0.28, "learning_rate": 3.5757680679347925e-05, "loss": 2.7998, "step": 79800 }, { "epoch": 0.29, "learning_rate": 3.5739833161402247e-05, "loss": 2.7972, "step": 79900 }, { "epoch": 0.29, "learning_rate": 3.572198564345657e-05, "loss": 2.8224, "step": 80000 }, { "epoch": 0.29, "learning_rate": 3.570413812551089e-05, "loss": 2.7974, "step": 80100 }, { "epoch": 0.29, "learning_rate": 3.568629060756521e-05, "loss": 2.803, "step": 80200 }, { "epoch": 0.29, "learning_rate": 3.566844308961953e-05, "loss": 2.8, "step": 80300 }, { "epoch": 0.29, "learning_rate": 3.565059557167385e-05, "loss": 2.8029, "step": 80400 }, { "epoch": 0.29, "learning_rate": 3.5632748053728174e-05, "loss": 2.7944, "step": 80500 }, { "epoch": 0.29, "learning_rate": 3.561490053578249e-05, "loss": 2.7996, "step": 80600 }, { "epoch": 0.29, "learning_rate": 3.559705301783681e-05, "loss": 2.8045, "step": 80700 }, { "epoch": 0.29, "learning_rate": 3.557920549989113e-05, "loss": 2.8062, "step": 80800 }, { "epoch": 0.29, "learning_rate": 3.556135798194545e-05, "loss": 2.7918, "step": 80900 }, { "epoch": 0.29, "learning_rate": 3.554351046399977e-05, "loss": 2.7998, "step": 81000 }, { "epoch": 0.29, "learning_rate": 3.5525662946054094e-05, "loss": 2.8071, "step": 81100 }, { "epoch": 0.29, "learning_rate": 3.5507815428108416e-05, "loss": 2.7921, "step": 81200 }, { "epoch": 0.29, "learning_rate": 3.548996791016274e-05, "loss": 2.7935, "step": 81300 }, { "epoch": 0.29, "learning_rate": 3.547212039221706e-05, "loss": 2.7901, "step": 81400 }, { "epoch": 0.29, "learning_rate": 3.545427287427138e-05, "loss": 2.7961, "step": 81500 }, { "epoch": 0.29, "learning_rate": 3.54364253563257e-05, "loss": 2.8023, "step": 81600 }, { "epoch": 0.29, "learning_rate": 3.5418577838380015e-05, "loss": 2.7976, "step": 81700 }, { "epoch": 0.29, "learning_rate": 3.5400730320434336e-05, "loss": 2.8003, "step": 81800 }, { "epoch": 0.29, "learning_rate": 3.538288280248866e-05, "loss": 2.7947, "step": 81900 }, { "epoch": 0.29, "learning_rate": 3.536503528454298e-05, "loss": 2.7876, "step": 82000 }, { "epoch": 0.29, "learning_rate": 3.53471877665973e-05, "loss": 2.8015, "step": 82100 }, { "epoch": 0.29, "learning_rate": 3.532934024865162e-05, "loss": 2.8107, "step": 82200 }, { "epoch": 0.29, "learning_rate": 3.531149273070594e-05, "loss": 2.7999, "step": 82300 }, { "epoch": 0.29, "learning_rate": 3.529364521276026e-05, "loss": 2.795, "step": 82400 }, { "epoch": 0.29, "learning_rate": 3.5275797694814585e-05, "loss": 2.7994, "step": 82500 }, { "epoch": 0.29, "learning_rate": 3.5257950176868906e-05, "loss": 2.8099, "step": 82600 }, { "epoch": 0.3, "learning_rate": 3.524010265892323e-05, "loss": 2.8062, "step": 82700 }, { "epoch": 0.3, "learning_rate": 3.522225514097754e-05, "loss": 2.795, "step": 82800 }, { "epoch": 0.3, "learning_rate": 3.520440762303186e-05, "loss": 2.8096, "step": 82900 }, { "epoch": 0.3, "learning_rate": 3.5186560105086184e-05, "loss": 2.7915, "step": 83000 }, { "epoch": 0.3, "learning_rate": 3.5168712587140505e-05, "loss": 2.7943, "step": 83100 }, { "epoch": 0.3, "learning_rate": 3.5150865069194826e-05, "loss": 2.7945, "step": 83200 }, { "epoch": 0.3, "learning_rate": 3.513301755124915e-05, "loss": 2.796, "step": 83300 }, { "epoch": 0.3, "learning_rate": 3.511517003330347e-05, "loss": 2.8073, "step": 83400 }, { "epoch": 0.3, "learning_rate": 3.509732251535779e-05, "loss": 2.8, "step": 83500 }, { "epoch": 0.3, "learning_rate": 3.507947499741211e-05, "loss": 2.7775, "step": 83600 }, { "epoch": 0.3, "learning_rate": 3.506162747946643e-05, "loss": 2.8005, "step": 83700 }, { "epoch": 0.3, "learning_rate": 3.5043779961520754e-05, "loss": 2.7855, "step": 83800 }, { "epoch": 0.3, "learning_rate": 3.502593244357507e-05, "loss": 2.7942, "step": 83900 }, { "epoch": 0.3, "learning_rate": 3.500808492562939e-05, "loss": 2.7927, "step": 84000 }, { "epoch": 0.3, "learning_rate": 3.499023740768371e-05, "loss": 2.7909, "step": 84100 }, { "epoch": 0.3, "learning_rate": 3.497238988973803e-05, "loss": 2.7891, "step": 84200 }, { "epoch": 0.3, "learning_rate": 3.495454237179236e-05, "loss": 2.7811, "step": 84300 }, { "epoch": 0.3, "learning_rate": 3.493669485384668e-05, "loss": 2.7972, "step": 84400 }, { "epoch": 0.3, "learning_rate": 3.4918847335901e-05, "loss": 2.7969, "step": 84500 }, { "epoch": 0.3, "learning_rate": 3.490099981795532e-05, "loss": 2.786, "step": 84600 }, { "epoch": 0.3, "learning_rate": 3.4883152300009645e-05, "loss": 2.7791, "step": 84700 }, { "epoch": 0.3, "learning_rate": 3.486530478206396e-05, "loss": 2.7897, "step": 84800 }, { "epoch": 0.3, "learning_rate": 3.484745726411828e-05, "loss": 2.7829, "step": 84900 }, { "epoch": 0.3, "learning_rate": 3.48296097461726e-05, "loss": 2.7877, "step": 85000 }, { "epoch": 0.3, "learning_rate": 3.481176222822692e-05, "loss": 2.7965, "step": 85100 }, { "epoch": 0.3, "learning_rate": 3.4793914710281244e-05, "loss": 2.7986, "step": 85200 }, { "epoch": 0.3, "learning_rate": 3.4776067192335565e-05, "loss": 2.7859, "step": 85300 }, { "epoch": 0.3, "learning_rate": 3.4758219674389886e-05, "loss": 2.7989, "step": 85400 }, { "epoch": 0.31, "learning_rate": 3.474037215644421e-05, "loss": 2.7942, "step": 85500 }, { "epoch": 0.31, "learning_rate": 3.472252463849853e-05, "loss": 2.793, "step": 85600 }, { "epoch": 0.31, "learning_rate": 3.470467712055285e-05, "loss": 2.7894, "step": 85700 }, { "epoch": 0.31, "learning_rate": 3.468682960260717e-05, "loss": 2.7974, "step": 85800 }, { "epoch": 0.31, "learning_rate": 3.4668982084661486e-05, "loss": 2.7912, "step": 85900 }, { "epoch": 0.31, "learning_rate": 3.465113456671581e-05, "loss": 2.8016, "step": 86000 }, { "epoch": 0.31, "learning_rate": 3.463328704877013e-05, "loss": 2.7912, "step": 86100 }, { "epoch": 0.31, "learning_rate": 3.461543953082445e-05, "loss": 2.7997, "step": 86200 }, { "epoch": 0.31, "learning_rate": 3.459759201287877e-05, "loss": 2.7946, "step": 86300 }, { "epoch": 0.31, "learning_rate": 3.457974449493309e-05, "loss": 2.7986, "step": 86400 }, { "epoch": 0.31, "learning_rate": 3.456189697698741e-05, "loss": 2.7806, "step": 86500 }, { "epoch": 0.31, "learning_rate": 3.4544049459041734e-05, "loss": 2.8009, "step": 86600 }, { "epoch": 0.31, "learning_rate": 3.4526201941096055e-05, "loss": 2.7881, "step": 86700 }, { "epoch": 0.31, "learning_rate": 3.4508354423150376e-05, "loss": 2.7896, "step": 86800 }, { "epoch": 0.31, "learning_rate": 3.44905069052047e-05, "loss": 2.7916, "step": 86900 }, { "epoch": 0.31, "learning_rate": 3.447265938725901e-05, "loss": 2.7909, "step": 87000 }, { "epoch": 0.31, "learning_rate": 3.445481186931333e-05, "loss": 2.7981, "step": 87100 }, { "epoch": 0.31, "learning_rate": 3.4436964351367655e-05, "loss": 2.7981, "step": 87200 }, { "epoch": 0.31, "learning_rate": 3.4419116833421976e-05, "loss": 2.7944, "step": 87300 }, { "epoch": 0.31, "learning_rate": 3.44012693154763e-05, "loss": 2.7774, "step": 87400 }, { "epoch": 0.31, "learning_rate": 3.438342179753062e-05, "loss": 2.7957, "step": 87500 }, { "epoch": 0.31, "learning_rate": 3.436557427958494e-05, "loss": 2.7897, "step": 87600 }, { "epoch": 0.31, "learning_rate": 3.434772676163926e-05, "loss": 2.7931, "step": 87700 }, { "epoch": 0.31, "learning_rate": 3.432987924369358e-05, "loss": 2.8006, "step": 87800 }, { "epoch": 0.31, "learning_rate": 3.43120317257479e-05, "loss": 2.7896, "step": 87900 }, { "epoch": 0.31, "learning_rate": 3.4294184207802224e-05, "loss": 2.7926, "step": 88000 }, { "epoch": 0.31, "learning_rate": 3.427633668985654e-05, "loss": 2.7889, "step": 88100 }, { "epoch": 0.31, "learning_rate": 3.425848917191086e-05, "loss": 2.7929, "step": 88200 }, { "epoch": 0.32, "learning_rate": 3.424064165396518e-05, "loss": 2.8003, "step": 88300 }, { "epoch": 0.32, "learning_rate": 3.42227941360195e-05, "loss": 2.7875, "step": 88400 }, { "epoch": 0.32, "learning_rate": 3.4204946618073824e-05, "loss": 2.7897, "step": 88500 }, { "epoch": 0.32, "learning_rate": 3.4187099100128145e-05, "loss": 2.7913, "step": 88600 }, { "epoch": 0.32, "learning_rate": 3.4169251582182466e-05, "loss": 2.7967, "step": 88700 }, { "epoch": 0.32, "learning_rate": 3.415140406423679e-05, "loss": 2.7979, "step": 88800 }, { "epoch": 0.32, "learning_rate": 3.413355654629111e-05, "loss": 2.7955, "step": 88900 }, { "epoch": 0.32, "learning_rate": 3.411570902834543e-05, "loss": 2.7939, "step": 89000 }, { "epoch": 0.32, "learning_rate": 3.409786151039975e-05, "loss": 2.7744, "step": 89100 }, { "epoch": 0.32, "learning_rate": 3.4080013992454065e-05, "loss": 2.7845, "step": 89200 }, { "epoch": 0.32, "learning_rate": 3.4062166474508387e-05, "loss": 2.7793, "step": 89300 }, { "epoch": 0.32, "learning_rate": 3.4044318956562715e-05, "loss": 2.7996, "step": 89400 }, { "epoch": 0.32, "learning_rate": 3.4026471438617036e-05, "loss": 2.7876, "step": 89500 }, { "epoch": 0.32, "learning_rate": 3.400862392067136e-05, "loss": 2.7879, "step": 89600 }, { "epoch": 0.32, "learning_rate": 3.399077640272568e-05, "loss": 2.7849, "step": 89700 }, { "epoch": 0.32, "learning_rate": 3.397292888478e-05, "loss": 2.7854, "step": 89800 }, { "epoch": 0.32, "learning_rate": 3.395508136683432e-05, "loss": 2.7948, "step": 89900 }, { "epoch": 0.32, "learning_rate": 3.393723384888864e-05, "loss": 2.7837, "step": 90000 }, { "epoch": 0.32, "learning_rate": 3.3919386330942956e-05, "loss": 2.7937, "step": 90100 }, { "epoch": 0.32, "learning_rate": 3.390153881299728e-05, "loss": 2.787, "step": 90200 }, { "epoch": 0.32, "learning_rate": 3.38836912950516e-05, "loss": 2.7907, "step": 90300 }, { "epoch": 0.32, "learning_rate": 3.386584377710592e-05, "loss": 2.7884, "step": 90400 }, { "epoch": 0.32, "learning_rate": 3.384799625916024e-05, "loss": 2.7879, "step": 90500 }, { "epoch": 0.32, "learning_rate": 3.383014874121456e-05, "loss": 2.8006, "step": 90600 }, { "epoch": 0.32, "learning_rate": 3.3812301223268884e-05, "loss": 2.8, "step": 90700 }, { "epoch": 0.32, "learning_rate": 3.3794453705323205e-05, "loss": 2.7887, "step": 90800 }, { "epoch": 0.32, "learning_rate": 3.3776606187377526e-05, "loss": 2.7826, "step": 90900 }, { "epoch": 0.32, "learning_rate": 3.375875866943185e-05, "loss": 2.7828, "step": 91000 }, { "epoch": 0.33, "learning_rate": 3.374091115148617e-05, "loss": 2.7941, "step": 91100 }, { "epoch": 0.33, "learning_rate": 3.372306363354048e-05, "loss": 2.7942, "step": 91200 }, { "epoch": 0.33, "learning_rate": 3.3705216115594804e-05, "loss": 2.7859, "step": 91300 }, { "epoch": 0.33, "learning_rate": 3.3687368597649125e-05, "loss": 2.7765, "step": 91400 }, { "epoch": 0.33, "learning_rate": 3.3669521079703446e-05, "loss": 2.7843, "step": 91500 }, { "epoch": 0.33, "learning_rate": 3.365167356175777e-05, "loss": 2.7845, "step": 91600 }, { "epoch": 0.33, "learning_rate": 3.363382604381209e-05, "loss": 2.7959, "step": 91700 }, { "epoch": 0.33, "learning_rate": 3.361597852586641e-05, "loss": 2.7833, "step": 91800 }, { "epoch": 0.33, "learning_rate": 3.359813100792073e-05, "loss": 2.7813, "step": 91900 }, { "epoch": 0.33, "learning_rate": 3.358028348997505e-05, "loss": 2.7833, "step": 92000 }, { "epoch": 0.33, "learning_rate": 3.3562435972029374e-05, "loss": 2.7833, "step": 92100 }, { "epoch": 0.33, "learning_rate": 3.3544588454083695e-05, "loss": 2.7922, "step": 92200 }, { "epoch": 0.33, "learning_rate": 3.352674093613801e-05, "loss": 2.7835, "step": 92300 }, { "epoch": 0.33, "learning_rate": 3.350889341819233e-05, "loss": 2.7863, "step": 92400 }, { "epoch": 0.33, "learning_rate": 3.349104590024665e-05, "loss": 2.7915, "step": 92500 }, { "epoch": 0.33, "learning_rate": 3.347319838230097e-05, "loss": 2.7777, "step": 92600 }, { "epoch": 0.33, "learning_rate": 3.3455350864355294e-05, "loss": 2.7776, "step": 92700 }, { "epoch": 0.33, "learning_rate": 3.3437503346409615e-05, "loss": 2.7845, "step": 92800 }, { "epoch": 0.33, "learning_rate": 3.341965582846394e-05, "loss": 2.7885, "step": 92900 }, { "epoch": 0.33, "learning_rate": 3.340180831051826e-05, "loss": 2.7868, "step": 93000 }, { "epoch": 0.33, "learning_rate": 3.338396079257258e-05, "loss": 2.7877, "step": 93100 }, { "epoch": 0.33, "learning_rate": 3.33661132746269e-05, "loss": 2.7766, "step": 93200 }, { "epoch": 0.33, "learning_rate": 3.334826575668122e-05, "loss": 2.777, "step": 93300 }, { "epoch": 0.33, "learning_rate": 3.3330418238735536e-05, "loss": 2.7879, "step": 93400 }, { "epoch": 0.33, "learning_rate": 3.331257072078986e-05, "loss": 2.7862, "step": 93500 }, { "epoch": 0.33, "learning_rate": 3.329472320284418e-05, "loss": 2.7689, "step": 93600 }, { "epoch": 0.33, "learning_rate": 3.32768756848985e-05, "loss": 2.7883, "step": 93700 }, { "epoch": 0.33, "learning_rate": 3.325902816695282e-05, "loss": 2.7862, "step": 93800 }, { "epoch": 0.34, "learning_rate": 3.324118064900714e-05, "loss": 2.789, "step": 93900 }, { "epoch": 0.34, "learning_rate": 3.322333313106146e-05, "loss": 2.7981, "step": 94000 }, { "epoch": 0.34, "learning_rate": 3.3205485613115785e-05, "loss": 2.78, "step": 94100 }, { "epoch": 0.34, "learning_rate": 3.3187638095170106e-05, "loss": 2.7938, "step": 94200 }, { "epoch": 0.34, "learning_rate": 3.316979057722443e-05, "loss": 2.7848, "step": 94300 }, { "epoch": 0.34, "learning_rate": 3.315194305927875e-05, "loss": 2.7788, "step": 94400 }, { "epoch": 0.34, "learning_rate": 3.313409554133307e-05, "loss": 2.7731, "step": 94500 }, { "epoch": 0.34, "learning_rate": 3.311624802338739e-05, "loss": 2.7879, "step": 94600 }, { "epoch": 0.34, "learning_rate": 3.309840050544171e-05, "loss": 2.7804, "step": 94700 }, { "epoch": 0.34, "learning_rate": 3.308055298749603e-05, "loss": 2.7745, "step": 94800 }, { "epoch": 0.34, "learning_rate": 3.3062705469550354e-05, "loss": 2.7767, "step": 94900 }, { "epoch": 0.34, "learning_rate": 3.3044857951604675e-05, "loss": 2.7902, "step": 95000 }, { "epoch": 0.34, "learning_rate": 3.3027010433659e-05, "loss": 2.7728, "step": 95100 }, { "epoch": 0.34, "learning_rate": 3.300916291571332e-05, "loss": 2.7857, "step": 95200 }, { "epoch": 0.34, "learning_rate": 3.299131539776764e-05, "loss": 2.7792, "step": 95300 }, { "epoch": 0.34, "learning_rate": 3.2973467879821954e-05, "loss": 2.7858, "step": 95400 }, { "epoch": 0.34, "learning_rate": 3.2955620361876275e-05, "loss": 2.7917, "step": 95500 }, { "epoch": 0.34, "learning_rate": 3.2937772843930596e-05, "loss": 2.7761, "step": 95600 }, { "epoch": 0.34, "learning_rate": 3.291992532598492e-05, "loss": 2.7859, "step": 95700 }, { "epoch": 0.34, "learning_rate": 3.290207780803924e-05, "loss": 2.7807, "step": 95800 }, { "epoch": 0.34, "learning_rate": 3.288423029009356e-05, "loss": 2.7925, "step": 95900 }, { "epoch": 0.34, "learning_rate": 3.286638277214788e-05, "loss": 2.7724, "step": 96000 }, { "epoch": 0.34, "learning_rate": 3.28485352542022e-05, "loss": 2.7807, "step": 96100 }, { "epoch": 0.34, "learning_rate": 3.283068773625652e-05, "loss": 2.7829, "step": 96200 }, { "epoch": 0.34, "learning_rate": 3.2812840218310844e-05, "loss": 2.7864, "step": 96300 }, { "epoch": 0.34, "learning_rate": 3.2794992700365166e-05, "loss": 2.7896, "step": 96400 }, { "epoch": 0.34, "learning_rate": 3.277714518241948e-05, "loss": 2.7769, "step": 96500 }, { "epoch": 0.34, "learning_rate": 3.27592976644738e-05, "loss": 2.7887, "step": 96600 }, { "epoch": 0.35, "learning_rate": 3.274145014652812e-05, "loss": 2.7704, "step": 96700 }, { "epoch": 0.35, "learning_rate": 3.2723602628582444e-05, "loss": 2.784, "step": 96800 }, { "epoch": 0.35, "learning_rate": 3.2705755110636765e-05, "loss": 2.7876, "step": 96900 }, { "epoch": 0.35, "learning_rate": 3.2687907592691086e-05, "loss": 2.7886, "step": 97000 }, { "epoch": 0.35, "learning_rate": 3.267006007474541e-05, "loss": 2.7802, "step": 97100 }, { "epoch": 0.35, "learning_rate": 3.265221255679973e-05, "loss": 2.7813, "step": 97200 }, { "epoch": 0.35, "learning_rate": 3.263436503885405e-05, "loss": 2.7827, "step": 97300 }, { "epoch": 0.35, "learning_rate": 3.261651752090837e-05, "loss": 2.7824, "step": 97400 }, { "epoch": 0.35, "learning_rate": 3.259867000296269e-05, "loss": 2.7709, "step": 97500 }, { "epoch": 0.35, "learning_rate": 3.258082248501701e-05, "loss": 2.7766, "step": 97600 }, { "epoch": 0.35, "learning_rate": 3.256297496707133e-05, "loss": 2.7816, "step": 97700 }, { "epoch": 0.35, "learning_rate": 3.254512744912565e-05, "loss": 2.774, "step": 97800 }, { "epoch": 0.35, "learning_rate": 3.252727993117997e-05, "loss": 2.7819, "step": 97900 }, { "epoch": 0.35, "learning_rate": 3.250943241323429e-05, "loss": 2.7736, "step": 98000 }, { "epoch": 0.35, "learning_rate": 3.249158489528861e-05, "loss": 2.7824, "step": 98100 }, { "epoch": 0.35, "learning_rate": 3.2473737377342934e-05, "loss": 2.7786, "step": 98200 }, { "epoch": 0.35, "learning_rate": 3.2455889859397255e-05, "loss": 2.7774, "step": 98300 }, { "epoch": 0.35, "learning_rate": 3.2438042341451576e-05, "loss": 2.7831, "step": 98400 }, { "epoch": 0.35, "learning_rate": 3.24201948235059e-05, "loss": 2.778, "step": 98500 }, { "epoch": 0.35, "learning_rate": 3.240234730556022e-05, "loss": 2.7847, "step": 98600 }, { "epoch": 0.35, "learning_rate": 3.238449978761453e-05, "loss": 2.7897, "step": 98700 }, { "epoch": 0.35, "learning_rate": 3.2366652269668855e-05, "loss": 2.7819, "step": 98800 }, { "epoch": 0.35, "learning_rate": 3.2348804751723176e-05, "loss": 2.777, "step": 98900 }, { "epoch": 0.35, "learning_rate": 3.23309572337775e-05, "loss": 2.7767, "step": 99000 }, { "epoch": 0.35, "learning_rate": 3.231310971583182e-05, "loss": 2.7747, "step": 99100 }, { "epoch": 0.35, "learning_rate": 3.229526219788614e-05, "loss": 2.7872, "step": 99200 }, { "epoch": 0.35, "learning_rate": 3.227741467994046e-05, "loss": 2.7743, "step": 99300 }, { "epoch": 0.35, "learning_rate": 3.225956716199478e-05, "loss": 2.7711, "step": 99400 }, { "epoch": 0.36, "learning_rate": 3.22417196440491e-05, "loss": 2.7764, "step": 99500 }, { "epoch": 0.36, "learning_rate": 3.2223872126103424e-05, "loss": 2.7801, "step": 99600 }, { "epoch": 0.36, "learning_rate": 3.2206024608157745e-05, "loss": 2.7806, "step": 99700 }, { "epoch": 0.36, "learning_rate": 3.218817709021207e-05, "loss": 2.7855, "step": 99800 }, { "epoch": 0.36, "learning_rate": 3.217032957226639e-05, "loss": 2.7787, "step": 99900 }, { "epoch": 0.36, "learning_rate": 3.215248205432071e-05, "loss": 2.7816, "step": 100000 }, { "epoch": 0.36, "learning_rate": 3.213463453637503e-05, "loss": 2.7644, "step": 100100 }, { "epoch": 0.36, "learning_rate": 3.211678701842935e-05, "loss": 2.7853, "step": 100200 }, { "epoch": 0.36, "learning_rate": 3.209893950048367e-05, "loss": 2.7787, "step": 100300 }, { "epoch": 0.36, "learning_rate": 3.2081091982537994e-05, "loss": 2.7706, "step": 100400 }, { "epoch": 0.36, "learning_rate": 3.2063244464592315e-05, "loss": 2.7756, "step": 100500 }, { "epoch": 0.36, "learning_rate": 3.2045396946646636e-05, "loss": 2.7726, "step": 100600 }, { "epoch": 0.36, "learning_rate": 3.202754942870095e-05, "loss": 2.7702, "step": 100700 }, { "epoch": 0.36, "learning_rate": 3.200970191075527e-05, "loss": 2.7696, "step": 100800 }, { "epoch": 0.36, "learning_rate": 3.199185439280959e-05, "loss": 2.7758, "step": 100900 }, { "epoch": 0.36, "learning_rate": 3.1974006874863914e-05, "loss": 2.7863, "step": 101000 }, { "epoch": 0.36, "learning_rate": 3.1956159356918236e-05, "loss": 2.7811, "step": 101100 }, { "epoch": 0.36, "learning_rate": 3.193831183897256e-05, "loss": 2.7844, "step": 101200 }, { "epoch": 0.36, "learning_rate": 3.192046432102688e-05, "loss": 2.781, "step": 101300 }, { "epoch": 0.36, "learning_rate": 3.19026168030812e-05, "loss": 2.7812, "step": 101400 }, { "epoch": 0.36, "learning_rate": 3.188476928513552e-05, "loss": 2.7776, "step": 101500 }, { "epoch": 0.36, "learning_rate": 3.186692176718984e-05, "loss": 2.7818, "step": 101600 }, { "epoch": 0.36, "learning_rate": 3.184907424924416e-05, "loss": 2.7723, "step": 101700 }, { "epoch": 0.36, "learning_rate": 3.183122673129848e-05, "loss": 2.7708, "step": 101800 }, { "epoch": 0.36, "learning_rate": 3.18133792133528e-05, "loss": 2.7878, "step": 101900 }, { "epoch": 0.36, "learning_rate": 3.179553169540712e-05, "loss": 2.7813, "step": 102000 }, { "epoch": 0.36, "learning_rate": 3.177768417746144e-05, "loss": 2.7802, "step": 102100 }, { "epoch": 0.36, "learning_rate": 3.175983665951576e-05, "loss": 2.773, "step": 102200 }, { "epoch": 0.37, "learning_rate": 3.1741989141570083e-05, "loss": 2.7742, "step": 102300 }, { "epoch": 0.37, "learning_rate": 3.1724141623624405e-05, "loss": 2.7707, "step": 102400 }, { "epoch": 0.37, "learning_rate": 3.1706294105678726e-05, "loss": 2.783, "step": 102500 }, { "epoch": 0.37, "learning_rate": 3.168844658773305e-05, "loss": 2.7739, "step": 102600 }, { "epoch": 0.37, "learning_rate": 3.167059906978737e-05, "loss": 2.7649, "step": 102700 }, { "epoch": 0.37, "learning_rate": 3.165275155184169e-05, "loss": 2.7831, "step": 102800 }, { "epoch": 0.37, "learning_rate": 3.1634904033896004e-05, "loss": 2.7838, "step": 102900 }, { "epoch": 0.37, "learning_rate": 3.1617056515950325e-05, "loss": 2.7795, "step": 103000 }, { "epoch": 0.37, "learning_rate": 3.1599208998004646e-05, "loss": 2.7791, "step": 103100 }, { "epoch": 0.37, "learning_rate": 3.158136148005897e-05, "loss": 2.7725, "step": 103200 }, { "epoch": 0.37, "learning_rate": 3.156351396211329e-05, "loss": 2.7783, "step": 103300 }, { "epoch": 0.37, "learning_rate": 3.154566644416761e-05, "loss": 2.7832, "step": 103400 }, { "epoch": 0.37, "learning_rate": 3.152781892622193e-05, "loss": 2.7692, "step": 103500 }, { "epoch": 0.37, "learning_rate": 3.150997140827625e-05, "loss": 2.7771, "step": 103600 }, { "epoch": 0.37, "learning_rate": 3.1492123890330574e-05, "loss": 2.7737, "step": 103700 }, { "epoch": 0.37, "learning_rate": 3.1474276372384895e-05, "loss": 2.7709, "step": 103800 }, { "epoch": 0.37, "learning_rate": 3.145642885443921e-05, "loss": 2.768, "step": 103900 }, { "epoch": 0.37, "learning_rate": 3.143858133649353e-05, "loss": 2.7733, "step": 104000 }, { "epoch": 0.37, "learning_rate": 3.142073381854785e-05, "loss": 2.7802, "step": 104100 }, { "epoch": 0.37, "learning_rate": 3.140288630060217e-05, "loss": 2.7794, "step": 104200 }, { "epoch": 0.37, "learning_rate": 3.1385038782656494e-05, "loss": 2.764, "step": 104300 }, { "epoch": 0.37, "learning_rate": 3.1367191264710815e-05, "loss": 2.7789, "step": 104400 }, { "epoch": 0.37, "learning_rate": 3.134934374676514e-05, "loss": 2.7717, "step": 104500 }, { "epoch": 0.37, "learning_rate": 3.133149622881946e-05, "loss": 2.7823, "step": 104600 }, { "epoch": 0.37, "learning_rate": 3.1313648710873786e-05, "loss": 2.7802, "step": 104700 }, { "epoch": 0.37, "learning_rate": 3.12958011929281e-05, "loss": 2.771, "step": 104800 }, { "epoch": 0.37, "learning_rate": 3.127795367498242e-05, "loss": 2.7752, "step": 104900 }, { "epoch": 0.37, "learning_rate": 3.126010615703674e-05, "loss": 2.7721, "step": 105000 }, { "epoch": 0.38, "learning_rate": 3.1242258639091064e-05, "loss": 2.7679, "step": 105100 }, { "epoch": 0.38, "learning_rate": 3.1224411121145385e-05, "loss": 2.7666, "step": 105200 }, { "epoch": 0.38, "learning_rate": 3.1206563603199706e-05, "loss": 2.7789, "step": 105300 }, { "epoch": 0.38, "learning_rate": 3.118871608525403e-05, "loss": 2.7649, "step": 105400 }, { "epoch": 0.38, "learning_rate": 3.117086856730835e-05, "loss": 2.7786, "step": 105500 }, { "epoch": 0.38, "learning_rate": 3.115302104936267e-05, "loss": 2.7672, "step": 105600 }, { "epoch": 0.38, "learning_rate": 3.113517353141699e-05, "loss": 2.772, "step": 105700 }, { "epoch": 0.38, "learning_rate": 3.111732601347131e-05, "loss": 2.7773, "step": 105800 }, { "epoch": 0.38, "learning_rate": 3.109947849552563e-05, "loss": 2.7653, "step": 105900 }, { "epoch": 0.38, "learning_rate": 3.108163097757995e-05, "loss": 2.7805, "step": 106000 }, { "epoch": 0.38, "learning_rate": 3.106378345963427e-05, "loss": 2.7822, "step": 106100 }, { "epoch": 0.38, "learning_rate": 3.104593594168859e-05, "loss": 2.7739, "step": 106200 }, { "epoch": 0.38, "learning_rate": 3.102808842374291e-05, "loss": 2.7786, "step": 106300 }, { "epoch": 0.38, "learning_rate": 3.101024090579723e-05, "loss": 2.7724, "step": 106400 }, { "epoch": 0.38, "learning_rate": 3.0992393387851554e-05, "loss": 2.7703, "step": 106500 }, { "epoch": 0.38, "learning_rate": 3.0974545869905875e-05, "loss": 2.7688, "step": 106600 }, { "epoch": 0.38, "learning_rate": 3.0956698351960197e-05, "loss": 2.7611, "step": 106700 }, { "epoch": 0.38, "learning_rate": 3.093885083401452e-05, "loss": 2.7784, "step": 106800 }, { "epoch": 0.38, "learning_rate": 3.092100331606884e-05, "loss": 2.7738, "step": 106900 }, { "epoch": 0.38, "learning_rate": 3.0903155798123153e-05, "loss": 2.7673, "step": 107000 }, { "epoch": 0.38, "learning_rate": 3.0885308280177475e-05, "loss": 2.7668, "step": 107100 }, { "epoch": 0.38, "learning_rate": 3.0867460762231796e-05, "loss": 2.7788, "step": 107200 }, { "epoch": 0.38, "learning_rate": 3.084961324428612e-05, "loss": 2.767, "step": 107300 }, { "epoch": 0.38, "learning_rate": 3.083176572634044e-05, "loss": 2.7629, "step": 107400 }, { "epoch": 0.38, "learning_rate": 3.081391820839476e-05, "loss": 2.7755, "step": 107500 }, { "epoch": 0.38, "learning_rate": 3.079607069044908e-05, "loss": 2.776, "step": 107600 }, { "epoch": 0.38, "learning_rate": 3.07782231725034e-05, "loss": 2.7787, "step": 107700 }, { "epoch": 0.38, "learning_rate": 3.076037565455772e-05, "loss": 2.772, "step": 107800 }, { "epoch": 0.39, "learning_rate": 3.0742528136612044e-05, "loss": 2.7732, "step": 107900 }, { "epoch": 0.39, "learning_rate": 3.0724680618666366e-05, "loss": 2.7695, "step": 108000 }, { "epoch": 0.39, "learning_rate": 3.070683310072068e-05, "loss": 2.7793, "step": 108100 }, { "epoch": 0.39, "learning_rate": 3.0688985582775e-05, "loss": 2.7695, "step": 108200 }, { "epoch": 0.39, "learning_rate": 3.067113806482932e-05, "loss": 2.7743, "step": 108300 }, { "epoch": 0.39, "learning_rate": 3.0653290546883644e-05, "loss": 2.7815, "step": 108400 }, { "epoch": 0.39, "learning_rate": 3.0635443028937965e-05, "loss": 2.7798, "step": 108500 }, { "epoch": 0.39, "learning_rate": 3.0617595510992286e-05, "loss": 2.7774, "step": 108600 }, { "epoch": 0.39, "learning_rate": 3.059974799304661e-05, "loss": 2.7779, "step": 108700 }, { "epoch": 0.39, "learning_rate": 3.058190047510093e-05, "loss": 2.7747, "step": 108800 }, { "epoch": 0.39, "learning_rate": 3.056405295715525e-05, "loss": 2.7721, "step": 108900 }, { "epoch": 0.39, "learning_rate": 3.054620543920957e-05, "loss": 2.7749, "step": 109000 }, { "epoch": 0.39, "learning_rate": 3.052835792126389e-05, "loss": 2.767, "step": 109100 }, { "epoch": 0.39, "learning_rate": 3.051051040331821e-05, "loss": 2.7563, "step": 109200 }, { "epoch": 0.39, "learning_rate": 3.049266288537253e-05, "loss": 2.7759, "step": 109300 }, { "epoch": 0.39, "learning_rate": 3.0474815367426852e-05, "loss": 2.766, "step": 109400 }, { "epoch": 0.39, "learning_rate": 3.045696784948117e-05, "loss": 2.7694, "step": 109500 }, { "epoch": 0.39, "learning_rate": 3.043912033153549e-05, "loss": 2.7632, "step": 109600 }, { "epoch": 0.39, "learning_rate": 3.0421272813589813e-05, "loss": 2.7565, "step": 109700 }, { "epoch": 0.39, "learning_rate": 3.0403425295644137e-05, "loss": 2.7679, "step": 109800 }, { "epoch": 0.39, "learning_rate": 3.038557777769846e-05, "loss": 2.7621, "step": 109900 }, { "epoch": 0.39, "learning_rate": 3.036773025975278e-05, "loss": 2.7688, "step": 110000 }, { "epoch": 0.39, "learning_rate": 3.03498827418071e-05, "loss": 2.7615, "step": 110100 }, { "epoch": 0.39, "learning_rate": 3.0332035223861422e-05, "loss": 2.7697, "step": 110200 }, { "epoch": 0.39, "learning_rate": 3.0314187705915743e-05, "loss": 2.7759, "step": 110300 }, { "epoch": 0.39, "learning_rate": 3.029634018797006e-05, "loss": 2.7709, "step": 110400 }, { "epoch": 0.39, "learning_rate": 3.0278492670024382e-05, "loss": 2.7642, "step": 110500 }, { "epoch": 0.39, "learning_rate": 3.0260645152078704e-05, "loss": 2.764, "step": 110600 }, { "epoch": 0.4, "learning_rate": 3.0242797634133025e-05, "loss": 2.7764, "step": 110700 }, { "epoch": 0.4, "learning_rate": 3.0224950116187346e-05, "loss": 2.7683, "step": 110800 }, { "epoch": 0.4, "learning_rate": 3.0207102598241664e-05, "loss": 2.7657, "step": 110900 }, { "epoch": 0.4, "learning_rate": 3.0189255080295985e-05, "loss": 2.7731, "step": 111000 }, { "epoch": 0.4, "learning_rate": 3.0171407562350306e-05, "loss": 2.7703, "step": 111100 }, { "epoch": 0.4, "learning_rate": 3.0153560044404628e-05, "loss": 2.7718, "step": 111200 }, { "epoch": 0.4, "learning_rate": 3.013571252645895e-05, "loss": 2.7739, "step": 111300 }, { "epoch": 0.4, "learning_rate": 3.011786500851327e-05, "loss": 2.7715, "step": 111400 }, { "epoch": 0.4, "learning_rate": 3.0100017490567588e-05, "loss": 2.7594, "step": 111500 }, { "epoch": 0.4, "learning_rate": 3.008216997262191e-05, "loss": 2.7705, "step": 111600 }, { "epoch": 0.4, "learning_rate": 3.006432245467623e-05, "loss": 2.7714, "step": 111700 }, { "epoch": 0.4, "learning_rate": 3.004647493673055e-05, "loss": 2.7601, "step": 111800 }, { "epoch": 0.4, "learning_rate": 3.0028627418784873e-05, "loss": 2.7659, "step": 111900 }, { "epoch": 0.4, "learning_rate": 3.001077990083919e-05, "loss": 2.7733, "step": 112000 }, { "epoch": 0.4, "learning_rate": 2.9992932382893512e-05, "loss": 2.7685, "step": 112100 }, { "epoch": 0.4, "learning_rate": 2.9975084864947833e-05, "loss": 2.7684, "step": 112200 }, { "epoch": 0.4, "learning_rate": 2.9957237347002154e-05, "loss": 2.7722, "step": 112300 }, { "epoch": 0.4, "learning_rate": 2.9939389829056475e-05, "loss": 2.7639, "step": 112400 }, { "epoch": 0.4, "learning_rate": 2.9921542311110797e-05, "loss": 2.7693, "step": 112500 }, { "epoch": 0.4, "learning_rate": 2.9903694793165114e-05, "loss": 2.759, "step": 112600 }, { "epoch": 0.4, "learning_rate": 2.9885847275219436e-05, "loss": 2.7726, "step": 112700 }, { "epoch": 0.4, "learning_rate": 2.9867999757273757e-05, "loss": 2.7667, "step": 112800 }, { "epoch": 0.4, "learning_rate": 2.9850152239328078e-05, "loss": 2.7621, "step": 112900 }, { "epoch": 0.4, "learning_rate": 2.98323047213824e-05, "loss": 2.772, "step": 113000 }, { "epoch": 0.4, "learning_rate": 2.9814457203436717e-05, "loss": 2.7647, "step": 113100 }, { "epoch": 0.4, "learning_rate": 2.9796609685491038e-05, "loss": 2.7604, "step": 113200 }, { "epoch": 0.4, "learning_rate": 2.977876216754536e-05, "loss": 2.7798, "step": 113300 }, { "epoch": 0.4, "learning_rate": 2.976091464959968e-05, "loss": 2.7606, "step": 113400 }, { "epoch": 0.41, "learning_rate": 2.9743067131654002e-05, "loss": 2.7628, "step": 113500 }, { "epoch": 0.41, "learning_rate": 2.9725219613708323e-05, "loss": 2.7717, "step": 113600 }, { "epoch": 0.41, "learning_rate": 2.970737209576264e-05, "loss": 2.7719, "step": 113700 }, { "epoch": 0.41, "learning_rate": 2.9689524577816962e-05, "loss": 2.7632, "step": 113800 }, { "epoch": 0.41, "learning_rate": 2.9671677059871283e-05, "loss": 2.7752, "step": 113900 }, { "epoch": 0.41, "learning_rate": 2.9653829541925605e-05, "loss": 2.7617, "step": 114000 }, { "epoch": 0.41, "learning_rate": 2.9635982023979926e-05, "loss": 2.7669, "step": 114100 }, { "epoch": 0.41, "learning_rate": 2.9618134506034244e-05, "loss": 2.7616, "step": 114200 }, { "epoch": 0.41, "learning_rate": 2.9600286988088565e-05, "loss": 2.7716, "step": 114300 }, { "epoch": 0.41, "learning_rate": 2.9582439470142886e-05, "loss": 2.7509, "step": 114400 }, { "epoch": 0.41, "learning_rate": 2.9564591952197207e-05, "loss": 2.7704, "step": 114500 }, { "epoch": 0.41, "learning_rate": 2.954674443425153e-05, "loss": 2.7653, "step": 114600 }, { "epoch": 0.41, "learning_rate": 2.952889691630585e-05, "loss": 2.7519, "step": 114700 }, { "epoch": 0.41, "learning_rate": 2.9511049398360168e-05, "loss": 2.7644, "step": 114800 }, { "epoch": 0.41, "learning_rate": 2.9493201880414496e-05, "loss": 2.7755, "step": 114900 }, { "epoch": 0.41, "learning_rate": 2.9475354362468817e-05, "loss": 2.7644, "step": 115000 }, { "epoch": 0.41, "learning_rate": 2.9457506844523135e-05, "loss": 2.7642, "step": 115100 }, { "epoch": 0.41, "learning_rate": 2.9439659326577456e-05, "loss": 2.7685, "step": 115200 }, { "epoch": 0.41, "learning_rate": 2.9421811808631777e-05, "loss": 2.7611, "step": 115300 }, { "epoch": 0.41, "learning_rate": 2.9403964290686098e-05, "loss": 2.7727, "step": 115400 }, { "epoch": 0.41, "learning_rate": 2.938611677274042e-05, "loss": 2.7658, "step": 115500 }, { "epoch": 0.41, "learning_rate": 2.936826925479474e-05, "loss": 2.7597, "step": 115600 }, { "epoch": 0.41, "learning_rate": 2.935042173684906e-05, "loss": 2.761, "step": 115700 }, { "epoch": 0.41, "learning_rate": 2.933257421890338e-05, "loss": 2.7679, "step": 115800 }, { "epoch": 0.41, "learning_rate": 2.93147267009577e-05, "loss": 2.7595, "step": 115900 }, { "epoch": 0.41, "learning_rate": 2.9296879183012022e-05, "loss": 2.7625, "step": 116000 }, { "epoch": 0.41, "learning_rate": 2.9279031665066343e-05, "loss": 2.7687, "step": 116100 }, { "epoch": 0.41, "learning_rate": 2.926118414712066e-05, "loss": 2.7706, "step": 116200 }, { "epoch": 0.42, "learning_rate": 2.9243336629174982e-05, "loss": 2.7599, "step": 116300 }, { "epoch": 0.42, "learning_rate": 2.9225489111229304e-05, "loss": 2.7678, "step": 116400 }, { "epoch": 0.42, "learning_rate": 2.9207641593283625e-05, "loss": 2.7669, "step": 116500 }, { "epoch": 0.42, "learning_rate": 2.9189794075337946e-05, "loss": 2.7563, "step": 116600 }, { "epoch": 0.42, "learning_rate": 2.9171946557392267e-05, "loss": 2.7615, "step": 116700 }, { "epoch": 0.42, "learning_rate": 2.9154099039446585e-05, "loss": 2.7517, "step": 116800 }, { "epoch": 0.42, "learning_rate": 2.9136251521500906e-05, "loss": 2.7608, "step": 116900 }, { "epoch": 0.42, "learning_rate": 2.9118404003555228e-05, "loss": 2.769, "step": 117000 }, { "epoch": 0.42, "learning_rate": 2.910055648560955e-05, "loss": 2.7616, "step": 117100 }, { "epoch": 0.42, "learning_rate": 2.908270896766387e-05, "loss": 2.7562, "step": 117200 }, { "epoch": 0.42, "learning_rate": 2.9064861449718188e-05, "loss": 2.7626, "step": 117300 }, { "epoch": 0.42, "learning_rate": 2.904701393177251e-05, "loss": 2.7589, "step": 117400 }, { "epoch": 0.42, "learning_rate": 2.902916641382683e-05, "loss": 2.7657, "step": 117500 }, { "epoch": 0.42, "learning_rate": 2.901131889588115e-05, "loss": 2.7595, "step": 117600 }, { "epoch": 0.42, "learning_rate": 2.8993471377935473e-05, "loss": 2.7686, "step": 117700 }, { "epoch": 0.42, "learning_rate": 2.8975623859989794e-05, "loss": 2.7624, "step": 117800 }, { "epoch": 0.42, "learning_rate": 2.895777634204411e-05, "loss": 2.7675, "step": 117900 }, { "epoch": 0.42, "learning_rate": 2.8939928824098433e-05, "loss": 2.759, "step": 118000 }, { "epoch": 0.42, "learning_rate": 2.8922081306152754e-05, "loss": 2.7601, "step": 118100 }, { "epoch": 0.42, "learning_rate": 2.8904233788207075e-05, "loss": 2.7665, "step": 118200 }, { "epoch": 0.42, "learning_rate": 2.8886386270261397e-05, "loss": 2.7626, "step": 118300 }, { "epoch": 0.42, "learning_rate": 2.8868538752315714e-05, "loss": 2.7513, "step": 118400 }, { "epoch": 0.42, "learning_rate": 2.8850691234370036e-05, "loss": 2.7645, "step": 118500 }, { "epoch": 0.42, "learning_rate": 2.8832843716424357e-05, "loss": 2.7585, "step": 118600 }, { "epoch": 0.42, "learning_rate": 2.8814996198478678e-05, "loss": 2.7556, "step": 118700 }, { "epoch": 0.42, "learning_rate": 2.8797148680533e-05, "loss": 2.7694, "step": 118800 }, { "epoch": 0.42, "learning_rate": 2.877930116258732e-05, "loss": 2.7676, "step": 118900 }, { "epoch": 0.42, "learning_rate": 2.8761453644641638e-05, "loss": 2.7649, "step": 119000 }, { "epoch": 0.43, "learning_rate": 2.874360612669596e-05, "loss": 2.7552, "step": 119100 }, { "epoch": 0.43, "learning_rate": 2.872575860875028e-05, "loss": 2.7612, "step": 119200 }, { "epoch": 0.43, "learning_rate": 2.8707911090804602e-05, "loss": 2.7602, "step": 119300 }, { "epoch": 0.43, "learning_rate": 2.8690063572858923e-05, "loss": 2.7551, "step": 119400 }, { "epoch": 0.43, "learning_rate": 2.867221605491324e-05, "loss": 2.7676, "step": 119500 }, { "epoch": 0.43, "learning_rate": 2.8654368536967562e-05, "loss": 2.7624, "step": 119600 }, { "epoch": 0.43, "learning_rate": 2.8636521019021883e-05, "loss": 2.7723, "step": 119700 }, { "epoch": 0.43, "learning_rate": 2.8618673501076205e-05, "loss": 2.754, "step": 119800 }, { "epoch": 0.43, "learning_rate": 2.8600825983130526e-05, "loss": 2.763, "step": 119900 }, { "epoch": 0.43, "learning_rate": 2.8582978465184847e-05, "loss": 2.7562, "step": 120000 }, { "epoch": 0.43, "learning_rate": 2.856513094723917e-05, "loss": 2.7504, "step": 120100 }, { "epoch": 0.43, "learning_rate": 2.8547283429293493e-05, "loss": 2.7601, "step": 120200 }, { "epoch": 0.43, "learning_rate": 2.8529435911347814e-05, "loss": 2.7528, "step": 120300 }, { "epoch": 0.43, "learning_rate": 2.8511588393402132e-05, "loss": 2.7636, "step": 120400 }, { "epoch": 0.43, "learning_rate": 2.8493740875456453e-05, "loss": 2.7589, "step": 120500 }, { "epoch": 0.43, "learning_rate": 2.8475893357510774e-05, "loss": 2.7598, "step": 120600 }, { "epoch": 0.43, "learning_rate": 2.8458045839565096e-05, "loss": 2.7556, "step": 120700 }, { "epoch": 0.43, "learning_rate": 2.8440198321619417e-05, "loss": 2.757, "step": 120800 }, { "epoch": 0.43, "learning_rate": 2.8422350803673735e-05, "loss": 2.7574, "step": 120900 }, { "epoch": 0.43, "learning_rate": 2.8404503285728056e-05, "loss": 2.7673, "step": 121000 }, { "epoch": 0.43, "learning_rate": 2.8386655767782377e-05, "loss": 2.7559, "step": 121100 }, { "epoch": 0.43, "learning_rate": 2.8368808249836698e-05, "loss": 2.7568, "step": 121200 }, { "epoch": 0.43, "learning_rate": 2.835096073189102e-05, "loss": 2.7604, "step": 121300 }, { "epoch": 0.43, "learning_rate": 2.833311321394534e-05, "loss": 2.7564, "step": 121400 }, { "epoch": 0.43, "learning_rate": 2.831526569599966e-05, "loss": 2.7626, "step": 121500 }, { "epoch": 0.43, "learning_rate": 2.829741817805398e-05, "loss": 2.7636, "step": 121600 }, { "epoch": 0.43, "learning_rate": 2.82795706601083e-05, "loss": 2.7605, "step": 121700 }, { "epoch": 0.43, "learning_rate": 2.8261723142162622e-05, "loss": 2.7583, "step": 121800 }, { "epoch": 0.44, "learning_rate": 2.8243875624216943e-05, "loss": 2.7727, "step": 121900 }, { "epoch": 0.44, "learning_rate": 2.822602810627126e-05, "loss": 2.7519, "step": 122000 }, { "epoch": 0.44, "learning_rate": 2.8208180588325582e-05, "loss": 2.7585, "step": 122100 }, { "epoch": 0.44, "learning_rate": 2.8190333070379904e-05, "loss": 2.7645, "step": 122200 }, { "epoch": 0.44, "learning_rate": 2.8172485552434225e-05, "loss": 2.7689, "step": 122300 }, { "epoch": 0.44, "learning_rate": 2.8154638034488546e-05, "loss": 2.7523, "step": 122400 }, { "epoch": 0.44, "learning_rate": 2.8136790516542867e-05, "loss": 2.7602, "step": 122500 }, { "epoch": 0.44, "learning_rate": 2.8118942998597185e-05, "loss": 2.756, "step": 122600 }, { "epoch": 0.44, "learning_rate": 2.8101095480651506e-05, "loss": 2.7626, "step": 122700 }, { "epoch": 0.44, "learning_rate": 2.8083247962705828e-05, "loss": 2.7616, "step": 122800 }, { "epoch": 0.44, "learning_rate": 2.806540044476015e-05, "loss": 2.7545, "step": 122900 }, { "epoch": 0.44, "learning_rate": 2.804755292681447e-05, "loss": 2.7735, "step": 123000 }, { "epoch": 0.44, "learning_rate": 2.8029705408868788e-05, "loss": 2.761, "step": 123100 }, { "epoch": 0.44, "learning_rate": 2.801185789092311e-05, "loss": 2.7621, "step": 123200 }, { "epoch": 0.44, "learning_rate": 2.799401037297743e-05, "loss": 2.7577, "step": 123300 }, { "epoch": 0.44, "learning_rate": 2.797616285503175e-05, "loss": 2.7498, "step": 123400 }, { "epoch": 0.44, "learning_rate": 2.7958315337086073e-05, "loss": 2.7634, "step": 123500 }, { "epoch": 0.44, "learning_rate": 2.7940467819140394e-05, "loss": 2.764, "step": 123600 }, { "epoch": 0.44, "learning_rate": 2.792262030119471e-05, "loss": 2.7741, "step": 123700 }, { "epoch": 0.44, "learning_rate": 2.7904772783249033e-05, "loss": 2.7674, "step": 123800 }, { "epoch": 0.44, "learning_rate": 2.7886925265303354e-05, "loss": 2.748, "step": 123900 }, { "epoch": 0.44, "learning_rate": 2.7869077747357675e-05, "loss": 2.7595, "step": 124000 }, { "epoch": 0.44, "learning_rate": 2.7851230229411997e-05, "loss": 2.7542, "step": 124100 }, { "epoch": 0.44, "learning_rate": 2.7833382711466314e-05, "loss": 2.7677, "step": 124200 }, { "epoch": 0.44, "learning_rate": 2.7815535193520636e-05, "loss": 2.7678, "step": 124300 }, { "epoch": 0.44, "learning_rate": 2.7797687675574957e-05, "loss": 2.7632, "step": 124400 }, { "epoch": 0.44, "learning_rate": 2.7779840157629278e-05, "loss": 2.7536, "step": 124500 }, { "epoch": 0.44, "learning_rate": 2.77619926396836e-05, "loss": 2.7657, "step": 124600 }, { "epoch": 0.45, "learning_rate": 2.774414512173792e-05, "loss": 2.7621, "step": 124700 }, { "epoch": 0.45, "learning_rate": 2.7726297603792238e-05, "loss": 2.7572, "step": 124800 }, { "epoch": 0.45, "learning_rate": 2.770845008584656e-05, "loss": 2.759, "step": 124900 }, { "epoch": 0.45, "learning_rate": 2.769060256790088e-05, "loss": 2.7554, "step": 125000 }, { "epoch": 0.45, "learning_rate": 2.7672755049955202e-05, "loss": 2.7481, "step": 125100 }, { "epoch": 0.45, "learning_rate": 2.7654907532009526e-05, "loss": 2.7598, "step": 125200 }, { "epoch": 0.45, "learning_rate": 2.7637060014063848e-05, "loss": 2.7662, "step": 125300 }, { "epoch": 0.45, "learning_rate": 2.761921249611817e-05, "loss": 2.765, "step": 125400 }, { "epoch": 0.45, "learning_rate": 2.760136497817249e-05, "loss": 2.7548, "step": 125500 }, { "epoch": 0.45, "learning_rate": 2.758351746022681e-05, "loss": 2.7529, "step": 125600 }, { "epoch": 0.45, "learning_rate": 2.756566994228113e-05, "loss": 2.7479, "step": 125700 }, { "epoch": 0.45, "learning_rate": 2.754782242433545e-05, "loss": 2.7477, "step": 125800 }, { "epoch": 0.45, "learning_rate": 2.752997490638977e-05, "loss": 2.745, "step": 125900 }, { "epoch": 0.45, "learning_rate": 2.7512127388444093e-05, "loss": 2.7561, "step": 126000 }, { "epoch": 0.45, "learning_rate": 2.7494279870498414e-05, "loss": 2.7503, "step": 126100 }, { "epoch": 0.45, "learning_rate": 2.7476432352552732e-05, "loss": 2.7533, "step": 126200 }, { "epoch": 0.45, "learning_rate": 2.7458584834607053e-05, "loss": 2.7483, "step": 126300 }, { "epoch": 0.45, "learning_rate": 2.7440737316661374e-05, "loss": 2.7628, "step": 126400 }, { "epoch": 0.45, "learning_rate": 2.7422889798715696e-05, "loss": 2.7571, "step": 126500 }, { "epoch": 0.45, "learning_rate": 2.7405042280770017e-05, "loss": 2.7596, "step": 126600 }, { "epoch": 0.45, "learning_rate": 2.7387194762824338e-05, "loss": 2.7522, "step": 126700 }, { "epoch": 0.45, "learning_rate": 2.7369347244878656e-05, "loss": 2.7577, "step": 126800 }, { "epoch": 0.45, "learning_rate": 2.7351499726932977e-05, "loss": 2.7501, "step": 126900 }, { "epoch": 0.45, "learning_rate": 2.7333652208987298e-05, "loss": 2.7453, "step": 127000 }, { "epoch": 0.45, "learning_rate": 2.731580469104162e-05, "loss": 2.7565, "step": 127100 }, { "epoch": 0.45, "learning_rate": 2.729795717309594e-05, "loss": 2.7527, "step": 127200 }, { "epoch": 0.45, "learning_rate": 2.728010965515026e-05, "loss": 2.7426, "step": 127300 }, { "epoch": 0.45, "learning_rate": 2.726226213720458e-05, "loss": 2.7606, "step": 127400 }, { "epoch": 0.46, "learning_rate": 2.72444146192589e-05, "loss": 2.752, "step": 127500 }, { "epoch": 0.46, "learning_rate": 2.7226567101313222e-05, "loss": 2.7606, "step": 127600 }, { "epoch": 0.46, "learning_rate": 2.7208719583367543e-05, "loss": 2.765, "step": 127700 }, { "epoch": 0.46, "learning_rate": 2.7190872065421865e-05, "loss": 2.7667, "step": 127800 }, { "epoch": 0.46, "learning_rate": 2.7173024547476182e-05, "loss": 2.7577, "step": 127900 }, { "epoch": 0.46, "learning_rate": 2.7155177029530504e-05, "loss": 2.7512, "step": 128000 }, { "epoch": 0.46, "learning_rate": 2.7137329511584825e-05, "loss": 2.7527, "step": 128100 }, { "epoch": 0.46, "learning_rate": 2.7119481993639146e-05, "loss": 2.7469, "step": 128200 }, { "epoch": 0.46, "learning_rate": 2.7101634475693467e-05, "loss": 2.7507, "step": 128300 }, { "epoch": 0.46, "learning_rate": 2.7083786957747785e-05, "loss": 2.7518, "step": 128400 }, { "epoch": 0.46, "learning_rate": 2.7065939439802106e-05, "loss": 2.751, "step": 128500 }, { "epoch": 0.46, "learning_rate": 2.7048091921856427e-05, "loss": 2.7627, "step": 128600 }, { "epoch": 0.46, "learning_rate": 2.703024440391075e-05, "loss": 2.7516, "step": 128700 }, { "epoch": 0.46, "learning_rate": 2.701239688596507e-05, "loss": 2.7515, "step": 128800 }, { "epoch": 0.46, "learning_rate": 2.699454936801939e-05, "loss": 2.7556, "step": 128900 }, { "epoch": 0.46, "learning_rate": 2.697670185007371e-05, "loss": 2.7491, "step": 129000 }, { "epoch": 0.46, "learning_rate": 2.695885433212803e-05, "loss": 2.7524, "step": 129100 }, { "epoch": 0.46, "learning_rate": 2.694100681418235e-05, "loss": 2.7487, "step": 129200 }, { "epoch": 0.46, "learning_rate": 2.6923159296236673e-05, "loss": 2.763, "step": 129300 }, { "epoch": 0.46, "learning_rate": 2.6905311778290994e-05, "loss": 2.739, "step": 129400 }, { "epoch": 0.46, "learning_rate": 2.688746426034531e-05, "loss": 2.7549, "step": 129500 }, { "epoch": 0.46, "learning_rate": 2.6869616742399633e-05, "loss": 2.761, "step": 129600 }, { "epoch": 0.46, "learning_rate": 2.6851769224453954e-05, "loss": 2.7487, "step": 129700 }, { "epoch": 0.46, "learning_rate": 2.6833921706508275e-05, "loss": 2.7585, "step": 129800 }, { "epoch": 0.46, "learning_rate": 2.6816074188562596e-05, "loss": 2.7546, "step": 129900 }, { "epoch": 0.46, "learning_rate": 2.6798226670616918e-05, "loss": 2.7487, "step": 130000 }, { "epoch": 0.46, "learning_rate": 2.6780379152671236e-05, "loss": 2.7442, "step": 130100 }, { "epoch": 0.46, "learning_rate": 2.6762531634725557e-05, "loss": 2.7464, "step": 130200 }, { "epoch": 0.47, "learning_rate": 2.6744684116779885e-05, "loss": 2.7618, "step": 130300 }, { "epoch": 0.47, "learning_rate": 2.6726836598834203e-05, "loss": 2.7588, "step": 130400 }, { "epoch": 0.47, "learning_rate": 2.6708989080888524e-05, "loss": 2.7507, "step": 130500 }, { "epoch": 0.47, "learning_rate": 2.6691141562942845e-05, "loss": 2.7527, "step": 130600 }, { "epoch": 0.47, "learning_rate": 2.6673294044997166e-05, "loss": 2.7451, "step": 130700 }, { "epoch": 0.47, "learning_rate": 2.6655446527051487e-05, "loss": 2.747, "step": 130800 }, { "epoch": 0.47, "learning_rate": 2.663759900910581e-05, "loss": 2.7497, "step": 130900 }, { "epoch": 0.47, "learning_rate": 2.6619751491160126e-05, "loss": 2.7488, "step": 131000 }, { "epoch": 0.47, "learning_rate": 2.6601903973214448e-05, "loss": 2.7602, "step": 131100 }, { "epoch": 0.47, "learning_rate": 2.658405645526877e-05, "loss": 2.7607, "step": 131200 }, { "epoch": 0.47, "learning_rate": 2.656620893732309e-05, "loss": 2.755, "step": 131300 }, { "epoch": 0.47, "learning_rate": 2.654836141937741e-05, "loss": 2.7475, "step": 131400 }, { "epoch": 0.47, "learning_rate": 2.653051390143173e-05, "loss": 2.748, "step": 131500 }, { "epoch": 0.47, "learning_rate": 2.651266638348605e-05, "loss": 2.7503, "step": 131600 }, { "epoch": 0.47, "learning_rate": 2.649481886554037e-05, "loss": 2.7487, "step": 131700 }, { "epoch": 0.47, "learning_rate": 2.6476971347594693e-05, "loss": 2.7458, "step": 131800 }, { "epoch": 0.47, "learning_rate": 2.6459123829649014e-05, "loss": 2.7588, "step": 131900 }, { "epoch": 0.47, "learning_rate": 2.6441276311703332e-05, "loss": 2.7382, "step": 132000 }, { "epoch": 0.47, "learning_rate": 2.6423428793757653e-05, "loss": 2.757, "step": 132100 }, { "epoch": 0.47, "learning_rate": 2.6405581275811974e-05, "loss": 2.762, "step": 132200 }, { "epoch": 0.47, "learning_rate": 2.6387733757866295e-05, "loss": 2.7397, "step": 132300 }, { "epoch": 0.47, "learning_rate": 2.6369886239920617e-05, "loss": 2.7499, "step": 132400 }, { "epoch": 0.47, "learning_rate": 2.6352038721974938e-05, "loss": 2.7543, "step": 132500 }, { "epoch": 0.47, "learning_rate": 2.6334191204029256e-05, "loss": 2.7473, "step": 132600 }, { "epoch": 0.47, "learning_rate": 2.6316343686083577e-05, "loss": 2.7637, "step": 132700 }, { "epoch": 0.47, "learning_rate": 2.6298496168137898e-05, "loss": 2.7512, "step": 132800 }, { "epoch": 0.47, "learning_rate": 2.628064865019222e-05, "loss": 2.7512, "step": 132900 }, { "epoch": 0.47, "learning_rate": 2.626280113224654e-05, "loss": 2.7439, "step": 133000 }, { "epoch": 0.48, "learning_rate": 2.624495361430086e-05, "loss": 2.7528, "step": 133100 }, { "epoch": 0.48, "learning_rate": 2.622710609635518e-05, "loss": 2.7516, "step": 133200 }, { "epoch": 0.48, "learning_rate": 2.62092585784095e-05, "loss": 2.7441, "step": 133300 }, { "epoch": 0.48, "learning_rate": 2.6191411060463822e-05, "loss": 2.7522, "step": 133400 }, { "epoch": 0.48, "learning_rate": 2.6173563542518143e-05, "loss": 2.7407, "step": 133500 }, { "epoch": 0.48, "learning_rate": 2.6155716024572465e-05, "loss": 2.7493, "step": 133600 }, { "epoch": 0.48, "learning_rate": 2.6137868506626782e-05, "loss": 2.738, "step": 133700 }, { "epoch": 0.48, "learning_rate": 2.6120020988681104e-05, "loss": 2.7546, "step": 133800 }, { "epoch": 0.48, "learning_rate": 2.6102173470735425e-05, "loss": 2.7533, "step": 133900 }, { "epoch": 0.48, "learning_rate": 2.6084325952789746e-05, "loss": 2.7519, "step": 134000 }, { "epoch": 0.48, "learning_rate": 2.6066478434844067e-05, "loss": 2.7701, "step": 134100 }, { "epoch": 0.48, "learning_rate": 2.6048630916898385e-05, "loss": 2.7568, "step": 134200 }, { "epoch": 0.48, "learning_rate": 2.6030783398952706e-05, "loss": 2.7504, "step": 134300 }, { "epoch": 0.48, "learning_rate": 2.6012935881007027e-05, "loss": 2.7449, "step": 134400 }, { "epoch": 0.48, "learning_rate": 2.599508836306135e-05, "loss": 2.7515, "step": 134500 }, { "epoch": 0.48, "learning_rate": 2.597724084511567e-05, "loss": 2.751, "step": 134600 }, { "epoch": 0.48, "learning_rate": 2.595939332716999e-05, "loss": 2.7471, "step": 134700 }, { "epoch": 0.48, "learning_rate": 2.594154580922431e-05, "loss": 2.7567, "step": 134800 }, { "epoch": 0.48, "learning_rate": 2.592369829127863e-05, "loss": 2.7525, "step": 134900 }, { "epoch": 0.48, "learning_rate": 2.590585077333295e-05, "loss": 2.7495, "step": 135000 }, { "epoch": 0.48, "learning_rate": 2.5888003255387273e-05, "loss": 2.7366, "step": 135100 }, { "epoch": 0.48, "learning_rate": 2.5870155737441594e-05, "loss": 2.7473, "step": 135200 }, { "epoch": 0.48, "learning_rate": 2.585230821949591e-05, "loss": 2.7443, "step": 135300 }, { "epoch": 0.48, "learning_rate": 2.583446070155024e-05, "loss": 2.7467, "step": 135400 }, { "epoch": 0.48, "learning_rate": 2.581661318360456e-05, "loss": 2.738, "step": 135500 }, { "epoch": 0.48, "learning_rate": 2.5798765665658882e-05, "loss": 2.7387, "step": 135600 }, { "epoch": 0.48, "learning_rate": 2.57809181477132e-05, "loss": 2.7489, "step": 135700 }, { "epoch": 0.48, "learning_rate": 2.576307062976752e-05, "loss": 2.7453, "step": 135800 }, { "epoch": 0.49, "learning_rate": 2.5745223111821842e-05, "loss": 2.7451, "step": 135900 }, { "epoch": 0.49, "learning_rate": 2.5727375593876163e-05, "loss": 2.7548, "step": 136000 }, { "epoch": 0.49, "learning_rate": 2.5709528075930485e-05, "loss": 2.7495, "step": 136100 }, { "epoch": 0.49, "learning_rate": 2.5691680557984803e-05, "loss": 2.7477, "step": 136200 }, { "epoch": 0.49, "learning_rate": 2.5673833040039124e-05, "loss": 2.7342, "step": 136300 }, { "epoch": 0.49, "learning_rate": 2.5655985522093445e-05, "loss": 2.747, "step": 136400 }, { "epoch": 0.49, "learning_rate": 2.5638138004147766e-05, "loss": 2.7491, "step": 136500 }, { "epoch": 0.49, "learning_rate": 2.5620290486202087e-05, "loss": 2.7354, "step": 136600 }, { "epoch": 0.49, "learning_rate": 2.560244296825641e-05, "loss": 2.7453, "step": 136700 }, { "epoch": 0.49, "learning_rate": 2.5584595450310726e-05, "loss": 2.7476, "step": 136800 }, { "epoch": 0.49, "learning_rate": 2.5566747932365048e-05, "loss": 2.7495, "step": 136900 }, { "epoch": 0.49, "learning_rate": 2.554890041441937e-05, "loss": 2.7475, "step": 137000 }, { "epoch": 0.49, "learning_rate": 2.553105289647369e-05, "loss": 2.7538, "step": 137100 }, { "epoch": 0.49, "learning_rate": 2.551320537852801e-05, "loss": 2.7488, "step": 137200 }, { "epoch": 0.49, "learning_rate": 2.549535786058233e-05, "loss": 2.7532, "step": 137300 }, { "epoch": 0.49, "learning_rate": 2.547751034263665e-05, "loss": 2.7602, "step": 137400 }, { "epoch": 0.49, "learning_rate": 2.545966282469097e-05, "loss": 2.7702, "step": 137500 }, { "epoch": 0.49, "learning_rate": 2.5441815306745293e-05, "loss": 2.7445, "step": 137600 }, { "epoch": 0.49, "learning_rate": 2.5423967788799614e-05, "loss": 2.7486, "step": 137700 }, { "epoch": 0.49, "learning_rate": 2.5406120270853935e-05, "loss": 2.7399, "step": 137800 }, { "epoch": 0.49, "learning_rate": 2.5388272752908253e-05, "loss": 2.7414, "step": 137900 }, { "epoch": 0.49, "learning_rate": 2.5370425234962574e-05, "loss": 2.7525, "step": 138000 }, { "epoch": 0.49, "learning_rate": 2.5352577717016895e-05, "loss": 2.7458, "step": 138100 }, { "epoch": 0.49, "learning_rate": 2.5334730199071217e-05, "loss": 2.752, "step": 138200 }, { "epoch": 0.49, "learning_rate": 2.5316882681125538e-05, "loss": 2.7483, "step": 138300 }, { "epoch": 0.49, "learning_rate": 2.5299035163179856e-05, "loss": 2.74, "step": 138400 }, { "epoch": 0.49, "learning_rate": 2.5281187645234177e-05, "loss": 2.7632, "step": 138500 }, { "epoch": 0.49, "learning_rate": 2.5263340127288498e-05, "loss": 2.7447, "step": 138600 }, { "epoch": 0.5, "learning_rate": 2.524549260934282e-05, "loss": 2.7355, "step": 138700 }, { "epoch": 0.5, "learning_rate": 2.522764509139714e-05, "loss": 2.7461, "step": 138800 }, { "epoch": 0.5, "learning_rate": 2.5209797573451462e-05, "loss": 2.7478, "step": 138900 }, { "epoch": 0.5, "learning_rate": 2.519195005550578e-05, "loss": 2.7389, "step": 139000 }, { "epoch": 0.5, "learning_rate": 2.51741025375601e-05, "loss": 2.7421, "step": 139100 }, { "epoch": 0.5, "learning_rate": 2.5156255019614422e-05, "loss": 2.7595, "step": 139200 }, { "epoch": 0.5, "learning_rate": 2.5138407501668743e-05, "loss": 2.7452, "step": 139300 }, { "epoch": 0.5, "learning_rate": 2.5120559983723064e-05, "loss": 2.746, "step": 139400 }, { "epoch": 0.5, "learning_rate": 2.5102712465777382e-05, "loss": 2.7434, "step": 139500 }, { "epoch": 0.5, "learning_rate": 2.5084864947831704e-05, "loss": 2.7466, "step": 139600 }, { "epoch": 0.5, "learning_rate": 2.5067017429886025e-05, "loss": 2.7494, "step": 139700 }, { "epoch": 0.5, "learning_rate": 2.5049169911940346e-05, "loss": 2.7442, "step": 139800 }, { "epoch": 0.5, "learning_rate": 2.5031322393994667e-05, "loss": 2.736, "step": 139900 }, { "epoch": 0.5, "learning_rate": 2.501347487604899e-05, "loss": 2.7473, "step": 140000 }, { "epoch": 0.5, "learning_rate": 2.499562735810331e-05, "loss": 2.7432, "step": 140100 }, { "epoch": 0.5, "learning_rate": 2.497777984015763e-05, "loss": 2.7406, "step": 140200 }, { "epoch": 0.5, "learning_rate": 2.4959932322211952e-05, "loss": 2.7418, "step": 140300 }, { "epoch": 0.5, "learning_rate": 2.4942084804266273e-05, "loss": 2.7518, "step": 140400 }, { "epoch": 0.5, "learning_rate": 2.492423728632059e-05, "loss": 2.7358, "step": 140500 }, { "epoch": 0.5, "learning_rate": 2.4906389768374912e-05, "loss": 2.7471, "step": 140600 }, { "epoch": 0.5, "learning_rate": 2.4888542250429233e-05, "loss": 2.7565, "step": 140700 }, { "epoch": 0.5, "learning_rate": 2.4870694732483555e-05, "loss": 2.7375, "step": 140800 }, { "epoch": 0.5, "learning_rate": 2.4852847214537876e-05, "loss": 2.7518, "step": 140900 }, { "epoch": 0.5, "learning_rate": 2.4834999696592197e-05, "loss": 2.7417, "step": 141000 }, { "epoch": 0.5, "learning_rate": 2.4817152178646515e-05, "loss": 2.7378, "step": 141100 }, { "epoch": 0.5, "learning_rate": 2.4799304660700836e-05, "loss": 2.7404, "step": 141200 }, { "epoch": 0.5, "learning_rate": 2.4781457142755157e-05, "loss": 2.7438, "step": 141300 }, { "epoch": 0.5, "learning_rate": 2.476360962480948e-05, "loss": 2.7558, "step": 141400 }, { "epoch": 0.51, "learning_rate": 2.47457621068638e-05, "loss": 2.7436, "step": 141500 }, { "epoch": 0.51, "learning_rate": 2.4727914588918118e-05, "loss": 2.7479, "step": 141600 }, { "epoch": 0.51, "learning_rate": 2.471006707097244e-05, "loss": 2.7583, "step": 141700 }, { "epoch": 0.51, "learning_rate": 2.4692219553026763e-05, "loss": 2.7381, "step": 141800 }, { "epoch": 0.51, "learning_rate": 2.4674372035081085e-05, "loss": 2.7397, "step": 141900 }, { "epoch": 0.51, "learning_rate": 2.4656524517135406e-05, "loss": 2.744, "step": 142000 }, { "epoch": 0.51, "learning_rate": 2.4638676999189724e-05, "loss": 2.7566, "step": 142100 }, { "epoch": 0.51, "learning_rate": 2.4620829481244045e-05, "loss": 2.7493, "step": 142200 }, { "epoch": 0.51, "learning_rate": 2.4602981963298366e-05, "loss": 2.7335, "step": 142300 }, { "epoch": 0.51, "learning_rate": 2.4585134445352687e-05, "loss": 2.7386, "step": 142400 }, { "epoch": 0.51, "learning_rate": 2.456728692740701e-05, "loss": 2.7502, "step": 142500 }, { "epoch": 0.51, "learning_rate": 2.4549439409461326e-05, "loss": 2.7461, "step": 142600 }, { "epoch": 0.51, "learning_rate": 2.4531591891515648e-05, "loss": 2.7352, "step": 142700 }, { "epoch": 0.51, "learning_rate": 2.451374437356997e-05, "loss": 2.7417, "step": 142800 }, { "epoch": 0.51, "learning_rate": 2.449589685562429e-05, "loss": 2.7435, "step": 142900 }, { "epoch": 0.51, "learning_rate": 2.447804933767861e-05, "loss": 2.748, "step": 143000 }, { "epoch": 0.51, "learning_rate": 2.446020181973293e-05, "loss": 2.7458, "step": 143100 }, { "epoch": 0.51, "learning_rate": 2.444235430178725e-05, "loss": 2.7492, "step": 143200 }, { "epoch": 0.51, "learning_rate": 2.442450678384157e-05, "loss": 2.7469, "step": 143300 }, { "epoch": 0.51, "learning_rate": 2.4406659265895893e-05, "loss": 2.7519, "step": 143400 }, { "epoch": 0.51, "learning_rate": 2.4388811747950214e-05, "loss": 2.7411, "step": 143500 }, { "epoch": 0.51, "learning_rate": 2.4370964230004535e-05, "loss": 2.7353, "step": 143600 }, { "epoch": 0.51, "learning_rate": 2.4353116712058853e-05, "loss": 2.7478, "step": 143700 }, { "epoch": 0.51, "learning_rate": 2.4335269194113174e-05, "loss": 2.7418, "step": 143800 }, { "epoch": 0.51, "learning_rate": 2.4317421676167495e-05, "loss": 2.7269, "step": 143900 }, { "epoch": 0.51, "learning_rate": 2.4299574158221817e-05, "loss": 2.7446, "step": 144000 }, { "epoch": 0.51, "learning_rate": 2.4281726640276138e-05, "loss": 2.7508, "step": 144100 }, { "epoch": 0.51, "learning_rate": 2.4263879122330456e-05, "loss": 2.7431, "step": 144200 }, { "epoch": 0.52, "learning_rate": 2.424603160438478e-05, "loss": 2.7478, "step": 144300 }, { "epoch": 0.52, "learning_rate": 2.42281840864391e-05, "loss": 2.7388, "step": 144400 }, { "epoch": 0.52, "learning_rate": 2.4210336568493423e-05, "loss": 2.7506, "step": 144500 }, { "epoch": 0.52, "learning_rate": 2.4192489050547744e-05, "loss": 2.7405, "step": 144600 }, { "epoch": 0.52, "learning_rate": 2.4174641532602062e-05, "loss": 2.7351, "step": 144700 }, { "epoch": 0.52, "learning_rate": 2.4156794014656383e-05, "loss": 2.7508, "step": 144800 }, { "epoch": 0.52, "learning_rate": 2.4138946496710704e-05, "loss": 2.7595, "step": 144900 }, { "epoch": 0.52, "learning_rate": 2.4121098978765025e-05, "loss": 2.7456, "step": 145000 }, { "epoch": 0.52, "learning_rate": 2.4103251460819347e-05, "loss": 2.7487, "step": 145100 }, { "epoch": 0.52, "learning_rate": 2.4085403942873664e-05, "loss": 2.7473, "step": 145200 }, { "epoch": 0.52, "learning_rate": 2.4067556424927986e-05, "loss": 2.7426, "step": 145300 }, { "epoch": 0.52, "learning_rate": 2.4049708906982307e-05, "loss": 2.7394, "step": 145400 }, { "epoch": 0.52, "learning_rate": 2.4031861389036628e-05, "loss": 2.7362, "step": 145500 }, { "epoch": 0.52, "learning_rate": 2.401401387109095e-05, "loss": 2.7418, "step": 145600 }, { "epoch": 0.52, "learning_rate": 2.399616635314527e-05, "loss": 2.7537, "step": 145700 }, { "epoch": 0.52, "learning_rate": 2.397831883519959e-05, "loss": 2.7451, "step": 145800 }, { "epoch": 0.52, "learning_rate": 2.396047131725391e-05, "loss": 2.7504, "step": 145900 }, { "epoch": 0.52, "learning_rate": 2.394262379930823e-05, "loss": 2.7472, "step": 146000 }, { "epoch": 0.52, "learning_rate": 2.3924776281362552e-05, "loss": 2.7374, "step": 146100 }, { "epoch": 0.52, "learning_rate": 2.3906928763416873e-05, "loss": 2.7421, "step": 146200 }, { "epoch": 0.52, "learning_rate": 2.388908124547119e-05, "loss": 2.745, "step": 146300 }, { "epoch": 0.52, "learning_rate": 2.3871233727525512e-05, "loss": 2.7375, "step": 146400 }, { "epoch": 0.52, "learning_rate": 2.3853386209579833e-05, "loss": 2.7346, "step": 146500 }, { "epoch": 0.52, "learning_rate": 2.3835538691634155e-05, "loss": 2.7361, "step": 146600 }, { "epoch": 0.52, "learning_rate": 2.3817691173688476e-05, "loss": 2.7455, "step": 146700 }, { "epoch": 0.52, "learning_rate": 2.3799843655742797e-05, "loss": 2.7492, "step": 146800 }, { "epoch": 0.52, "learning_rate": 2.378199613779712e-05, "loss": 2.7403, "step": 146900 }, { "epoch": 0.52, "learning_rate": 2.376414861985144e-05, "loss": 2.739, "step": 147000 }, { "epoch": 0.53, "learning_rate": 2.374630110190576e-05, "loss": 2.7435, "step": 147100 }, { "epoch": 0.53, "learning_rate": 2.3728453583960082e-05, "loss": 2.7487, "step": 147200 }, { "epoch": 0.53, "learning_rate": 2.37106060660144e-05, "loss": 2.7402, "step": 147300 }, { "epoch": 0.53, "learning_rate": 2.369275854806872e-05, "loss": 2.7438, "step": 147400 }, { "epoch": 0.53, "learning_rate": 2.3674911030123042e-05, "loss": 2.7439, "step": 147500 }, { "epoch": 0.53, "learning_rate": 2.3657063512177363e-05, "loss": 2.7467, "step": 147600 }, { "epoch": 0.53, "learning_rate": 2.3639215994231685e-05, "loss": 2.7309, "step": 147700 }, { "epoch": 0.53, "learning_rate": 2.3621368476286006e-05, "loss": 2.7259, "step": 147800 }, { "epoch": 0.53, "learning_rate": 2.3603520958340324e-05, "loss": 2.7411, "step": 147900 }, { "epoch": 0.53, "learning_rate": 2.3585673440394645e-05, "loss": 2.731, "step": 148000 }, { "epoch": 0.53, "learning_rate": 2.3567825922448966e-05, "loss": 2.7355, "step": 148100 }, { "epoch": 0.53, "learning_rate": 2.3549978404503287e-05, "loss": 2.7483, "step": 148200 }, { "epoch": 0.53, "learning_rate": 2.353213088655761e-05, "loss": 2.7381, "step": 148300 }, { "epoch": 0.53, "learning_rate": 2.3514283368611926e-05, "loss": 2.7356, "step": 148400 }, { "epoch": 0.53, "learning_rate": 2.3496435850666248e-05, "loss": 2.7512, "step": 148500 }, { "epoch": 0.53, "learning_rate": 2.347858833272057e-05, "loss": 2.7407, "step": 148600 }, { "epoch": 0.53, "learning_rate": 2.346074081477489e-05, "loss": 2.7338, "step": 148700 }, { "epoch": 0.53, "learning_rate": 2.344289329682921e-05, "loss": 2.7281, "step": 148800 }, { "epoch": 0.53, "learning_rate": 2.3425045778883532e-05, "loss": 2.7326, "step": 148900 }, { "epoch": 0.53, "learning_rate": 2.340719826093785e-05, "loss": 2.7227, "step": 149000 }, { "epoch": 0.53, "learning_rate": 2.338935074299217e-05, "loss": 2.7437, "step": 149100 }, { "epoch": 0.53, "learning_rate": 2.3371503225046493e-05, "loss": 2.739, "step": 149200 }, { "epoch": 0.53, "learning_rate": 2.3353655707100814e-05, "loss": 2.7323, "step": 149300 }, { "epoch": 0.53, "learning_rate": 2.3335808189155135e-05, "loss": 2.7362, "step": 149400 }, { "epoch": 0.53, "learning_rate": 2.3317960671209456e-05, "loss": 2.7386, "step": 149500 }, { "epoch": 0.53, "learning_rate": 2.3300113153263778e-05, "loss": 2.7508, "step": 149600 }, { "epoch": 0.53, "learning_rate": 2.32822656353181e-05, "loss": 2.7415, "step": 149700 }, { "epoch": 0.53, "learning_rate": 2.326441811737242e-05, "loss": 2.7551, "step": 149800 }, { "epoch": 0.54, "learning_rate": 2.324657059942674e-05, "loss": 2.7312, "step": 149900 }, { "epoch": 0.54, "learning_rate": 2.322872308148106e-05, "loss": 2.7526, "step": 150000 }, { "epoch": 0.54, "learning_rate": 2.321087556353538e-05, "loss": 2.7379, "step": 150100 }, { "epoch": 0.54, "learning_rate": 2.31930280455897e-05, "loss": 2.7416, "step": 150200 }, { "epoch": 0.54, "learning_rate": 2.3175180527644023e-05, "loss": 2.7331, "step": 150300 }, { "epoch": 0.54, "learning_rate": 2.3157333009698344e-05, "loss": 2.7272, "step": 150400 }, { "epoch": 0.54, "learning_rate": 2.3139485491752662e-05, "loss": 2.7433, "step": 150500 }, { "epoch": 0.54, "learning_rate": 2.3121637973806983e-05, "loss": 2.7358, "step": 150600 }, { "epoch": 0.54, "learning_rate": 2.3103790455861304e-05, "loss": 2.7218, "step": 150700 }, { "epoch": 0.54, "learning_rate": 2.3085942937915625e-05, "loss": 2.73, "step": 150800 }, { "epoch": 0.54, "learning_rate": 2.3068095419969947e-05, "loss": 2.734, "step": 150900 }, { "epoch": 0.54, "learning_rate": 2.3050247902024268e-05, "loss": 2.7469, "step": 151000 }, { "epoch": 0.54, "learning_rate": 2.3032400384078586e-05, "loss": 2.7356, "step": 151100 }, { "epoch": 0.54, "learning_rate": 2.3014552866132907e-05, "loss": 2.7421, "step": 151200 }, { "epoch": 0.54, "learning_rate": 2.2996705348187228e-05, "loss": 2.7388, "step": 151300 }, { "epoch": 0.54, "learning_rate": 2.297885783024155e-05, "loss": 2.7337, "step": 151400 }, { "epoch": 0.54, "learning_rate": 2.296101031229587e-05, "loss": 2.74, "step": 151500 }, { "epoch": 0.54, "learning_rate": 2.294316279435019e-05, "loss": 2.7336, "step": 151600 }, { "epoch": 0.54, "learning_rate": 2.292531527640451e-05, "loss": 2.7438, "step": 151700 }, { "epoch": 0.54, "learning_rate": 2.290746775845883e-05, "loss": 2.7376, "step": 151800 }, { "epoch": 0.54, "learning_rate": 2.2889620240513152e-05, "loss": 2.7321, "step": 151900 }, { "epoch": 0.54, "learning_rate": 2.2871772722567477e-05, "loss": 2.7423, "step": 152000 }, { "epoch": 0.54, "learning_rate": 2.2853925204621794e-05, "loss": 2.7312, "step": 152100 }, { "epoch": 0.54, "learning_rate": 2.2836077686676116e-05, "loss": 2.7442, "step": 152200 }, { "epoch": 0.54, "learning_rate": 2.2818230168730437e-05, "loss": 2.7308, "step": 152300 }, { "epoch": 0.54, "learning_rate": 2.2800382650784758e-05, "loss": 2.7277, "step": 152400 }, { "epoch": 0.54, "learning_rate": 2.278253513283908e-05, "loss": 2.7335, "step": 152500 }, { "epoch": 0.54, "learning_rate": 2.2764687614893397e-05, "loss": 2.7382, "step": 152600 }, { "epoch": 0.55, "learning_rate": 2.2746840096947718e-05, "loss": 2.7324, "step": 152700 }, { "epoch": 0.55, "learning_rate": 2.272899257900204e-05, "loss": 2.7361, "step": 152800 }, { "epoch": 0.55, "learning_rate": 2.271114506105636e-05, "loss": 2.7301, "step": 152900 }, { "epoch": 0.55, "learning_rate": 2.2693297543110682e-05, "loss": 2.7327, "step": 153000 }, { "epoch": 0.55, "learning_rate": 2.2675450025165003e-05, "loss": 2.7254, "step": 153100 }, { "epoch": 0.55, "learning_rate": 2.265760250721932e-05, "loss": 2.7393, "step": 153200 }, { "epoch": 0.55, "learning_rate": 2.2639754989273642e-05, "loss": 2.7282, "step": 153300 }, { "epoch": 0.55, "learning_rate": 2.2621907471327963e-05, "loss": 2.7323, "step": 153400 }, { "epoch": 0.55, "learning_rate": 2.2604059953382285e-05, "loss": 2.7472, "step": 153500 }, { "epoch": 0.55, "learning_rate": 2.2586212435436606e-05, "loss": 2.7348, "step": 153600 }, { "epoch": 0.55, "learning_rate": 2.2568364917490924e-05, "loss": 2.7405, "step": 153700 }, { "epoch": 0.55, "learning_rate": 2.2550517399545245e-05, "loss": 2.7285, "step": 153800 }, { "epoch": 0.55, "learning_rate": 2.2532669881599566e-05, "loss": 2.7314, "step": 153900 }, { "epoch": 0.55, "learning_rate": 2.2514822363653887e-05, "loss": 2.737, "step": 154000 }, { "epoch": 0.55, "learning_rate": 2.249697484570821e-05, "loss": 2.7351, "step": 154100 }, { "epoch": 0.55, "learning_rate": 2.247912732776253e-05, "loss": 2.7373, "step": 154200 }, { "epoch": 0.55, "learning_rate": 2.2461279809816848e-05, "loss": 2.7375, "step": 154300 }, { "epoch": 0.55, "learning_rate": 2.244343229187117e-05, "loss": 2.7319, "step": 154400 }, { "epoch": 0.55, "learning_rate": 2.2425584773925493e-05, "loss": 2.7364, "step": 154500 }, { "epoch": 0.55, "learning_rate": 2.2407737255979815e-05, "loss": 2.74, "step": 154600 }, { "epoch": 0.55, "learning_rate": 2.2389889738034132e-05, "loss": 2.7392, "step": 154700 }, { "epoch": 0.55, "learning_rate": 2.2372042220088454e-05, "loss": 2.7344, "step": 154800 }, { "epoch": 0.55, "learning_rate": 2.2354194702142775e-05, "loss": 2.7385, "step": 154900 }, { "epoch": 0.55, "learning_rate": 2.2336347184197096e-05, "loss": 2.7381, "step": 155000 }, { "epoch": 0.55, "learning_rate": 2.2318499666251417e-05, "loss": 2.7338, "step": 155100 }, { "epoch": 0.55, "learning_rate": 2.2300652148305735e-05, "loss": 2.735, "step": 155200 }, { "epoch": 0.55, "learning_rate": 2.2282804630360056e-05, "loss": 2.7421, "step": 155300 }, { "epoch": 0.55, "learning_rate": 2.2264957112414378e-05, "loss": 2.7342, "step": 155400 }, { "epoch": 0.56, "learning_rate": 2.22471095944687e-05, "loss": 2.7246, "step": 155500 }, { "epoch": 0.56, "learning_rate": 2.222926207652302e-05, "loss": 2.7373, "step": 155600 }, { "epoch": 0.56, "learning_rate": 2.221141455857734e-05, "loss": 2.7279, "step": 155700 }, { "epoch": 0.56, "learning_rate": 2.219356704063166e-05, "loss": 2.7401, "step": 155800 }, { "epoch": 0.56, "learning_rate": 2.217571952268598e-05, "loss": 2.7328, "step": 155900 }, { "epoch": 0.56, "learning_rate": 2.21578720047403e-05, "loss": 2.7382, "step": 156000 }, { "epoch": 0.56, "learning_rate": 2.2140024486794623e-05, "loss": 2.7401, "step": 156100 }, { "epoch": 0.56, "learning_rate": 2.2122176968848944e-05, "loss": 2.7447, "step": 156200 }, { "epoch": 0.56, "learning_rate": 2.2104329450903262e-05, "loss": 2.7364, "step": 156300 }, { "epoch": 0.56, "learning_rate": 2.2086481932957583e-05, "loss": 2.7355, "step": 156400 }, { "epoch": 0.56, "learning_rate": 2.2068634415011904e-05, "loss": 2.7373, "step": 156500 }, { "epoch": 0.56, "learning_rate": 2.2050786897066225e-05, "loss": 2.7383, "step": 156600 }, { "epoch": 0.56, "learning_rate": 2.2032939379120547e-05, "loss": 2.7296, "step": 156700 }, { "epoch": 0.56, "learning_rate": 2.2015091861174868e-05, "loss": 2.742, "step": 156800 }, { "epoch": 0.56, "learning_rate": 2.1997244343229186e-05, "loss": 2.7389, "step": 156900 }, { "epoch": 0.56, "learning_rate": 2.1979396825283507e-05, "loss": 2.7347, "step": 157000 }, { "epoch": 0.56, "learning_rate": 2.196154930733783e-05, "loss": 2.7319, "step": 157100 }, { "epoch": 0.56, "learning_rate": 2.1943701789392153e-05, "loss": 2.7376, "step": 157200 }, { "epoch": 0.56, "learning_rate": 2.192585427144647e-05, "loss": 2.7379, "step": 157300 }, { "epoch": 0.56, "learning_rate": 2.190800675350079e-05, "loss": 2.7259, "step": 157400 }, { "epoch": 0.56, "learning_rate": 2.1890159235555113e-05, "loss": 2.7286, "step": 157500 }, { "epoch": 0.56, "learning_rate": 2.1872311717609434e-05, "loss": 2.7355, "step": 157600 }, { "epoch": 0.56, "learning_rate": 2.1854464199663755e-05, "loss": 2.7356, "step": 157700 }, { "epoch": 0.56, "learning_rate": 2.1836616681718077e-05, "loss": 2.74, "step": 157800 }, { "epoch": 0.56, "learning_rate": 2.1818769163772394e-05, "loss": 2.7324, "step": 157900 }, { "epoch": 0.56, "learning_rate": 2.1800921645826716e-05, "loss": 2.7417, "step": 158000 }, { "epoch": 0.56, "learning_rate": 2.1783074127881037e-05, "loss": 2.7372, "step": 158100 }, { "epoch": 0.56, "learning_rate": 2.1765226609935358e-05, "loss": 2.7353, "step": 158200 }, { "epoch": 0.57, "learning_rate": 2.174737909198968e-05, "loss": 2.7333, "step": 158300 }, { "epoch": 0.57, "learning_rate": 2.1729531574043997e-05, "loss": 2.7282, "step": 158400 }, { "epoch": 0.57, "learning_rate": 2.1711684056098318e-05, "loss": 2.7233, "step": 158500 }, { "epoch": 0.57, "learning_rate": 2.169383653815264e-05, "loss": 2.7364, "step": 158600 }, { "epoch": 0.57, "learning_rate": 2.167598902020696e-05, "loss": 2.7266, "step": 158700 }, { "epoch": 0.57, "learning_rate": 2.1658141502261282e-05, "loss": 2.7387, "step": 158800 }, { "epoch": 0.57, "learning_rate": 2.1640293984315603e-05, "loss": 2.7276, "step": 158900 }, { "epoch": 0.57, "learning_rate": 2.162244646636992e-05, "loss": 2.731, "step": 159000 }, { "epoch": 0.57, "learning_rate": 2.1604598948424242e-05, "loss": 2.7318, "step": 159100 }, { "epoch": 0.57, "learning_rate": 2.1586751430478563e-05, "loss": 2.7236, "step": 159200 }, { "epoch": 0.57, "learning_rate": 2.1568903912532885e-05, "loss": 2.7312, "step": 159300 }, { "epoch": 0.57, "learning_rate": 2.1551056394587206e-05, "loss": 2.731, "step": 159400 }, { "epoch": 0.57, "learning_rate": 2.1533208876641524e-05, "loss": 2.731, "step": 159500 }, { "epoch": 0.57, "learning_rate": 2.1515361358695848e-05, "loss": 2.7295, "step": 159600 }, { "epoch": 0.57, "learning_rate": 2.149751384075017e-05, "loss": 2.7353, "step": 159700 }, { "epoch": 0.57, "learning_rate": 2.147966632280449e-05, "loss": 2.7237, "step": 159800 }, { "epoch": 0.57, "learning_rate": 2.1461818804858812e-05, "loss": 2.7234, "step": 159900 }, { "epoch": 0.57, "learning_rate": 2.144397128691313e-05, "loss": 2.7391, "step": 160000 }, { "epoch": 0.57, "learning_rate": 2.142612376896745e-05, "loss": 2.7354, "step": 160100 }, { "epoch": 0.57, "learning_rate": 2.1408276251021772e-05, "loss": 2.7322, "step": 160200 }, { "epoch": 0.57, "learning_rate": 2.1390428733076093e-05, "loss": 2.734, "step": 160300 }, { "epoch": 0.57, "learning_rate": 2.1372581215130415e-05, "loss": 2.7288, "step": 160400 }, { "epoch": 0.57, "learning_rate": 2.1354733697184732e-05, "loss": 2.7248, "step": 160500 }, { "epoch": 0.57, "learning_rate": 2.1336886179239054e-05, "loss": 2.7324, "step": 160600 }, { "epoch": 0.57, "learning_rate": 2.1319038661293375e-05, "loss": 2.7315, "step": 160700 }, { "epoch": 0.57, "learning_rate": 2.1301191143347696e-05, "loss": 2.7247, "step": 160800 }, { "epoch": 0.57, "learning_rate": 2.1283343625402017e-05, "loss": 2.7203, "step": 160900 }, { "epoch": 0.57, "learning_rate": 2.126549610745634e-05, "loss": 2.7303, "step": 161000 }, { "epoch": 0.58, "learning_rate": 2.1247648589510656e-05, "loss": 2.7359, "step": 161100 }, { "epoch": 0.58, "learning_rate": 2.1229801071564978e-05, "loss": 2.7361, "step": 161200 }, { "epoch": 0.58, "learning_rate": 2.12119535536193e-05, "loss": 2.7247, "step": 161300 }, { "epoch": 0.58, "learning_rate": 2.119410603567362e-05, "loss": 2.7246, "step": 161400 }, { "epoch": 0.58, "learning_rate": 2.117625851772794e-05, "loss": 2.7382, "step": 161500 }, { "epoch": 0.58, "learning_rate": 2.115841099978226e-05, "loss": 2.7319, "step": 161600 }, { "epoch": 0.58, "learning_rate": 2.114056348183658e-05, "loss": 2.7361, "step": 161700 }, { "epoch": 0.58, "learning_rate": 2.11227159638909e-05, "loss": 2.7345, "step": 161800 }, { "epoch": 0.58, "learning_rate": 2.1104868445945223e-05, "loss": 2.7344, "step": 161900 }, { "epoch": 0.58, "learning_rate": 2.1087020927999544e-05, "loss": 2.7123, "step": 162000 }, { "epoch": 0.58, "learning_rate": 2.1069173410053865e-05, "loss": 2.7427, "step": 162100 }, { "epoch": 0.58, "learning_rate": 2.1051325892108186e-05, "loss": 2.7383, "step": 162200 }, { "epoch": 0.58, "learning_rate": 2.1033478374162507e-05, "loss": 2.7346, "step": 162300 }, { "epoch": 0.58, "learning_rate": 2.101563085621683e-05, "loss": 2.7368, "step": 162400 }, { "epoch": 0.58, "learning_rate": 2.099778333827115e-05, "loss": 2.7379, "step": 162500 }, { "epoch": 0.58, "learning_rate": 2.0979935820325468e-05, "loss": 2.7412, "step": 162600 }, { "epoch": 0.58, "learning_rate": 2.096208830237979e-05, "loss": 2.7277, "step": 162700 }, { "epoch": 0.58, "learning_rate": 2.094424078443411e-05, "loss": 2.7263, "step": 162800 }, { "epoch": 0.58, "learning_rate": 2.092639326648843e-05, "loss": 2.743, "step": 162900 }, { "epoch": 0.58, "learning_rate": 2.0908545748542753e-05, "loss": 2.7327, "step": 163000 }, { "epoch": 0.58, "learning_rate": 2.0890698230597074e-05, "loss": 2.7256, "step": 163100 }, { "epoch": 0.58, "learning_rate": 2.087285071265139e-05, "loss": 2.7262, "step": 163200 }, { "epoch": 0.58, "learning_rate": 2.0855003194705713e-05, "loss": 2.7296, "step": 163300 }, { "epoch": 0.58, "learning_rate": 2.0837155676760034e-05, "loss": 2.7268, "step": 163400 }, { "epoch": 0.58, "learning_rate": 2.0819308158814355e-05, "loss": 2.7389, "step": 163500 }, { "epoch": 0.58, "learning_rate": 2.0801460640868677e-05, "loss": 2.7382, "step": 163600 }, { "epoch": 0.58, "learning_rate": 2.0783613122922994e-05, "loss": 2.7265, "step": 163700 }, { "epoch": 0.58, "learning_rate": 2.0765765604977316e-05, "loss": 2.7362, "step": 163800 }, { "epoch": 0.59, "learning_rate": 2.0747918087031637e-05, "loss": 2.7381, "step": 163900 }, { "epoch": 0.59, "learning_rate": 2.0730070569085958e-05, "loss": 2.7294, "step": 164000 }, { "epoch": 0.59, "learning_rate": 2.071222305114028e-05, "loss": 2.734, "step": 164100 }, { "epoch": 0.59, "learning_rate": 2.06943755331946e-05, "loss": 2.7334, "step": 164200 }, { "epoch": 0.59, "learning_rate": 2.0676528015248918e-05, "loss": 2.7351, "step": 164300 }, { "epoch": 0.59, "learning_rate": 2.065868049730324e-05, "loss": 2.7395, "step": 164400 }, { "epoch": 0.59, "learning_rate": 2.064083297935756e-05, "loss": 2.7273, "step": 164500 }, { "epoch": 0.59, "learning_rate": 2.0622985461411882e-05, "loss": 2.7393, "step": 164600 }, { "epoch": 0.59, "learning_rate": 2.0605137943466203e-05, "loss": 2.7142, "step": 164700 }, { "epoch": 0.59, "learning_rate": 2.0587290425520524e-05, "loss": 2.7215, "step": 164800 }, { "epoch": 0.59, "learning_rate": 2.0569442907574846e-05, "loss": 2.7487, "step": 164900 }, { "epoch": 0.59, "learning_rate": 2.0551595389629167e-05, "loss": 2.7345, "step": 165000 }, { "epoch": 0.59, "learning_rate": 2.0533747871683488e-05, "loss": 2.7379, "step": 165100 }, { "epoch": 0.59, "learning_rate": 2.051590035373781e-05, "loss": 2.7307, "step": 165200 }, { "epoch": 0.59, "learning_rate": 2.0498052835792127e-05, "loss": 2.7303, "step": 165300 }, { "epoch": 0.59, "learning_rate": 2.0480205317846448e-05, "loss": 2.7239, "step": 165400 }, { "epoch": 0.59, "learning_rate": 2.046235779990077e-05, "loss": 2.7309, "step": 165500 }, { "epoch": 0.59, "learning_rate": 2.044451028195509e-05, "loss": 2.7322, "step": 165600 }, { "epoch": 0.59, "learning_rate": 2.0426662764009412e-05, "loss": 2.733, "step": 165700 }, { "epoch": 0.59, "learning_rate": 2.040881524606373e-05, "loss": 2.7296, "step": 165800 }, { "epoch": 0.59, "learning_rate": 2.039096772811805e-05, "loss": 2.7305, "step": 165900 }, { "epoch": 0.59, "learning_rate": 2.0373120210172372e-05, "loss": 2.7323, "step": 166000 }, { "epoch": 0.59, "learning_rate": 2.0355272692226693e-05, "loss": 2.7186, "step": 166100 }, { "epoch": 0.59, "learning_rate": 2.0337425174281015e-05, "loss": 2.7257, "step": 166200 }, { "epoch": 0.59, "learning_rate": 2.0319577656335332e-05, "loss": 2.724, "step": 166300 }, { "epoch": 0.59, "learning_rate": 2.0301730138389654e-05, "loss": 2.744, "step": 166400 }, { "epoch": 0.59, "learning_rate": 2.0283882620443975e-05, "loss": 2.7325, "step": 166500 }, { "epoch": 0.59, "learning_rate": 2.0266035102498296e-05, "loss": 2.7355, "step": 166600 }, { "epoch": 0.6, "learning_rate": 2.0248187584552617e-05, "loss": 2.7348, "step": 166700 }, { "epoch": 0.6, "learning_rate": 2.023034006660694e-05, "loss": 2.7232, "step": 166800 }, { "epoch": 0.6, "learning_rate": 2.0212492548661256e-05, "loss": 2.7342, "step": 166900 }, { "epoch": 0.6, "learning_rate": 2.0194645030715577e-05, "loss": 2.736, "step": 167000 }, { "epoch": 0.6, "learning_rate": 2.01767975127699e-05, "loss": 2.7335, "step": 167100 }, { "epoch": 0.6, "learning_rate": 2.015894999482422e-05, "loss": 2.7272, "step": 167200 }, { "epoch": 0.6, "learning_rate": 2.014110247687854e-05, "loss": 2.7345, "step": 167300 }, { "epoch": 0.6, "learning_rate": 2.0123254958932862e-05, "loss": 2.7304, "step": 167400 }, { "epoch": 0.6, "learning_rate": 2.0105407440987184e-05, "loss": 2.7237, "step": 167500 }, { "epoch": 0.6, "learning_rate": 2.0087559923041505e-05, "loss": 2.7293, "step": 167600 }, { "epoch": 0.6, "learning_rate": 2.0069712405095826e-05, "loss": 2.7264, "step": 167700 }, { "epoch": 0.6, "learning_rate": 2.0051864887150147e-05, "loss": 2.7075, "step": 167800 }, { "epoch": 0.6, "learning_rate": 2.0034017369204465e-05, "loss": 2.736, "step": 167900 }, { "epoch": 0.6, "learning_rate": 2.0016169851258786e-05, "loss": 2.7302, "step": 168000 }, { "epoch": 0.6, "learning_rate": 1.9998322333313107e-05, "loss": 2.7386, "step": 168100 }, { "epoch": 0.6, "learning_rate": 1.998047481536743e-05, "loss": 2.7324, "step": 168200 }, { "epoch": 0.6, "learning_rate": 1.996262729742175e-05, "loss": 2.7342, "step": 168300 }, { "epoch": 0.6, "learning_rate": 1.9944779779476068e-05, "loss": 2.7263, "step": 168400 }, { "epoch": 0.6, "learning_rate": 1.992693226153039e-05, "loss": 2.728, "step": 168500 }, { "epoch": 0.6, "learning_rate": 1.990908474358471e-05, "loss": 2.7226, "step": 168600 }, { "epoch": 0.6, "learning_rate": 1.989123722563903e-05, "loss": 2.7231, "step": 168700 }, { "epoch": 0.6, "learning_rate": 1.9873389707693353e-05, "loss": 2.7286, "step": 168800 }, { "epoch": 0.6, "learning_rate": 1.9855542189747674e-05, "loss": 2.7298, "step": 168900 }, { "epoch": 0.6, "learning_rate": 1.983769467180199e-05, "loss": 2.727, "step": 169000 }, { "epoch": 0.6, "learning_rate": 1.9819847153856313e-05, "loss": 2.728, "step": 169100 }, { "epoch": 0.6, "learning_rate": 1.9801999635910634e-05, "loss": 2.7248, "step": 169200 }, { "epoch": 0.6, "learning_rate": 1.9784152117964955e-05, "loss": 2.7209, "step": 169300 }, { "epoch": 0.6, "learning_rate": 1.9766304600019276e-05, "loss": 2.7347, "step": 169400 }, { "epoch": 0.61, "learning_rate": 1.9748457082073594e-05, "loss": 2.723, "step": 169500 }, { "epoch": 0.61, "learning_rate": 1.9730609564127916e-05, "loss": 2.7269, "step": 169600 }, { "epoch": 0.61, "learning_rate": 1.9712762046182237e-05, "loss": 2.7281, "step": 169700 }, { "epoch": 0.61, "learning_rate": 1.969491452823656e-05, "loss": 2.7226, "step": 169800 }, { "epoch": 0.61, "learning_rate": 1.9677067010290883e-05, "loss": 2.7264, "step": 169900 }, { "epoch": 0.61, "learning_rate": 1.96592194923452e-05, "loss": 2.7239, "step": 170000 }, { "epoch": 0.61, "learning_rate": 1.964137197439952e-05, "loss": 2.7346, "step": 170100 }, { "epoch": 0.61, "learning_rate": 1.9623524456453843e-05, "loss": 2.7238, "step": 170200 }, { "epoch": 0.61, "learning_rate": 1.9605676938508164e-05, "loss": 2.7189, "step": 170300 }, { "epoch": 0.61, "learning_rate": 1.9587829420562485e-05, "loss": 2.71, "step": 170400 }, { "epoch": 0.61, "learning_rate": 1.9569981902616803e-05, "loss": 2.7351, "step": 170500 }, { "epoch": 0.61, "learning_rate": 1.9552134384671124e-05, "loss": 2.7317, "step": 170600 }, { "epoch": 0.61, "learning_rate": 1.9534286866725445e-05, "loss": 2.7294, "step": 170700 }, { "epoch": 0.61, "learning_rate": 1.9516439348779767e-05, "loss": 2.7275, "step": 170800 }, { "epoch": 0.61, "learning_rate": 1.9498591830834088e-05, "loss": 2.7257, "step": 170900 }, { "epoch": 0.61, "learning_rate": 1.948074431288841e-05, "loss": 2.7292, "step": 171000 }, { "epoch": 0.61, "learning_rate": 1.9462896794942727e-05, "loss": 2.7256, "step": 171100 }, { "epoch": 0.61, "learning_rate": 1.9445049276997048e-05, "loss": 2.7299, "step": 171200 }, { "epoch": 0.61, "learning_rate": 1.942720175905137e-05, "loss": 2.7187, "step": 171300 }, { "epoch": 0.61, "learning_rate": 1.940935424110569e-05, "loss": 2.7382, "step": 171400 }, { "epoch": 0.61, "learning_rate": 1.9391506723160012e-05, "loss": 2.7323, "step": 171500 }, { "epoch": 0.61, "learning_rate": 1.937365920521433e-05, "loss": 2.7233, "step": 171600 }, { "epoch": 0.61, "learning_rate": 1.935581168726865e-05, "loss": 2.7148, "step": 171700 }, { "epoch": 0.61, "learning_rate": 1.9337964169322972e-05, "loss": 2.727, "step": 171800 }, { "epoch": 0.61, "learning_rate": 1.9320116651377293e-05, "loss": 2.731, "step": 171900 }, { "epoch": 0.61, "learning_rate": 1.9302269133431615e-05, "loss": 2.7285, "step": 172000 }, { "epoch": 0.61, "learning_rate": 1.9284421615485936e-05, "loss": 2.7351, "step": 172100 }, { "epoch": 0.61, "learning_rate": 1.9266574097540254e-05, "loss": 2.7242, "step": 172200 }, { "epoch": 0.62, "learning_rate": 1.9248726579594575e-05, "loss": 2.7252, "step": 172300 }, { "epoch": 0.62, "learning_rate": 1.92308790616489e-05, "loss": 2.7327, "step": 172400 }, { "epoch": 0.62, "learning_rate": 1.921303154370322e-05, "loss": 2.7288, "step": 172500 }, { "epoch": 0.62, "learning_rate": 1.919518402575754e-05, "loss": 2.7168, "step": 172600 }, { "epoch": 0.62, "learning_rate": 1.917733650781186e-05, "loss": 2.7302, "step": 172700 }, { "epoch": 0.62, "learning_rate": 1.915948898986618e-05, "loss": 2.7306, "step": 172800 }, { "epoch": 0.62, "learning_rate": 1.9141641471920502e-05, "loss": 2.7166, "step": 172900 }, { "epoch": 0.62, "learning_rate": 1.9123793953974823e-05, "loss": 2.7168, "step": 173000 }, { "epoch": 0.62, "learning_rate": 1.9105946436029144e-05, "loss": 2.736, "step": 173100 }, { "epoch": 0.62, "learning_rate": 1.9088098918083462e-05, "loss": 2.724, "step": 173200 }, { "epoch": 0.62, "learning_rate": 1.9070251400137784e-05, "loss": 2.7258, "step": 173300 }, { "epoch": 0.62, "learning_rate": 1.9052403882192105e-05, "loss": 2.723, "step": 173400 }, { "epoch": 0.62, "learning_rate": 1.9034556364246426e-05, "loss": 2.7322, "step": 173500 }, { "epoch": 0.62, "learning_rate": 1.9016708846300747e-05, "loss": 2.7307, "step": 173600 }, { "epoch": 0.62, "learning_rate": 1.8998861328355065e-05, "loss": 2.7214, "step": 173700 }, { "epoch": 0.62, "learning_rate": 1.8981013810409386e-05, "loss": 2.7123, "step": 173800 }, { "epoch": 0.62, "learning_rate": 1.8963166292463707e-05, "loss": 2.731, "step": 173900 }, { "epoch": 0.62, "learning_rate": 1.894531877451803e-05, "loss": 2.7262, "step": 174000 }, { "epoch": 0.62, "learning_rate": 1.892747125657235e-05, "loss": 2.7263, "step": 174100 }, { "epoch": 0.62, "learning_rate": 1.890962373862667e-05, "loss": 2.7317, "step": 174200 }, { "epoch": 0.62, "learning_rate": 1.889177622068099e-05, "loss": 2.7196, "step": 174300 }, { "epoch": 0.62, "learning_rate": 1.887392870273531e-05, "loss": 2.7145, "step": 174400 }, { "epoch": 0.62, "learning_rate": 1.885608118478963e-05, "loss": 2.731, "step": 174500 }, { "epoch": 0.62, "learning_rate": 1.8838233666843953e-05, "loss": 2.7168, "step": 174600 }, { "epoch": 0.62, "learning_rate": 1.8820386148898274e-05, "loss": 2.7146, "step": 174700 }, { "epoch": 0.62, "learning_rate": 1.880253863095259e-05, "loss": 2.7292, "step": 174800 }, { "epoch": 0.62, "learning_rate": 1.8784691113006913e-05, "loss": 2.7264, "step": 174900 }, { "epoch": 0.62, "learning_rate": 1.8766843595061237e-05, "loss": 2.7189, "step": 175000 }, { "epoch": 0.63, "learning_rate": 1.874899607711556e-05, "loss": 2.7263, "step": 175100 }, { "epoch": 0.63, "learning_rate": 1.873114855916988e-05, "loss": 2.7316, "step": 175200 }, { "epoch": 0.63, "learning_rate": 1.8713301041224198e-05, "loss": 2.7149, "step": 175300 }, { "epoch": 0.63, "learning_rate": 1.869545352327852e-05, "loss": 2.7134, "step": 175400 }, { "epoch": 0.63, "learning_rate": 1.867760600533284e-05, "loss": 2.7276, "step": 175500 }, { "epoch": 0.63, "learning_rate": 1.865975848738716e-05, "loss": 2.7289, "step": 175600 }, { "epoch": 0.63, "learning_rate": 1.8641910969441483e-05, "loss": 2.7301, "step": 175700 }, { "epoch": 0.63, "learning_rate": 1.86240634514958e-05, "loss": 2.7229, "step": 175800 }, { "epoch": 0.63, "learning_rate": 1.860621593355012e-05, "loss": 2.7207, "step": 175900 }, { "epoch": 0.63, "learning_rate": 1.8588368415604443e-05, "loss": 2.7187, "step": 176000 }, { "epoch": 0.63, "learning_rate": 1.8570520897658764e-05, "loss": 2.7242, "step": 176100 }, { "epoch": 0.63, "learning_rate": 1.8552673379713085e-05, "loss": 2.7331, "step": 176200 }, { "epoch": 0.63, "learning_rate": 1.8534825861767406e-05, "loss": 2.728, "step": 176300 }, { "epoch": 0.63, "learning_rate": 1.8516978343821724e-05, "loss": 2.7247, "step": 176400 }, { "epoch": 0.63, "learning_rate": 1.8499130825876045e-05, "loss": 2.7274, "step": 176500 }, { "epoch": 0.63, "learning_rate": 1.8481283307930367e-05, "loss": 2.7342, "step": 176600 }, { "epoch": 0.63, "learning_rate": 1.8463435789984688e-05, "loss": 2.7282, "step": 176700 }, { "epoch": 0.63, "learning_rate": 1.844558827203901e-05, "loss": 2.7143, "step": 176800 }, { "epoch": 0.63, "learning_rate": 1.8427740754093327e-05, "loss": 2.7194, "step": 176900 }, { "epoch": 0.63, "learning_rate": 1.8409893236147648e-05, "loss": 2.7324, "step": 177000 }, { "epoch": 0.63, "learning_rate": 1.839204571820197e-05, "loss": 2.7307, "step": 177100 }, { "epoch": 0.63, "learning_rate": 1.837419820025629e-05, "loss": 2.7296, "step": 177200 }, { "epoch": 0.63, "learning_rate": 1.8356350682310612e-05, "loss": 2.7236, "step": 177300 }, { "epoch": 0.63, "learning_rate": 1.833850316436493e-05, "loss": 2.7156, "step": 177400 }, { "epoch": 0.63, "learning_rate": 1.8320655646419254e-05, "loss": 2.7224, "step": 177500 }, { "epoch": 0.63, "learning_rate": 1.8302808128473575e-05, "loss": 2.7156, "step": 177600 }, { "epoch": 0.63, "learning_rate": 1.8284960610527897e-05, "loss": 2.724, "step": 177700 }, { "epoch": 0.63, "learning_rate": 1.8267113092582218e-05, "loss": 2.7244, "step": 177800 }, { "epoch": 0.64, "learning_rate": 1.8249265574636536e-05, "loss": 2.7247, "step": 177900 }, { "epoch": 0.64, "learning_rate": 1.8231418056690857e-05, "loss": 2.7214, "step": 178000 }, { "epoch": 0.64, "learning_rate": 1.8213570538745178e-05, "loss": 2.7257, "step": 178100 }, { "epoch": 0.64, "learning_rate": 1.81957230207995e-05, "loss": 2.7253, "step": 178200 }, { "epoch": 0.64, "learning_rate": 1.817787550285382e-05, "loss": 2.7244, "step": 178300 }, { "epoch": 0.64, "learning_rate": 1.816002798490814e-05, "loss": 2.7392, "step": 178400 }, { "epoch": 0.64, "learning_rate": 1.814218046696246e-05, "loss": 2.7217, "step": 178500 }, { "epoch": 0.64, "learning_rate": 1.812433294901678e-05, "loss": 2.7255, "step": 178600 }, { "epoch": 0.64, "learning_rate": 1.8106485431071102e-05, "loss": 2.7261, "step": 178700 }, { "epoch": 0.64, "learning_rate": 1.8088637913125423e-05, "loss": 2.7218, "step": 178800 }, { "epoch": 0.64, "learning_rate": 1.8070790395179744e-05, "loss": 2.7325, "step": 178900 }, { "epoch": 0.64, "learning_rate": 1.8052942877234062e-05, "loss": 2.7252, "step": 179000 }, { "epoch": 0.64, "learning_rate": 1.8035095359288384e-05, "loss": 2.7204, "step": 179100 }, { "epoch": 0.64, "learning_rate": 1.8017247841342705e-05, "loss": 2.7163, "step": 179200 }, { "epoch": 0.64, "learning_rate": 1.7999400323397026e-05, "loss": 2.7288, "step": 179300 }, { "epoch": 0.64, "learning_rate": 1.7981552805451347e-05, "loss": 2.7302, "step": 179400 }, { "epoch": 0.64, "learning_rate": 1.7963705287505665e-05, "loss": 2.7141, "step": 179500 }, { "epoch": 0.64, "learning_rate": 1.7945857769559986e-05, "loss": 2.7133, "step": 179600 }, { "epoch": 0.64, "learning_rate": 1.7928010251614307e-05, "loss": 2.7207, "step": 179700 }, { "epoch": 0.64, "learning_rate": 1.791016273366863e-05, "loss": 2.7191, "step": 179800 }, { "epoch": 0.64, "learning_rate": 1.789231521572295e-05, "loss": 2.7188, "step": 179900 }, { "epoch": 0.64, "learning_rate": 1.787446769777727e-05, "loss": 2.7159, "step": 180000 }, { "epoch": 0.64, "learning_rate": 1.7856620179831592e-05, "loss": 2.7176, "step": 180100 }, { "epoch": 0.64, "learning_rate": 1.7838772661885913e-05, "loss": 2.7271, "step": 180200 }, { "epoch": 0.64, "learning_rate": 1.7820925143940235e-05, "loss": 2.7298, "step": 180300 }, { "epoch": 0.64, "learning_rate": 1.7803077625994556e-05, "loss": 2.7158, "step": 180400 }, { "epoch": 0.64, "learning_rate": 1.7785230108048874e-05, "loss": 2.733, "step": 180500 }, { "epoch": 0.64, "learning_rate": 1.7767382590103195e-05, "loss": 2.7242, "step": 180600 }, { "epoch": 0.65, "learning_rate": 1.7749535072157516e-05, "loss": 2.7177, "step": 180700 }, { "epoch": 0.65, "learning_rate": 1.7731687554211837e-05, "loss": 2.7187, "step": 180800 }, { "epoch": 0.65, "learning_rate": 1.771384003626616e-05, "loss": 2.7196, "step": 180900 }, { "epoch": 0.65, "learning_rate": 1.769599251832048e-05, "loss": 2.7254, "step": 181000 }, { "epoch": 0.65, "learning_rate": 1.7678145000374798e-05, "loss": 2.717, "step": 181100 }, { "epoch": 0.65, "learning_rate": 1.766029748242912e-05, "loss": 2.7162, "step": 181200 }, { "epoch": 0.65, "learning_rate": 1.764244996448344e-05, "loss": 2.7167, "step": 181300 }, { "epoch": 0.65, "learning_rate": 1.762460244653776e-05, "loss": 2.7216, "step": 181400 }, { "epoch": 0.65, "learning_rate": 1.7606754928592082e-05, "loss": 2.7254, "step": 181500 }, { "epoch": 0.65, "learning_rate": 1.75889074106464e-05, "loss": 2.72, "step": 181600 }, { "epoch": 0.65, "learning_rate": 1.757105989270072e-05, "loss": 2.7132, "step": 181700 }, { "epoch": 0.65, "learning_rate": 1.7553212374755043e-05, "loss": 2.723, "step": 181800 }, { "epoch": 0.65, "learning_rate": 1.7535364856809364e-05, "loss": 2.7191, "step": 181900 }, { "epoch": 0.65, "learning_rate": 1.7517517338863685e-05, "loss": 2.7216, "step": 182000 }, { "epoch": 0.65, "learning_rate": 1.7499669820918006e-05, "loss": 2.7184, "step": 182100 }, { "epoch": 0.65, "learning_rate": 1.7481822302972324e-05, "loss": 2.7303, "step": 182200 }, { "epoch": 0.65, "learning_rate": 1.7463974785026645e-05, "loss": 2.7074, "step": 182300 }, { "epoch": 0.65, "learning_rate": 1.7446127267080967e-05, "loss": 2.7286, "step": 182400 }, { "epoch": 0.65, "learning_rate": 1.7428279749135288e-05, "loss": 2.7204, "step": 182500 }, { "epoch": 0.65, "learning_rate": 1.741043223118961e-05, "loss": 2.7159, "step": 182600 }, { "epoch": 0.65, "learning_rate": 1.739258471324393e-05, "loss": 2.7101, "step": 182700 }, { "epoch": 0.65, "learning_rate": 1.737473719529825e-05, "loss": 2.7151, "step": 182800 }, { "epoch": 0.65, "learning_rate": 1.7356889677352573e-05, "loss": 2.7056, "step": 182900 }, { "epoch": 0.65, "learning_rate": 1.7339042159406894e-05, "loss": 2.7251, "step": 183000 }, { "epoch": 0.65, "learning_rate": 1.7321194641461215e-05, "loss": 2.7264, "step": 183100 }, { "epoch": 0.65, "learning_rate": 1.7303347123515533e-05, "loss": 2.7249, "step": 183200 }, { "epoch": 0.65, "learning_rate": 1.7285499605569854e-05, "loss": 2.7095, "step": 183300 }, { "epoch": 0.65, "learning_rate": 1.7267652087624175e-05, "loss": 2.7235, "step": 183400 }, { "epoch": 0.66, "learning_rate": 1.7249804569678497e-05, "loss": 2.7213, "step": 183500 }, { "epoch": 0.66, "learning_rate": 1.7231957051732818e-05, "loss": 2.7328, "step": 183600 }, { "epoch": 0.66, "learning_rate": 1.7214109533787136e-05, "loss": 2.7092, "step": 183700 }, { "epoch": 0.66, "learning_rate": 1.7196262015841457e-05, "loss": 2.7146, "step": 183800 }, { "epoch": 0.66, "learning_rate": 1.7178414497895778e-05, "loss": 2.7208, "step": 183900 }, { "epoch": 0.66, "learning_rate": 1.71605669799501e-05, "loss": 2.723, "step": 184000 }, { "epoch": 0.66, "learning_rate": 1.714271946200442e-05, "loss": 2.7104, "step": 184100 }, { "epoch": 0.66, "learning_rate": 1.7124871944058742e-05, "loss": 2.7221, "step": 184200 }, { "epoch": 0.66, "learning_rate": 1.710702442611306e-05, "loss": 2.7253, "step": 184300 }, { "epoch": 0.66, "learning_rate": 1.708917690816738e-05, "loss": 2.712, "step": 184400 }, { "epoch": 0.66, "learning_rate": 1.7071329390221702e-05, "loss": 2.723, "step": 184500 }, { "epoch": 0.66, "learning_rate": 1.7053481872276023e-05, "loss": 2.7147, "step": 184600 }, { "epoch": 0.66, "learning_rate": 1.7035634354330344e-05, "loss": 2.7115, "step": 184700 }, { "epoch": 0.66, "learning_rate": 1.7017786836384662e-05, "loss": 2.7187, "step": 184800 }, { "epoch": 0.66, "learning_rate": 1.6999939318438983e-05, "loss": 2.7232, "step": 184900 }, { "epoch": 0.66, "learning_rate": 1.6982091800493305e-05, "loss": 2.7248, "step": 185000 }, { "epoch": 0.66, "learning_rate": 1.6964244282547626e-05, "loss": 2.7092, "step": 185100 }, { "epoch": 0.66, "learning_rate": 1.694639676460195e-05, "loss": 2.7228, "step": 185200 }, { "epoch": 0.66, "learning_rate": 1.692854924665627e-05, "loss": 2.7268, "step": 185300 }, { "epoch": 0.66, "learning_rate": 1.691070172871059e-05, "loss": 2.7142, "step": 185400 }, { "epoch": 0.66, "learning_rate": 1.689285421076491e-05, "loss": 2.7131, "step": 185500 }, { "epoch": 0.66, "learning_rate": 1.6875006692819232e-05, "loss": 2.7241, "step": 185600 }, { "epoch": 0.66, "learning_rate": 1.6857159174873553e-05, "loss": 2.7147, "step": 185700 }, { "epoch": 0.66, "learning_rate": 1.683931165692787e-05, "loss": 2.7145, "step": 185800 }, { "epoch": 0.66, "learning_rate": 1.6821464138982192e-05, "loss": 2.7177, "step": 185900 }, { "epoch": 0.66, "learning_rate": 1.6803616621036513e-05, "loss": 2.7221, "step": 186000 }, { "epoch": 0.66, "learning_rate": 1.6785769103090835e-05, "loss": 2.7218, "step": 186100 }, { "epoch": 0.66, "learning_rate": 1.6767921585145156e-05, "loss": 2.7185, "step": 186200 }, { "epoch": 0.66, "learning_rate": 1.6750074067199477e-05, "loss": 2.7084, "step": 186300 }, { "epoch": 0.67, "learning_rate": 1.6732226549253795e-05, "loss": 2.7109, "step": 186400 }, { "epoch": 0.67, "learning_rate": 1.6714379031308116e-05, "loss": 2.7236, "step": 186500 }, { "epoch": 0.67, "learning_rate": 1.6696531513362437e-05, "loss": 2.732, "step": 186600 }, { "epoch": 0.67, "learning_rate": 1.667868399541676e-05, "loss": 2.7222, "step": 186700 }, { "epoch": 0.67, "learning_rate": 1.666083647747108e-05, "loss": 2.7112, "step": 186800 }, { "epoch": 0.67, "learning_rate": 1.6642988959525398e-05, "loss": 2.7275, "step": 186900 }, { "epoch": 0.67, "learning_rate": 1.662514144157972e-05, "loss": 2.7227, "step": 187000 }, { "epoch": 0.67, "learning_rate": 1.660729392363404e-05, "loss": 2.723, "step": 187100 }, { "epoch": 0.67, "learning_rate": 1.658944640568836e-05, "loss": 2.732, "step": 187200 }, { "epoch": 0.67, "learning_rate": 1.6571598887742682e-05, "loss": 2.7154, "step": 187300 }, { "epoch": 0.67, "learning_rate": 1.6553751369797004e-05, "loss": 2.7151, "step": 187400 }, { "epoch": 0.67, "learning_rate": 1.653590385185132e-05, "loss": 2.7072, "step": 187500 }, { "epoch": 0.67, "learning_rate": 1.6518056333905643e-05, "loss": 2.7216, "step": 187600 }, { "epoch": 0.67, "learning_rate": 1.6500208815959967e-05, "loss": 2.7015, "step": 187700 }, { "epoch": 0.67, "learning_rate": 1.648236129801429e-05, "loss": 2.7158, "step": 187800 }, { "epoch": 0.67, "learning_rate": 1.6464513780068606e-05, "loss": 2.7235, "step": 187900 }, { "epoch": 0.67, "learning_rate": 1.6446666262122928e-05, "loss": 2.7088, "step": 188000 }, { "epoch": 0.67, "learning_rate": 1.642881874417725e-05, "loss": 2.7184, "step": 188100 }, { "epoch": 0.67, "learning_rate": 1.641097122623157e-05, "loss": 2.7156, "step": 188200 }, { "epoch": 0.67, "learning_rate": 1.639312370828589e-05, "loss": 2.7184, "step": 188300 }, { "epoch": 0.67, "learning_rate": 1.6375276190340212e-05, "loss": 2.7243, "step": 188400 }, { "epoch": 0.67, "learning_rate": 1.635742867239453e-05, "loss": 2.7248, "step": 188500 }, { "epoch": 0.67, "learning_rate": 1.633958115444885e-05, "loss": 2.7154, "step": 188600 }, { "epoch": 0.67, "learning_rate": 1.6321733636503173e-05, "loss": 2.7176, "step": 188700 }, { "epoch": 0.67, "learning_rate": 1.6303886118557494e-05, "loss": 2.7134, "step": 188800 }, { "epoch": 0.67, "learning_rate": 1.6286038600611815e-05, "loss": 2.7259, "step": 188900 }, { "epoch": 0.67, "learning_rate": 1.6268191082666133e-05, "loss": 2.7226, "step": 189000 }, { "epoch": 0.67, "learning_rate": 1.6250343564720454e-05, "loss": 2.7062, "step": 189100 }, { "epoch": 0.68, "learning_rate": 1.6232496046774775e-05, "loss": 2.7082, "step": 189200 }, { "epoch": 0.68, "learning_rate": 1.6214648528829097e-05, "loss": 2.712, "step": 189300 }, { "epoch": 0.68, "learning_rate": 1.6196801010883418e-05, "loss": 2.7137, "step": 189400 }, { "epoch": 0.68, "learning_rate": 1.6178953492937736e-05, "loss": 2.7242, "step": 189500 }, { "epoch": 0.68, "learning_rate": 1.6161105974992057e-05, "loss": 2.7074, "step": 189600 }, { "epoch": 0.68, "learning_rate": 1.6143258457046378e-05, "loss": 2.7128, "step": 189700 }, { "epoch": 0.68, "learning_rate": 1.61254109391007e-05, "loss": 2.726, "step": 189800 }, { "epoch": 0.68, "learning_rate": 1.610756342115502e-05, "loss": 2.7125, "step": 189900 }, { "epoch": 0.68, "learning_rate": 1.6089715903209342e-05, "loss": 2.7131, "step": 190000 }, { "epoch": 0.68, "learning_rate": 1.607186838526366e-05, "loss": 2.7301, "step": 190100 }, { "epoch": 0.68, "learning_rate": 1.605402086731798e-05, "loss": 2.7243, "step": 190200 }, { "epoch": 0.68, "learning_rate": 1.6036173349372305e-05, "loss": 2.715, "step": 190300 }, { "epoch": 0.68, "learning_rate": 1.6018325831426627e-05, "loss": 2.7066, "step": 190400 }, { "epoch": 0.68, "learning_rate": 1.6000478313480944e-05, "loss": 2.7182, "step": 190500 }, { "epoch": 0.68, "learning_rate": 1.5982630795535266e-05, "loss": 2.7147, "step": 190600 }, { "epoch": 0.68, "learning_rate": 1.5964783277589587e-05, "loss": 2.7144, "step": 190700 }, { "epoch": 0.68, "learning_rate": 1.5946935759643908e-05, "loss": 2.7222, "step": 190800 }, { "epoch": 0.68, "learning_rate": 1.592908824169823e-05, "loss": 2.7202, "step": 190900 }, { "epoch": 0.68, "learning_rate": 1.591124072375255e-05, "loss": 2.7086, "step": 191000 }, { "epoch": 0.68, "learning_rate": 1.589339320580687e-05, "loss": 2.7141, "step": 191100 }, { "epoch": 0.68, "learning_rate": 1.587554568786119e-05, "loss": 2.7148, "step": 191200 }, { "epoch": 0.68, "learning_rate": 1.585769816991551e-05, "loss": 2.7037, "step": 191300 }, { "epoch": 0.68, "learning_rate": 1.5839850651969832e-05, "loss": 2.7298, "step": 191400 }, { "epoch": 0.68, "learning_rate": 1.5822003134024153e-05, "loss": 2.7071, "step": 191500 }, { "epoch": 0.68, "learning_rate": 1.580415561607847e-05, "loss": 2.71, "step": 191600 }, { "epoch": 0.68, "learning_rate": 1.5786308098132792e-05, "loss": 2.7058, "step": 191700 }, { "epoch": 0.68, "learning_rate": 1.5768460580187113e-05, "loss": 2.7256, "step": 191800 }, { "epoch": 0.68, "learning_rate": 1.5750613062241435e-05, "loss": 2.7194, "step": 191900 }, { "epoch": 0.69, "learning_rate": 1.5732765544295756e-05, "loss": 2.7117, "step": 192000 }, { "epoch": 0.69, "learning_rate": 1.5714918026350077e-05, "loss": 2.7162, "step": 192100 }, { "epoch": 0.69, "learning_rate": 1.5697070508404395e-05, "loss": 2.7219, "step": 192200 }, { "epoch": 0.69, "learning_rate": 1.5679222990458716e-05, "loss": 2.716, "step": 192300 }, { "epoch": 0.69, "learning_rate": 1.5661375472513037e-05, "loss": 2.7156, "step": 192400 }, { "epoch": 0.69, "learning_rate": 1.564352795456736e-05, "loss": 2.7103, "step": 192500 }, { "epoch": 0.69, "learning_rate": 1.562568043662168e-05, "loss": 2.7085, "step": 192600 }, { "epoch": 0.69, "learning_rate": 1.5607832918675998e-05, "loss": 2.7151, "step": 192700 }, { "epoch": 0.69, "learning_rate": 1.5589985400730322e-05, "loss": 2.711, "step": 192800 }, { "epoch": 0.69, "learning_rate": 1.5572137882784643e-05, "loss": 2.7089, "step": 192900 }, { "epoch": 0.69, "learning_rate": 1.5554290364838965e-05, "loss": 2.7197, "step": 193000 }, { "epoch": 0.69, "learning_rate": 1.5536442846893286e-05, "loss": 2.7035, "step": 193100 }, { "epoch": 0.69, "learning_rate": 1.5518595328947604e-05, "loss": 2.724, "step": 193200 }, { "epoch": 0.69, "learning_rate": 1.5500747811001925e-05, "loss": 2.721, "step": 193300 }, { "epoch": 0.69, "learning_rate": 1.5482900293056246e-05, "loss": 2.7175, "step": 193400 }, { "epoch": 0.69, "learning_rate": 1.5465052775110567e-05, "loss": 2.7098, "step": 193500 }, { "epoch": 0.69, "learning_rate": 1.544720525716489e-05, "loss": 2.7112, "step": 193600 }, { "epoch": 0.69, "learning_rate": 1.5429357739219206e-05, "loss": 2.7137, "step": 193700 }, { "epoch": 0.69, "learning_rate": 1.5411510221273528e-05, "loss": 2.7177, "step": 193800 }, { "epoch": 0.69, "learning_rate": 1.539366270332785e-05, "loss": 2.7109, "step": 193900 }, { "epoch": 0.69, "learning_rate": 1.537581518538217e-05, "loss": 2.7027, "step": 194000 }, { "epoch": 0.69, "learning_rate": 1.535796766743649e-05, "loss": 2.7139, "step": 194100 }, { "epoch": 0.69, "learning_rate": 1.5340120149490812e-05, "loss": 2.7126, "step": 194200 }, { "epoch": 0.69, "learning_rate": 1.532227263154513e-05, "loss": 2.7025, "step": 194300 }, { "epoch": 0.69, "learning_rate": 1.530442511359945e-05, "loss": 2.7097, "step": 194400 }, { "epoch": 0.69, "learning_rate": 1.5286577595653773e-05, "loss": 2.7117, "step": 194500 }, { "epoch": 0.69, "learning_rate": 1.5268730077708094e-05, "loss": 2.7228, "step": 194600 }, { "epoch": 0.69, "learning_rate": 1.5250882559762413e-05, "loss": 2.7108, "step": 194700 }, { "epoch": 0.7, "learning_rate": 1.5233035041816735e-05, "loss": 2.6974, "step": 194800 }, { "epoch": 0.7, "learning_rate": 1.5215187523871056e-05, "loss": 2.7131, "step": 194900 }, { "epoch": 0.7, "learning_rate": 1.5197340005925375e-05, "loss": 2.7097, "step": 195000 }, { "epoch": 0.7, "learning_rate": 1.5179492487979697e-05, "loss": 2.7097, "step": 195100 }, { "epoch": 0.7, "learning_rate": 1.5161644970034016e-05, "loss": 2.7182, "step": 195200 }, { "epoch": 0.7, "learning_rate": 1.5143797452088337e-05, "loss": 2.7113, "step": 195300 }, { "epoch": 0.7, "learning_rate": 1.512594993414266e-05, "loss": 2.7196, "step": 195400 }, { "epoch": 0.7, "learning_rate": 1.5108102416196981e-05, "loss": 2.7211, "step": 195500 }, { "epoch": 0.7, "learning_rate": 1.5090254898251303e-05, "loss": 2.7129, "step": 195600 }, { "epoch": 0.7, "learning_rate": 1.5072407380305622e-05, "loss": 2.7124, "step": 195700 }, { "epoch": 0.7, "learning_rate": 1.5054559862359943e-05, "loss": 2.7142, "step": 195800 }, { "epoch": 0.7, "learning_rate": 1.5036712344414265e-05, "loss": 2.71, "step": 195900 }, { "epoch": 0.7, "learning_rate": 1.5018864826468584e-05, "loss": 2.7202, "step": 196000 }, { "epoch": 0.7, "learning_rate": 1.5001017308522905e-05, "loss": 2.7147, "step": 196100 }, { "epoch": 0.7, "learning_rate": 1.4983169790577225e-05, "loss": 2.705, "step": 196200 }, { "epoch": 0.7, "learning_rate": 1.4965322272631546e-05, "loss": 2.7123, "step": 196300 }, { "epoch": 0.7, "learning_rate": 1.4947474754685867e-05, "loss": 2.7156, "step": 196400 }, { "epoch": 0.7, "learning_rate": 1.4929627236740187e-05, "loss": 2.716, "step": 196500 }, { "epoch": 0.7, "learning_rate": 1.4911779718794508e-05, "loss": 2.7234, "step": 196600 }, { "epoch": 0.7, "learning_rate": 1.489393220084883e-05, "loss": 2.7084, "step": 196700 }, { "epoch": 0.7, "learning_rate": 1.4876084682903149e-05, "loss": 2.7275, "step": 196800 }, { "epoch": 0.7, "learning_rate": 1.485823716495747e-05, "loss": 2.7241, "step": 196900 }, { "epoch": 0.7, "learning_rate": 1.4840389647011791e-05, "loss": 2.7109, "step": 197000 }, { "epoch": 0.7, "learning_rate": 1.482254212906611e-05, "loss": 2.7211, "step": 197100 }, { "epoch": 0.7, "learning_rate": 1.4804694611120432e-05, "loss": 2.717, "step": 197200 }, { "epoch": 0.7, "learning_rate": 1.4786847093174751e-05, "loss": 2.7164, "step": 197300 }, { "epoch": 0.7, "learning_rate": 1.4768999575229073e-05, "loss": 2.7071, "step": 197400 }, { "epoch": 0.7, "learning_rate": 1.4751152057283394e-05, "loss": 2.7143, "step": 197500 }, { "epoch": 0.71, "learning_rate": 1.4733304539337713e-05, "loss": 2.7101, "step": 197600 }, { "epoch": 0.71, "learning_rate": 1.4715457021392035e-05, "loss": 2.703, "step": 197700 }, { "epoch": 0.71, "learning_rate": 1.4697609503446356e-05, "loss": 2.7191, "step": 197800 }, { "epoch": 0.71, "learning_rate": 1.4679761985500679e-05, "loss": 2.6988, "step": 197900 }, { "epoch": 0.71, "learning_rate": 1.4661914467554998e-05, "loss": 2.7139, "step": 198000 }, { "epoch": 0.71, "learning_rate": 1.464406694960932e-05, "loss": 2.7126, "step": 198100 }, { "epoch": 0.71, "learning_rate": 1.462621943166364e-05, "loss": 2.7098, "step": 198200 }, { "epoch": 0.71, "learning_rate": 1.460837191371796e-05, "loss": 2.7169, "step": 198300 }, { "epoch": 0.71, "learning_rate": 1.4590524395772281e-05, "loss": 2.7155, "step": 198400 }, { "epoch": 0.71, "learning_rate": 1.4572676877826603e-05, "loss": 2.7074, "step": 198500 }, { "epoch": 0.71, "learning_rate": 1.4554829359880922e-05, "loss": 2.7125, "step": 198600 }, { "epoch": 0.71, "learning_rate": 1.4536981841935243e-05, "loss": 2.7075, "step": 198700 }, { "epoch": 0.71, "learning_rate": 1.4519134323989565e-05, "loss": 2.7093, "step": 198800 }, { "epoch": 0.71, "learning_rate": 1.4501286806043884e-05, "loss": 2.723, "step": 198900 }, { "epoch": 0.71, "learning_rate": 1.4483439288098205e-05, "loss": 2.7153, "step": 199000 }, { "epoch": 0.71, "learning_rate": 1.4465591770152525e-05, "loss": 2.722, "step": 199100 }, { "epoch": 0.71, "learning_rate": 1.4447744252206846e-05, "loss": 2.7141, "step": 199200 }, { "epoch": 0.71, "learning_rate": 1.4429896734261167e-05, "loss": 2.7116, "step": 199300 }, { "epoch": 0.71, "learning_rate": 1.4412049216315487e-05, "loss": 2.7091, "step": 199400 }, { "epoch": 0.71, "learning_rate": 1.4394201698369808e-05, "loss": 2.7091, "step": 199500 }, { "epoch": 0.71, "learning_rate": 1.437635418042413e-05, "loss": 2.7183, "step": 199600 }, { "epoch": 0.71, "learning_rate": 1.4358506662478449e-05, "loss": 2.7198, "step": 199700 }, { "epoch": 0.71, "learning_rate": 1.434065914453277e-05, "loss": 2.7075, "step": 199800 }, { "epoch": 0.71, "learning_rate": 1.4322811626587091e-05, "loss": 2.7077, "step": 199900 }, { "epoch": 0.71, "learning_rate": 1.430496410864141e-05, "loss": 2.7082, "step": 200000 }, { "epoch": 0.71, "learning_rate": 1.4287116590695732e-05, "loss": 2.7021, "step": 200100 }, { "epoch": 0.71, "learning_rate": 1.4269269072750051e-05, "loss": 2.7069, "step": 200200 }, { "epoch": 0.71, "learning_rate": 1.4251421554804373e-05, "loss": 2.7115, "step": 200300 }, { "epoch": 0.72, "learning_rate": 1.4233574036858694e-05, "loss": 2.721, "step": 200400 }, { "epoch": 0.72, "learning_rate": 1.4215726518913017e-05, "loss": 2.7126, "step": 200500 }, { "epoch": 0.72, "learning_rate": 1.4197879000967338e-05, "loss": 2.71, "step": 200600 }, { "epoch": 0.72, "learning_rate": 1.4180031483021658e-05, "loss": 2.7111, "step": 200700 }, { "epoch": 0.72, "learning_rate": 1.4162183965075979e-05, "loss": 2.7108, "step": 200800 }, { "epoch": 0.72, "learning_rate": 1.41443364471303e-05, "loss": 2.7228, "step": 200900 }, { "epoch": 0.72, "learning_rate": 1.412648892918462e-05, "loss": 2.7085, "step": 201000 }, { "epoch": 0.72, "learning_rate": 1.410864141123894e-05, "loss": 2.7095, "step": 201100 }, { "epoch": 0.72, "learning_rate": 1.409079389329326e-05, "loss": 2.7077, "step": 201200 }, { "epoch": 0.72, "learning_rate": 1.4072946375347581e-05, "loss": 2.7039, "step": 201300 }, { "epoch": 0.72, "learning_rate": 1.4055098857401903e-05, "loss": 2.7041, "step": 201400 }, { "epoch": 0.72, "learning_rate": 1.4037251339456222e-05, "loss": 2.7023, "step": 201500 }, { "epoch": 0.72, "learning_rate": 1.4019403821510543e-05, "loss": 2.7144, "step": 201600 }, { "epoch": 0.72, "learning_rate": 1.4001556303564865e-05, "loss": 2.7156, "step": 201700 }, { "epoch": 0.72, "learning_rate": 1.3983708785619184e-05, "loss": 2.7142, "step": 201800 }, { "epoch": 0.72, "learning_rate": 1.3965861267673505e-05, "loss": 2.7201, "step": 201900 }, { "epoch": 0.72, "learning_rate": 1.3948013749727827e-05, "loss": 2.6971, "step": 202000 }, { "epoch": 0.72, "learning_rate": 1.3930166231782146e-05, "loss": 2.7061, "step": 202100 }, { "epoch": 0.72, "learning_rate": 1.3912318713836467e-05, "loss": 2.6988, "step": 202200 }, { "epoch": 0.72, "learning_rate": 1.3894471195890787e-05, "loss": 2.6982, "step": 202300 }, { "epoch": 0.72, "learning_rate": 1.3876623677945108e-05, "loss": 2.7014, "step": 202400 }, { "epoch": 0.72, "learning_rate": 1.385877615999943e-05, "loss": 2.7116, "step": 202500 }, { "epoch": 0.72, "learning_rate": 1.3840928642053749e-05, "loss": 2.7166, "step": 202600 }, { "epoch": 0.72, "learning_rate": 1.382308112410807e-05, "loss": 2.7059, "step": 202700 }, { "epoch": 0.72, "learning_rate": 1.3805233606162391e-05, "loss": 2.7062, "step": 202800 }, { "epoch": 0.72, "learning_rate": 1.378738608821671e-05, "loss": 2.7098, "step": 202900 }, { "epoch": 0.72, "learning_rate": 1.3769538570271035e-05, "loss": 2.7037, "step": 203000 }, { "epoch": 0.72, "learning_rate": 1.3751691052325355e-05, "loss": 2.6999, "step": 203100 }, { "epoch": 0.73, "learning_rate": 1.3733843534379676e-05, "loss": 2.7086, "step": 203200 }, { "epoch": 0.73, "learning_rate": 1.3715996016433996e-05, "loss": 2.6995, "step": 203300 }, { "epoch": 0.73, "learning_rate": 1.3698148498488317e-05, "loss": 2.711, "step": 203400 }, { "epoch": 0.73, "learning_rate": 1.3680300980542638e-05, "loss": 2.7133, "step": 203500 }, { "epoch": 0.73, "learning_rate": 1.3662453462596957e-05, "loss": 2.7038, "step": 203600 }, { "epoch": 0.73, "learning_rate": 1.3644605944651279e-05, "loss": 2.7063, "step": 203700 }, { "epoch": 0.73, "learning_rate": 1.36267584267056e-05, "loss": 2.7157, "step": 203800 }, { "epoch": 0.73, "learning_rate": 1.360891090875992e-05, "loss": 2.7108, "step": 203900 }, { "epoch": 0.73, "learning_rate": 1.359106339081424e-05, "loss": 2.7191, "step": 204000 }, { "epoch": 0.73, "learning_rate": 1.357321587286856e-05, "loss": 2.7154, "step": 204100 }, { "epoch": 0.73, "learning_rate": 1.3555368354922881e-05, "loss": 2.7079, "step": 204200 }, { "epoch": 0.73, "learning_rate": 1.3537520836977203e-05, "loss": 2.7155, "step": 204300 }, { "epoch": 0.73, "learning_rate": 1.3519673319031522e-05, "loss": 2.7104, "step": 204400 }, { "epoch": 0.73, "learning_rate": 1.3501825801085843e-05, "loss": 2.7177, "step": 204500 }, { "epoch": 0.73, "learning_rate": 1.3483978283140165e-05, "loss": 2.7027, "step": 204600 }, { "epoch": 0.73, "learning_rate": 1.3466130765194484e-05, "loss": 2.7069, "step": 204700 }, { "epoch": 0.73, "learning_rate": 1.3448283247248805e-05, "loss": 2.7113, "step": 204800 }, { "epoch": 0.73, "learning_rate": 1.3430435729303127e-05, "loss": 2.7089, "step": 204900 }, { "epoch": 0.73, "learning_rate": 1.3412588211357446e-05, "loss": 2.7084, "step": 205000 }, { "epoch": 0.73, "learning_rate": 1.3394740693411767e-05, "loss": 2.7063, "step": 205100 }, { "epoch": 0.73, "learning_rate": 1.3376893175466087e-05, "loss": 2.7047, "step": 205200 }, { "epoch": 0.73, "learning_rate": 1.3359045657520408e-05, "loss": 2.7045, "step": 205300 }, { "epoch": 0.73, "learning_rate": 1.334119813957473e-05, "loss": 2.7003, "step": 205400 }, { "epoch": 0.73, "learning_rate": 1.3323350621629049e-05, "loss": 2.7127, "step": 205500 }, { "epoch": 0.73, "learning_rate": 1.3305503103683373e-05, "loss": 2.706, "step": 205600 }, { "epoch": 0.73, "learning_rate": 1.3287655585737693e-05, "loss": 2.7087, "step": 205700 }, { "epoch": 0.73, "learning_rate": 1.3269808067792014e-05, "loss": 2.7042, "step": 205800 }, { "epoch": 0.73, "learning_rate": 1.3251960549846335e-05, "loss": 2.7102, "step": 205900 }, { "epoch": 0.74, "learning_rate": 1.3234113031900655e-05, "loss": 2.7052, "step": 206000 }, { "epoch": 0.74, "learning_rate": 1.3216265513954976e-05, "loss": 2.7132, "step": 206100 }, { "epoch": 0.74, "learning_rate": 1.3198417996009296e-05, "loss": 2.7066, "step": 206200 }, { "epoch": 0.74, "learning_rate": 1.3180570478063617e-05, "loss": 2.7044, "step": 206300 }, { "epoch": 0.74, "learning_rate": 1.3162722960117938e-05, "loss": 2.7034, "step": 206400 }, { "epoch": 0.74, "learning_rate": 1.3144875442172257e-05, "loss": 2.7098, "step": 206500 }, { "epoch": 0.74, "learning_rate": 1.3127027924226579e-05, "loss": 2.7157, "step": 206600 }, { "epoch": 0.74, "learning_rate": 1.31091804062809e-05, "loss": 2.7046, "step": 206700 }, { "epoch": 0.74, "learning_rate": 1.309133288833522e-05, "loss": 2.7127, "step": 206800 }, { "epoch": 0.74, "learning_rate": 1.307348537038954e-05, "loss": 2.7212, "step": 206900 }, { "epoch": 0.74, "learning_rate": 1.3055637852443862e-05, "loss": 2.7121, "step": 207000 }, { "epoch": 0.74, "learning_rate": 1.3037790334498181e-05, "loss": 2.7047, "step": 207100 }, { "epoch": 0.74, "learning_rate": 1.3019942816552503e-05, "loss": 2.7158, "step": 207200 }, { "epoch": 0.74, "learning_rate": 1.3002095298606822e-05, "loss": 2.7122, "step": 207300 }, { "epoch": 0.74, "learning_rate": 1.2984247780661143e-05, "loss": 2.7048, "step": 207400 }, { "epoch": 0.74, "learning_rate": 1.2966400262715465e-05, "loss": 2.6977, "step": 207500 }, { "epoch": 0.74, "learning_rate": 1.2948552744769784e-05, "loss": 2.7032, "step": 207600 }, { "epoch": 0.74, "learning_rate": 1.2930705226824105e-05, "loss": 2.7065, "step": 207700 }, { "epoch": 0.74, "learning_rate": 1.2912857708878426e-05, "loss": 2.7104, "step": 207800 }, { "epoch": 0.74, "learning_rate": 1.2895010190932746e-05, "loss": 2.7086, "step": 207900 }, { "epoch": 0.74, "learning_rate": 1.2877162672987067e-05, "loss": 2.7076, "step": 208000 }, { "epoch": 0.74, "learning_rate": 1.285931515504139e-05, "loss": 2.7008, "step": 208100 }, { "epoch": 0.74, "learning_rate": 1.2841467637095711e-05, "loss": 2.712, "step": 208200 }, { "epoch": 0.74, "learning_rate": 1.2823620119150031e-05, "loss": 2.7053, "step": 208300 }, { "epoch": 0.74, "learning_rate": 1.2805772601204352e-05, "loss": 2.7051, "step": 208400 }, { "epoch": 0.74, "learning_rate": 1.2787925083258673e-05, "loss": 2.7001, "step": 208500 }, { "epoch": 0.74, "learning_rate": 1.2770077565312993e-05, "loss": 2.697, "step": 208600 }, { "epoch": 0.74, "learning_rate": 1.2752230047367314e-05, "loss": 2.7099, "step": 208700 }, { "epoch": 0.75, "learning_rate": 1.2734382529421635e-05, "loss": 2.7105, "step": 208800 }, { "epoch": 0.75, "learning_rate": 1.2716535011475955e-05, "loss": 2.7111, "step": 208900 }, { "epoch": 0.75, "learning_rate": 1.2698687493530276e-05, "loss": 2.7154, "step": 209000 }, { "epoch": 0.75, "learning_rate": 1.2680839975584596e-05, "loss": 2.7073, "step": 209100 }, { "epoch": 0.75, "learning_rate": 1.2662992457638917e-05, "loss": 2.7087, "step": 209200 }, { "epoch": 0.75, "learning_rate": 1.2645144939693238e-05, "loss": 2.7062, "step": 209300 }, { "epoch": 0.75, "learning_rate": 1.2627297421747557e-05, "loss": 2.7118, "step": 209400 }, { "epoch": 0.75, "learning_rate": 1.2609449903801879e-05, "loss": 2.7088, "step": 209500 }, { "epoch": 0.75, "learning_rate": 1.25916023858562e-05, "loss": 2.7063, "step": 209600 }, { "epoch": 0.75, "learning_rate": 1.257375486791052e-05, "loss": 2.6977, "step": 209700 }, { "epoch": 0.75, "learning_rate": 1.255590734996484e-05, "loss": 2.704, "step": 209800 }, { "epoch": 0.75, "learning_rate": 1.2538059832019162e-05, "loss": 2.7019, "step": 209900 }, { "epoch": 0.75, "learning_rate": 1.2520212314073481e-05, "loss": 2.7064, "step": 210000 }, { "epoch": 0.75, "learning_rate": 1.2502364796127803e-05, "loss": 2.7135, "step": 210100 }, { "epoch": 0.75, "learning_rate": 1.2484517278182124e-05, "loss": 2.699, "step": 210200 }, { "epoch": 0.75, "learning_rate": 1.2466669760236445e-05, "loss": 2.6967, "step": 210300 }, { "epoch": 0.75, "learning_rate": 1.2448822242290766e-05, "loss": 2.6995, "step": 210400 }, { "epoch": 0.75, "learning_rate": 1.2430974724345086e-05, "loss": 2.7137, "step": 210500 }, { "epoch": 0.75, "learning_rate": 1.2413127206399407e-05, "loss": 2.7117, "step": 210600 }, { "epoch": 0.75, "learning_rate": 1.2395279688453726e-05, "loss": 2.7145, "step": 210700 }, { "epoch": 0.75, "learning_rate": 1.2377432170508048e-05, "loss": 2.7083, "step": 210800 }, { "epoch": 0.75, "learning_rate": 1.2359584652562369e-05, "loss": 2.7017, "step": 210900 }, { "epoch": 0.75, "learning_rate": 1.2341737134616688e-05, "loss": 2.7052, "step": 211000 }, { "epoch": 0.75, "learning_rate": 1.232388961667101e-05, "loss": 2.7031, "step": 211100 }, { "epoch": 0.75, "learning_rate": 1.2306042098725331e-05, "loss": 2.6986, "step": 211200 }, { "epoch": 0.75, "learning_rate": 1.2288194580779652e-05, "loss": 2.7134, "step": 211300 }, { "epoch": 0.75, "learning_rate": 1.2270347062833973e-05, "loss": 2.7121, "step": 211400 }, { "epoch": 0.75, "learning_rate": 1.2252499544888293e-05, "loss": 2.6999, "step": 211500 }, { "epoch": 0.76, "learning_rate": 1.2234652026942614e-05, "loss": 2.7063, "step": 211600 }, { "epoch": 0.76, "learning_rate": 1.2216804508996935e-05, "loss": 2.7068, "step": 211700 }, { "epoch": 0.76, "learning_rate": 1.2198956991051255e-05, "loss": 2.711, "step": 211800 }, { "epoch": 0.76, "learning_rate": 1.2181109473105576e-05, "loss": 2.7179, "step": 211900 }, { "epoch": 0.76, "learning_rate": 1.2163261955159897e-05, "loss": 2.7077, "step": 212000 }, { "epoch": 0.76, "learning_rate": 1.2145414437214217e-05, "loss": 2.703, "step": 212100 }, { "epoch": 0.76, "learning_rate": 1.2127566919268538e-05, "loss": 2.709, "step": 212200 }, { "epoch": 0.76, "learning_rate": 1.2109719401322857e-05, "loss": 2.703, "step": 212300 }, { "epoch": 0.76, "learning_rate": 1.2091871883377179e-05, "loss": 2.7035, "step": 212400 }, { "epoch": 0.76, "learning_rate": 1.20740243654315e-05, "loss": 2.6925, "step": 212500 }, { "epoch": 0.76, "learning_rate": 1.2056176847485821e-05, "loss": 2.7145, "step": 212600 }, { "epoch": 0.76, "learning_rate": 1.2038329329540142e-05, "loss": 2.7073, "step": 212700 }, { "epoch": 0.76, "learning_rate": 1.2020481811594462e-05, "loss": 2.7192, "step": 212800 }, { "epoch": 0.76, "learning_rate": 1.2002634293648783e-05, "loss": 2.7103, "step": 212900 }, { "epoch": 0.76, "learning_rate": 1.1984786775703104e-05, "loss": 2.7006, "step": 213000 }, { "epoch": 0.76, "learning_rate": 1.1966939257757424e-05, "loss": 2.7086, "step": 213100 }, { "epoch": 0.76, "learning_rate": 1.1949091739811745e-05, "loss": 2.7046, "step": 213200 }, { "epoch": 0.76, "learning_rate": 1.1931244221866066e-05, "loss": 2.7126, "step": 213300 }, { "epoch": 0.76, "learning_rate": 1.1913396703920386e-05, "loss": 2.6996, "step": 213400 }, { "epoch": 0.76, "learning_rate": 1.1895549185974707e-05, "loss": 2.7086, "step": 213500 }, { "epoch": 0.76, "learning_rate": 1.1877701668029028e-05, "loss": 2.7119, "step": 213600 }, { "epoch": 0.76, "learning_rate": 1.1859854150083348e-05, "loss": 2.6988, "step": 213700 }, { "epoch": 0.76, "learning_rate": 1.1842006632137669e-05, "loss": 2.7048, "step": 213800 }, { "epoch": 0.76, "learning_rate": 1.182415911419199e-05, "loss": 2.7112, "step": 213900 }, { "epoch": 0.76, "learning_rate": 1.1806311596246311e-05, "loss": 2.7044, "step": 214000 }, { "epoch": 0.76, "learning_rate": 1.1788464078300633e-05, "loss": 2.7009, "step": 214100 }, { "epoch": 0.76, "learning_rate": 1.1770616560354952e-05, "loss": 2.7011, "step": 214200 }, { "epoch": 0.76, "learning_rate": 1.1752769042409273e-05, "loss": 2.7038, "step": 214300 }, { "epoch": 0.77, "learning_rate": 1.1734921524463593e-05, "loss": 2.6994, "step": 214400 }, { "epoch": 0.77, "learning_rate": 1.1717074006517914e-05, "loss": 2.7019, "step": 214500 }, { "epoch": 0.77, "learning_rate": 1.1699226488572235e-05, "loss": 2.7028, "step": 214600 }, { "epoch": 0.77, "learning_rate": 1.1681378970626555e-05, "loss": 2.7122, "step": 214700 }, { "epoch": 0.77, "learning_rate": 1.1663531452680876e-05, "loss": 2.7032, "step": 214800 }, { "epoch": 0.77, "learning_rate": 1.1645683934735197e-05, "loss": 2.6944, "step": 214900 }, { "epoch": 0.77, "learning_rate": 1.1627836416789517e-05, "loss": 2.7105, "step": 215000 }, { "epoch": 0.77, "learning_rate": 1.1609988898843838e-05, "loss": 2.6958, "step": 215100 }, { "epoch": 0.77, "learning_rate": 1.1592141380898159e-05, "loss": 2.701, "step": 215200 }, { "epoch": 0.77, "learning_rate": 1.157429386295248e-05, "loss": 2.7047, "step": 215300 }, { "epoch": 0.77, "learning_rate": 1.1556446345006802e-05, "loss": 2.7097, "step": 215400 }, { "epoch": 0.77, "learning_rate": 1.1538598827061121e-05, "loss": 2.6957, "step": 215500 }, { "epoch": 0.77, "learning_rate": 1.1520751309115442e-05, "loss": 2.7055, "step": 215600 }, { "epoch": 0.77, "learning_rate": 1.1502903791169762e-05, "loss": 2.7012, "step": 215700 }, { "epoch": 0.77, "learning_rate": 1.1485056273224083e-05, "loss": 2.7047, "step": 215800 }, { "epoch": 0.77, "learning_rate": 1.1467208755278404e-05, "loss": 2.6991, "step": 215900 }, { "epoch": 0.77, "learning_rate": 1.1449361237332724e-05, "loss": 2.7148, "step": 216000 }, { "epoch": 0.77, "learning_rate": 1.1431513719387045e-05, "loss": 2.6978, "step": 216100 }, { "epoch": 0.77, "learning_rate": 1.1413666201441366e-05, "loss": 2.6998, "step": 216200 }, { "epoch": 0.77, "learning_rate": 1.1395818683495686e-05, "loss": 2.7136, "step": 216300 }, { "epoch": 0.77, "learning_rate": 1.1377971165550009e-05, "loss": 2.6882, "step": 216400 }, { "epoch": 0.77, "learning_rate": 1.1360123647604328e-05, "loss": 2.7058, "step": 216500 }, { "epoch": 0.77, "learning_rate": 1.134227612965865e-05, "loss": 2.7071, "step": 216600 }, { "epoch": 0.77, "learning_rate": 1.132442861171297e-05, "loss": 2.6967, "step": 216700 }, { "epoch": 0.77, "learning_rate": 1.130658109376729e-05, "loss": 2.6981, "step": 216800 }, { "epoch": 0.77, "learning_rate": 1.1288733575821611e-05, "loss": 2.6987, "step": 216900 }, { "epoch": 0.77, "learning_rate": 1.1270886057875933e-05, "loss": 2.7027, "step": 217000 }, { "epoch": 0.77, "learning_rate": 1.1253038539930252e-05, "loss": 2.7022, "step": 217100 }, { "epoch": 0.78, "learning_rate": 1.1235191021984573e-05, "loss": 2.7012, "step": 217200 }, { "epoch": 0.78, "learning_rate": 1.1217343504038893e-05, "loss": 2.6959, "step": 217300 }, { "epoch": 0.78, "learning_rate": 1.1199495986093214e-05, "loss": 2.7083, "step": 217400 }, { "epoch": 0.78, "learning_rate": 1.1181648468147535e-05, "loss": 2.705, "step": 217500 }, { "epoch": 0.78, "learning_rate": 1.1163800950201855e-05, "loss": 2.7049, "step": 217600 }, { "epoch": 0.78, "learning_rate": 1.1145953432256178e-05, "loss": 2.7069, "step": 217700 }, { "epoch": 0.78, "learning_rate": 1.1128105914310497e-05, "loss": 2.7007, "step": 217800 }, { "epoch": 0.78, "learning_rate": 1.1110258396364818e-05, "loss": 2.7038, "step": 217900 }, { "epoch": 0.78, "learning_rate": 1.109241087841914e-05, "loss": 2.7036, "step": 218000 }, { "epoch": 0.78, "learning_rate": 1.1074563360473459e-05, "loss": 2.7045, "step": 218100 }, { "epoch": 0.78, "learning_rate": 1.105671584252778e-05, "loss": 2.7068, "step": 218200 }, { "epoch": 0.78, "learning_rate": 1.1038868324582102e-05, "loss": 2.6995, "step": 218300 }, { "epoch": 0.78, "learning_rate": 1.1021020806636421e-05, "loss": 2.7003, "step": 218400 }, { "epoch": 0.78, "learning_rate": 1.1003173288690742e-05, "loss": 2.7071, "step": 218500 }, { "epoch": 0.78, "learning_rate": 1.0985325770745063e-05, "loss": 2.7033, "step": 218600 }, { "epoch": 0.78, "learning_rate": 1.0967478252799383e-05, "loss": 2.7064, "step": 218700 }, { "epoch": 0.78, "learning_rate": 1.0949630734853704e-05, "loss": 2.6959, "step": 218800 }, { "epoch": 0.78, "learning_rate": 1.0931783216908024e-05, "loss": 2.6931, "step": 218900 }, { "epoch": 0.78, "learning_rate": 1.0913935698962347e-05, "loss": 2.7059, "step": 219000 }, { "epoch": 0.78, "learning_rate": 1.0896088181016668e-05, "loss": 2.6962, "step": 219100 }, { "epoch": 0.78, "learning_rate": 1.0878240663070987e-05, "loss": 2.7064, "step": 219200 }, { "epoch": 0.78, "learning_rate": 1.0860393145125309e-05, "loss": 2.7002, "step": 219300 }, { "epoch": 0.78, "learning_rate": 1.0842545627179628e-05, "loss": 2.7041, "step": 219400 }, { "epoch": 0.78, "learning_rate": 1.082469810923395e-05, "loss": 2.7046, "step": 219500 }, { "epoch": 0.78, "learning_rate": 1.080685059128827e-05, "loss": 2.6958, "step": 219600 }, { "epoch": 0.78, "learning_rate": 1.078900307334259e-05, "loss": 2.7129, "step": 219700 }, { "epoch": 0.78, "learning_rate": 1.0771155555396911e-05, "loss": 2.7022, "step": 219800 }, { "epoch": 0.78, "learning_rate": 1.0753308037451233e-05, "loss": 2.7013, "step": 219900 }, { "epoch": 0.79, "learning_rate": 1.0735460519505552e-05, "loss": 2.6996, "step": 220000 }, { "epoch": 0.79, "learning_rate": 1.0717613001559873e-05, "loss": 2.7058, "step": 220100 }, { "epoch": 0.79, "learning_rate": 1.0699765483614194e-05, "loss": 2.7006, "step": 220200 }, { "epoch": 0.79, "learning_rate": 1.0681917965668516e-05, "loss": 2.7022, "step": 220300 }, { "epoch": 0.79, "learning_rate": 1.0664070447722837e-05, "loss": 2.7062, "step": 220400 }, { "epoch": 0.79, "learning_rate": 1.0646222929777156e-05, "loss": 2.6988, "step": 220500 }, { "epoch": 0.79, "learning_rate": 1.0628375411831478e-05, "loss": 2.7036, "step": 220600 }, { "epoch": 0.79, "learning_rate": 1.0610527893885797e-05, "loss": 2.6986, "step": 220700 }, { "epoch": 0.79, "learning_rate": 1.0592680375940118e-05, "loss": 2.7137, "step": 220800 }, { "epoch": 0.79, "learning_rate": 1.057483285799444e-05, "loss": 2.7025, "step": 220900 }, { "epoch": 0.79, "learning_rate": 1.0556985340048759e-05, "loss": 2.7126, "step": 221000 }, { "epoch": 0.79, "learning_rate": 1.053913782210308e-05, "loss": 2.7058, "step": 221100 }, { "epoch": 0.79, "learning_rate": 1.0521290304157402e-05, "loss": 2.7087, "step": 221200 }, { "epoch": 0.79, "learning_rate": 1.0503442786211721e-05, "loss": 2.6998, "step": 221300 }, { "epoch": 0.79, "learning_rate": 1.0485595268266042e-05, "loss": 2.7031, "step": 221400 }, { "epoch": 0.79, "learning_rate": 1.0467747750320363e-05, "loss": 2.7069, "step": 221500 }, { "epoch": 0.79, "learning_rate": 1.0449900232374685e-05, "loss": 2.708, "step": 221600 }, { "epoch": 0.79, "learning_rate": 1.0432052714429006e-05, "loss": 2.7032, "step": 221700 }, { "epoch": 0.79, "learning_rate": 1.0414205196483325e-05, "loss": 2.7039, "step": 221800 }, { "epoch": 0.79, "learning_rate": 1.0396357678537647e-05, "loss": 2.6945, "step": 221900 }, { "epoch": 0.79, "learning_rate": 1.0378510160591968e-05, "loss": 2.6955, "step": 222000 }, { "epoch": 0.79, "learning_rate": 1.0360662642646287e-05, "loss": 2.6966, "step": 222100 }, { "epoch": 0.79, "learning_rate": 1.0342815124700609e-05, "loss": 2.6987, "step": 222200 }, { "epoch": 0.79, "learning_rate": 1.0324967606754928e-05, "loss": 2.6986, "step": 222300 }, { "epoch": 0.79, "learning_rate": 1.030712008880925e-05, "loss": 2.7068, "step": 222400 }, { "epoch": 0.79, "learning_rate": 1.028927257086357e-05, "loss": 2.7002, "step": 222500 }, { "epoch": 0.79, "learning_rate": 1.027142505291789e-05, "loss": 2.6913, "step": 222600 }, { "epoch": 0.79, "learning_rate": 1.0253577534972211e-05, "loss": 2.6961, "step": 222700 }, { "epoch": 0.8, "learning_rate": 1.0235730017026532e-05, "loss": 2.6971, "step": 222800 }, { "epoch": 0.8, "learning_rate": 1.0217882499080854e-05, "loss": 2.7025, "step": 222900 }, { "epoch": 0.8, "learning_rate": 1.0200034981135175e-05, "loss": 2.7009, "step": 223000 }, { "epoch": 0.8, "learning_rate": 1.0182187463189494e-05, "loss": 2.698, "step": 223100 }, { "epoch": 0.8, "learning_rate": 1.0164339945243816e-05, "loss": 2.6932, "step": 223200 }, { "epoch": 0.8, "learning_rate": 1.0146492427298137e-05, "loss": 2.6892, "step": 223300 }, { "epoch": 0.8, "learning_rate": 1.0128644909352456e-05, "loss": 2.7013, "step": 223400 }, { "epoch": 0.8, "learning_rate": 1.0110797391406778e-05, "loss": 2.6979, "step": 223500 }, { "epoch": 0.8, "learning_rate": 1.0092949873461099e-05, "loss": 2.6968, "step": 223600 }, { "epoch": 0.8, "learning_rate": 1.0075102355515418e-05, "loss": 2.7025, "step": 223700 }, { "epoch": 0.8, "learning_rate": 1.005725483756974e-05, "loss": 2.7071, "step": 223800 }, { "epoch": 0.8, "learning_rate": 1.0039407319624059e-05, "loss": 2.7038, "step": 223900 }, { "epoch": 0.8, "learning_rate": 1.002155980167838e-05, "loss": 2.6933, "step": 224000 }, { "epoch": 0.8, "learning_rate": 1.0003712283732703e-05, "loss": 2.6946, "step": 224100 }, { "epoch": 0.8, "learning_rate": 9.985864765787023e-06, "loss": 2.7156, "step": 224200 }, { "epoch": 0.8, "learning_rate": 9.968017247841344e-06, "loss": 2.7014, "step": 224300 }, { "epoch": 0.8, "learning_rate": 9.950169729895663e-06, "loss": 2.7004, "step": 224400 }, { "epoch": 0.8, "learning_rate": 9.932322211949985e-06, "loss": 2.7012, "step": 224500 }, { "epoch": 0.8, "learning_rate": 9.914474694004306e-06, "loss": 2.7054, "step": 224600 }, { "epoch": 0.8, "learning_rate": 9.896627176058625e-06, "loss": 2.6914, "step": 224700 }, { "epoch": 0.8, "learning_rate": 9.878779658112947e-06, "loss": 2.7055, "step": 224800 }, { "epoch": 0.8, "learning_rate": 9.860932140167268e-06, "loss": 2.6997, "step": 224900 }, { "epoch": 0.8, "learning_rate": 9.843084622221587e-06, "loss": 2.7085, "step": 225000 }, { "epoch": 0.8, "learning_rate": 9.825237104275909e-06, "loss": 2.6912, "step": 225100 }, { "epoch": 0.8, "learning_rate": 9.80738958633023e-06, "loss": 2.7002, "step": 225200 }, { "epoch": 0.8, "learning_rate": 9.78954206838455e-06, "loss": 2.6993, "step": 225300 }, { "epoch": 0.8, "learning_rate": 9.771694550438872e-06, "loss": 2.6978, "step": 225400 }, { "epoch": 0.8, "learning_rate": 9.753847032493192e-06, "loss": 2.7073, "step": 225500 }, { "epoch": 0.81, "learning_rate": 9.735999514547513e-06, "loss": 2.6984, "step": 225600 }, { "epoch": 0.81, "learning_rate": 9.718151996601834e-06, "loss": 2.6929, "step": 225700 }, { "epoch": 0.81, "learning_rate": 9.700304478656154e-06, "loss": 2.6988, "step": 225800 }, { "epoch": 0.81, "learning_rate": 9.682456960710475e-06, "loss": 2.6957, "step": 225900 }, { "epoch": 0.81, "learning_rate": 9.664609442764794e-06, "loss": 2.7019, "step": 226000 }, { "epoch": 0.81, "learning_rate": 9.646761924819116e-06, "loss": 2.7073, "step": 226100 }, { "epoch": 0.81, "learning_rate": 9.628914406873437e-06, "loss": 2.7022, "step": 226200 }, { "epoch": 0.81, "learning_rate": 9.611066888927756e-06, "loss": 2.7146, "step": 226300 }, { "epoch": 0.81, "learning_rate": 9.593219370982078e-06, "loss": 2.6932, "step": 226400 }, { "epoch": 0.81, "learning_rate": 9.575371853036399e-06, "loss": 2.6939, "step": 226500 }, { "epoch": 0.81, "learning_rate": 9.557524335090718e-06, "loss": 2.6955, "step": 226600 }, { "epoch": 0.81, "learning_rate": 9.539676817145041e-06, "loss": 2.6939, "step": 226700 }, { "epoch": 0.81, "learning_rate": 9.52182929919936e-06, "loss": 2.6939, "step": 226800 }, { "epoch": 0.81, "learning_rate": 9.503981781253682e-06, "loss": 2.7074, "step": 226900 }, { "epoch": 0.81, "learning_rate": 9.486134263308003e-06, "loss": 2.7026, "step": 227000 }, { "epoch": 0.81, "learning_rate": 9.468286745362323e-06, "loss": 2.7055, "step": 227100 }, { "epoch": 0.81, "learning_rate": 9.450439227416644e-06, "loss": 2.6944, "step": 227200 }, { "epoch": 0.81, "learning_rate": 9.432591709470963e-06, "loss": 2.6982, "step": 227300 }, { "epoch": 0.81, "learning_rate": 9.414744191525285e-06, "loss": 2.6879, "step": 227400 }, { "epoch": 0.81, "learning_rate": 9.396896673579606e-06, "loss": 2.7001, "step": 227500 }, { "epoch": 0.81, "learning_rate": 9.379049155633925e-06, "loss": 2.704, "step": 227600 }, { "epoch": 0.81, "learning_rate": 9.361201637688247e-06, "loss": 2.6901, "step": 227700 }, { "epoch": 0.81, "learning_rate": 9.343354119742568e-06, "loss": 2.7065, "step": 227800 }, { "epoch": 0.81, "learning_rate": 9.325506601796889e-06, "loss": 2.696, "step": 227900 }, { "epoch": 0.81, "learning_rate": 9.30765908385121e-06, "loss": 2.7024, "step": 228000 }, { "epoch": 0.81, "learning_rate": 9.28981156590553e-06, "loss": 2.7057, "step": 228100 }, { "epoch": 0.81, "learning_rate": 9.271964047959851e-06, "loss": 2.6975, "step": 228200 }, { "epoch": 0.81, "learning_rate": 9.254116530014172e-06, "loss": 2.7037, "step": 228300 }, { "epoch": 0.82, "learning_rate": 9.236269012068492e-06, "loss": 2.6997, "step": 228400 }, { "epoch": 0.82, "learning_rate": 9.218421494122813e-06, "loss": 2.6944, "step": 228500 }, { "epoch": 0.82, "learning_rate": 9.200573976177134e-06, "loss": 2.6946, "step": 228600 }, { "epoch": 0.82, "learning_rate": 9.182726458231454e-06, "loss": 2.7034, "step": 228700 }, { "epoch": 0.82, "learning_rate": 9.164878940285775e-06, "loss": 2.7105, "step": 228800 }, { "epoch": 0.82, "learning_rate": 9.147031422340094e-06, "loss": 2.6954, "step": 228900 }, { "epoch": 0.82, "learning_rate": 9.129183904394416e-06, "loss": 2.6991, "step": 229000 }, { "epoch": 0.82, "learning_rate": 9.111336386448737e-06, "loss": 2.7004, "step": 229100 }, { "epoch": 0.82, "learning_rate": 9.093488868503058e-06, "loss": 2.6985, "step": 229200 }, { "epoch": 0.82, "learning_rate": 9.07564135055738e-06, "loss": 2.7126, "step": 229300 }, { "epoch": 0.82, "learning_rate": 9.057793832611699e-06, "loss": 2.7081, "step": 229400 }, { "epoch": 0.82, "learning_rate": 9.03994631466602e-06, "loss": 2.6968, "step": 229500 }, { "epoch": 0.82, "learning_rate": 9.022098796720341e-06, "loss": 2.6953, "step": 229600 }, { "epoch": 0.82, "learning_rate": 9.00425127877466e-06, "loss": 2.7023, "step": 229700 }, { "epoch": 0.82, "learning_rate": 8.986403760828982e-06, "loss": 2.6988, "step": 229800 }, { "epoch": 0.82, "learning_rate": 8.968556242883303e-06, "loss": 2.7011, "step": 229900 }, { "epoch": 0.82, "learning_rate": 8.950708724937623e-06, "loss": 2.7062, "step": 230000 }, { "epoch": 0.82, "learning_rate": 8.932861206991944e-06, "loss": 2.6967, "step": 230100 }, { "epoch": 0.82, "learning_rate": 8.915013689046265e-06, "loss": 2.6912, "step": 230200 }, { "epoch": 0.82, "learning_rate": 8.897166171100585e-06, "loss": 2.7043, "step": 230300 }, { "epoch": 0.82, "learning_rate": 8.879318653154906e-06, "loss": 2.6933, "step": 230400 }, { "epoch": 0.82, "learning_rate": 8.861471135209227e-06, "loss": 2.7098, "step": 230500 }, { "epoch": 0.82, "learning_rate": 8.843623617263548e-06, "loss": 2.6921, "step": 230600 }, { "epoch": 0.82, "learning_rate": 8.82577609931787e-06, "loss": 2.6961, "step": 230700 }, { "epoch": 0.82, "learning_rate": 8.807928581372189e-06, "loss": 2.6977, "step": 230800 }, { "epoch": 0.82, "learning_rate": 8.79008106342651e-06, "loss": 2.7012, "step": 230900 }, { "epoch": 0.82, "learning_rate": 8.77223354548083e-06, "loss": 2.7062, "step": 231000 }, { "epoch": 0.82, "learning_rate": 8.754386027535151e-06, "loss": 2.7066, "step": 231100 }, { "epoch": 0.83, "learning_rate": 8.736538509589472e-06, "loss": 2.6948, "step": 231200 }, { "epoch": 0.83, "learning_rate": 8.718690991643792e-06, "loss": 2.7013, "step": 231300 }, { "epoch": 0.83, "learning_rate": 8.700843473698113e-06, "loss": 2.7004, "step": 231400 }, { "epoch": 0.83, "learning_rate": 8.682995955752434e-06, "loss": 2.6943, "step": 231500 }, { "epoch": 0.83, "learning_rate": 8.665148437806754e-06, "loss": 2.6959, "step": 231600 }, { "epoch": 0.83, "learning_rate": 8.647300919861075e-06, "loss": 2.6993, "step": 231700 }, { "epoch": 0.83, "learning_rate": 8.629453401915396e-06, "loss": 2.6986, "step": 231800 }, { "epoch": 0.83, "learning_rate": 8.611605883969717e-06, "loss": 2.7002, "step": 231900 }, { "epoch": 0.83, "learning_rate": 8.593758366024039e-06, "loss": 2.6978, "step": 232000 }, { "epoch": 0.83, "learning_rate": 8.575910848078358e-06, "loss": 2.6906, "step": 232100 }, { "epoch": 0.83, "learning_rate": 8.55806333013268e-06, "loss": 2.6979, "step": 232200 }, { "epoch": 0.83, "learning_rate": 8.540215812186999e-06, "loss": 2.7052, "step": 232300 }, { "epoch": 0.83, "learning_rate": 8.52236829424132e-06, "loss": 2.6965, "step": 232400 }, { "epoch": 0.83, "learning_rate": 8.504520776295641e-06, "loss": 2.6953, "step": 232500 }, { "epoch": 0.83, "learning_rate": 8.48667325834996e-06, "loss": 2.6928, "step": 232600 }, { "epoch": 0.83, "learning_rate": 8.468825740404282e-06, "loss": 2.7064, "step": 232700 }, { "epoch": 0.83, "learning_rate": 8.450978222458603e-06, "loss": 2.6998, "step": 232800 }, { "epoch": 0.83, "learning_rate": 8.433130704512923e-06, "loss": 2.6953, "step": 232900 }, { "epoch": 0.83, "learning_rate": 8.415283186567246e-06, "loss": 2.691, "step": 233000 }, { "epoch": 0.83, "learning_rate": 8.397435668621565e-06, "loss": 2.6897, "step": 233100 }, { "epoch": 0.83, "learning_rate": 8.379588150675886e-06, "loss": 2.7072, "step": 233200 }, { "epoch": 0.83, "learning_rate": 8.361740632730208e-06, "loss": 2.6961, "step": 233300 }, { "epoch": 0.83, "learning_rate": 8.343893114784527e-06, "loss": 2.7006, "step": 233400 }, { "epoch": 0.83, "learning_rate": 8.326045596838848e-06, "loss": 2.695, "step": 233500 }, { "epoch": 0.83, "learning_rate": 8.30819807889317e-06, "loss": 2.6959, "step": 233600 }, { "epoch": 0.83, "learning_rate": 8.290350560947489e-06, "loss": 2.685, "step": 233700 }, { "epoch": 0.83, "learning_rate": 8.27250304300181e-06, "loss": 2.701, "step": 233800 }, { "epoch": 0.83, "learning_rate": 8.25465552505613e-06, "loss": 2.7057, "step": 233900 }, { "epoch": 0.84, "learning_rate": 8.236808007110451e-06, "loss": 2.6963, "step": 234000 }, { "epoch": 0.84, "learning_rate": 8.218960489164772e-06, "loss": 2.6934, "step": 234100 }, { "epoch": 0.84, "learning_rate": 8.201112971219092e-06, "loss": 2.7008, "step": 234200 }, { "epoch": 0.84, "learning_rate": 8.183265453273415e-06, "loss": 2.7019, "step": 234300 }, { "epoch": 0.84, "learning_rate": 8.165417935327734e-06, "loss": 2.6915, "step": 234400 }, { "epoch": 0.84, "learning_rate": 8.147570417382055e-06, "loss": 2.702, "step": 234500 }, { "epoch": 0.84, "learning_rate": 8.129722899436377e-06, "loss": 2.6931, "step": 234600 }, { "epoch": 0.84, "learning_rate": 8.111875381490696e-06, "loss": 2.6943, "step": 234700 }, { "epoch": 0.84, "learning_rate": 8.094027863545017e-06, "loss": 2.6886, "step": 234800 }, { "epoch": 0.84, "learning_rate": 8.076180345599339e-06, "loss": 2.6973, "step": 234900 }, { "epoch": 0.84, "learning_rate": 8.058332827653658e-06, "loss": 2.6945, "step": 235000 }, { "epoch": 0.84, "learning_rate": 8.04048530970798e-06, "loss": 2.6986, "step": 235100 }, { "epoch": 0.84, "learning_rate": 8.0226377917623e-06, "loss": 2.6973, "step": 235200 }, { "epoch": 0.84, "learning_rate": 8.00479027381662e-06, "loss": 2.6908, "step": 235300 }, { "epoch": 0.84, "learning_rate": 7.986942755870941e-06, "loss": 2.6937, "step": 235400 }, { "epoch": 0.84, "learning_rate": 7.96909523792526e-06, "loss": 2.7081, "step": 235500 }, { "epoch": 0.84, "learning_rate": 7.951247719979584e-06, "loss": 2.6974, "step": 235600 }, { "epoch": 0.84, "learning_rate": 7.933400202033905e-06, "loss": 2.6976, "step": 235700 }, { "epoch": 0.84, "learning_rate": 7.915552684088224e-06, "loss": 2.6937, "step": 235800 }, { "epoch": 0.84, "learning_rate": 7.897705166142546e-06, "loss": 2.6938, "step": 235900 }, { "epoch": 0.84, "learning_rate": 7.879857648196865e-06, "loss": 2.6963, "step": 236000 }, { "epoch": 0.84, "learning_rate": 7.862010130251186e-06, "loss": 2.6978, "step": 236100 }, { "epoch": 0.84, "learning_rate": 7.844162612305508e-06, "loss": 2.6944, "step": 236200 }, { "epoch": 0.84, "learning_rate": 7.826315094359827e-06, "loss": 2.6979, "step": 236300 }, { "epoch": 0.84, "learning_rate": 7.808467576414148e-06, "loss": 2.6917, "step": 236400 }, { "epoch": 0.84, "learning_rate": 7.79062005846847e-06, "loss": 2.6964, "step": 236500 }, { "epoch": 0.84, "learning_rate": 7.772772540522789e-06, "loss": 2.6937, "step": 236600 }, { "epoch": 0.84, "learning_rate": 7.75492502257711e-06, "loss": 2.7025, "step": 236700 }, { "epoch": 0.85, "learning_rate": 7.737077504631431e-06, "loss": 2.6998, "step": 236800 }, { "epoch": 0.85, "learning_rate": 7.719229986685753e-06, "loss": 2.6884, "step": 236900 }, { "epoch": 0.85, "learning_rate": 7.701382468740074e-06, "loss": 2.7001, "step": 237000 }, { "epoch": 0.85, "learning_rate": 7.683534950794393e-06, "loss": 2.6938, "step": 237100 }, { "epoch": 0.85, "learning_rate": 7.665687432848715e-06, "loss": 2.6937, "step": 237200 }, { "epoch": 0.85, "learning_rate": 7.647839914903036e-06, "loss": 2.6936, "step": 237300 }, { "epoch": 0.85, "learning_rate": 7.629992396957355e-06, "loss": 2.6893, "step": 237400 }, { "epoch": 0.85, "learning_rate": 7.6121448790116766e-06, "loss": 2.687, "step": 237500 }, { "epoch": 0.85, "learning_rate": 7.594297361065997e-06, "loss": 2.6948, "step": 237600 }, { "epoch": 0.85, "learning_rate": 7.576449843120317e-06, "loss": 2.6913, "step": 237700 }, { "epoch": 0.85, "learning_rate": 7.558602325174638e-06, "loss": 2.6908, "step": 237800 }, { "epoch": 0.85, "learning_rate": 7.540754807228959e-06, "loss": 2.6938, "step": 237900 }, { "epoch": 0.85, "learning_rate": 7.522907289283279e-06, "loss": 2.6906, "step": 238000 }, { "epoch": 0.85, "learning_rate": 7.505059771337601e-06, "loss": 2.6924, "step": 238100 }, { "epoch": 0.85, "learning_rate": 7.487212253391922e-06, "loss": 2.7051, "step": 238200 }, { "epoch": 0.85, "learning_rate": 7.469364735446242e-06, "loss": 2.6969, "step": 238300 }, { "epoch": 0.85, "learning_rate": 7.451517217500563e-06, "loss": 2.6976, "step": 238400 }, { "epoch": 0.85, "learning_rate": 7.433669699554884e-06, "loss": 2.6947, "step": 238500 }, { "epoch": 0.85, "learning_rate": 7.415822181609204e-06, "loss": 2.6945, "step": 238600 }, { "epoch": 0.85, "learning_rate": 7.397974663663524e-06, "loss": 2.6951, "step": 238700 }, { "epoch": 0.85, "learning_rate": 7.3801271457178456e-06, "loss": 2.6876, "step": 238800 }, { "epoch": 0.85, "learning_rate": 7.362279627772166e-06, "loss": 2.6975, "step": 238900 }, { "epoch": 0.85, "learning_rate": 7.344432109826486e-06, "loss": 2.6884, "step": 239000 }, { "epoch": 0.85, "learning_rate": 7.326584591880807e-06, "loss": 2.6911, "step": 239100 }, { "epoch": 0.85, "learning_rate": 7.308737073935128e-06, "loss": 2.696, "step": 239200 }, { "epoch": 0.85, "learning_rate": 7.290889555989448e-06, "loss": 2.6989, "step": 239300 }, { "epoch": 0.85, "learning_rate": 7.27304203804377e-06, "loss": 2.6914, "step": 239400 }, { "epoch": 0.85, "learning_rate": 7.255194520098091e-06, "loss": 2.6928, "step": 239500 }, { "epoch": 0.86, "learning_rate": 7.237347002152411e-06, "loss": 2.6887, "step": 239600 }, { "epoch": 0.86, "learning_rate": 7.219499484206732e-06, "loss": 2.7037, "step": 239700 }, { "epoch": 0.86, "learning_rate": 7.201651966261053e-06, "loss": 2.6903, "step": 239800 }, { "epoch": 0.86, "learning_rate": 7.183804448315373e-06, "loss": 2.6958, "step": 239900 }, { "epoch": 0.86, "learning_rate": 7.165956930369694e-06, "loss": 2.6902, "step": 240000 }, { "epoch": 0.86, "learning_rate": 7.148109412424015e-06, "loss": 2.6947, "step": 240100 }, { "epoch": 0.86, "learning_rate": 7.130261894478335e-06, "loss": 2.7103, "step": 240200 }, { "epoch": 0.86, "learning_rate": 7.112414376532655e-06, "loss": 2.7019, "step": 240300 }, { "epoch": 0.86, "learning_rate": 7.0945668585869765e-06, "loss": 2.7102, "step": 240400 }, { "epoch": 0.86, "learning_rate": 7.076719340641297e-06, "loss": 2.6914, "step": 240500 }, { "epoch": 0.86, "learning_rate": 7.058871822695617e-06, "loss": 2.6956, "step": 240600 }, { "epoch": 0.86, "learning_rate": 7.041024304749939e-06, "loss": 2.6888, "step": 240700 }, { "epoch": 0.86, "learning_rate": 7.02317678680426e-06, "loss": 2.6916, "step": 240800 }, { "epoch": 0.86, "learning_rate": 7.005329268858581e-06, "loss": 2.6881, "step": 240900 }, { "epoch": 0.86, "learning_rate": 6.987481750912901e-06, "loss": 2.6921, "step": 241000 }, { "epoch": 0.86, "learning_rate": 6.969634232967222e-06, "loss": 2.7049, "step": 241100 }, { "epoch": 0.86, "learning_rate": 6.951786715021542e-06, "loss": 2.6928, "step": 241200 }, { "epoch": 0.86, "learning_rate": 6.933939197075863e-06, "loss": 2.6975, "step": 241300 }, { "epoch": 0.86, "learning_rate": 6.916091679130184e-06, "loss": 2.6956, "step": 241400 }, { "epoch": 0.86, "learning_rate": 6.898244161184504e-06, "loss": 2.6812, "step": 241500 }, { "epoch": 0.86, "learning_rate": 6.880396643238825e-06, "loss": 2.6887, "step": 241600 }, { "epoch": 0.86, "learning_rate": 6.8625491252931456e-06, "loss": 2.6962, "step": 241700 }, { "epoch": 0.86, "learning_rate": 6.844701607347466e-06, "loss": 2.689, "step": 241800 }, { "epoch": 0.86, "learning_rate": 6.826854089401786e-06, "loss": 2.689, "step": 241900 }, { "epoch": 0.86, "learning_rate": 6.809006571456108e-06, "loss": 2.6999, "step": 242000 }, { "epoch": 0.86, "learning_rate": 6.791159053510429e-06, "loss": 2.6883, "step": 242100 }, { "epoch": 0.86, "learning_rate": 6.77331153556475e-06, "loss": 2.6956, "step": 242200 }, { "epoch": 0.86, "learning_rate": 6.75546401761907e-06, "loss": 2.7015, "step": 242300 }, { "epoch": 0.87, "learning_rate": 6.737616499673391e-06, "loss": 2.688, "step": 242400 }, { "epoch": 0.87, "learning_rate": 6.719768981727712e-06, "loss": 2.694, "step": 242500 }, { "epoch": 0.87, "learning_rate": 6.701921463782032e-06, "loss": 2.688, "step": 242600 }, { "epoch": 0.87, "learning_rate": 6.684073945836353e-06, "loss": 2.6889, "step": 242700 }, { "epoch": 0.87, "learning_rate": 6.666226427890673e-06, "loss": 2.6904, "step": 242800 }, { "epoch": 0.87, "learning_rate": 6.648378909944994e-06, "loss": 2.6881, "step": 242900 }, { "epoch": 0.87, "learning_rate": 6.6305313919993146e-06, "loss": 2.689, "step": 243000 }, { "epoch": 0.87, "learning_rate": 6.612683874053635e-06, "loss": 2.6946, "step": 243100 }, { "epoch": 0.87, "learning_rate": 6.594836356107955e-06, "loss": 2.6888, "step": 243200 }, { "epoch": 0.87, "learning_rate": 6.576988838162277e-06, "loss": 2.6866, "step": 243300 }, { "epoch": 0.87, "learning_rate": 6.559141320216599e-06, "loss": 2.7064, "step": 243400 }, { "epoch": 0.87, "learning_rate": 6.541293802270919e-06, "loss": 2.6871, "step": 243500 }, { "epoch": 0.87, "learning_rate": 6.523446284325239e-06, "loss": 2.6867, "step": 243600 }, { "epoch": 0.87, "learning_rate": 6.50559876637956e-06, "loss": 2.6865, "step": 243700 }, { "epoch": 0.87, "learning_rate": 6.487751248433881e-06, "loss": 2.6934, "step": 243800 }, { "epoch": 0.87, "learning_rate": 6.469903730488201e-06, "loss": 2.6892, "step": 243900 }, { "epoch": 0.87, "learning_rate": 6.452056212542522e-06, "loss": 2.7051, "step": 244000 }, { "epoch": 0.87, "learning_rate": 6.434208694596843e-06, "loss": 2.6944, "step": 244100 }, { "epoch": 0.87, "learning_rate": 6.416361176651163e-06, "loss": 2.7003, "step": 244200 }, { "epoch": 0.87, "learning_rate": 6.398513658705484e-06, "loss": 2.6952, "step": 244300 }, { "epoch": 0.87, "learning_rate": 6.380666140759804e-06, "loss": 2.6977, "step": 244400 }, { "epoch": 0.87, "learning_rate": 6.362818622814126e-06, "loss": 2.6896, "step": 244500 }, { "epoch": 0.87, "learning_rate": 6.344971104868447e-06, "loss": 2.6967, "step": 244600 }, { "epoch": 0.87, "learning_rate": 6.327123586922768e-06, "loss": 2.6798, "step": 244700 }, { "epoch": 0.87, "learning_rate": 6.309276068977088e-06, "loss": 2.6945, "step": 244800 }, { "epoch": 0.87, "learning_rate": 6.291428551031408e-06, "loss": 2.6842, "step": 244900 }, { "epoch": 0.87, "learning_rate": 6.2735810330857296e-06, "loss": 2.6888, "step": 245000 }, { "epoch": 0.87, "learning_rate": 6.25573351514005e-06, "loss": 2.6997, "step": 245100 }, { "epoch": 0.88, "learning_rate": 6.23788599719437e-06, "loss": 2.6885, "step": 245200 }, { "epoch": 0.88, "learning_rate": 6.220038479248691e-06, "loss": 2.7027, "step": 245300 }, { "epoch": 0.88, "learning_rate": 6.202190961303012e-06, "loss": 2.6942, "step": 245400 }, { "epoch": 0.88, "learning_rate": 6.184343443357333e-06, "loss": 2.6893, "step": 245500 }, { "epoch": 0.88, "learning_rate": 6.1664959254116535e-06, "loss": 2.7014, "step": 245600 }, { "epoch": 0.88, "learning_rate": 6.148648407465974e-06, "loss": 2.6891, "step": 245700 }, { "epoch": 0.88, "learning_rate": 6.130800889520295e-06, "loss": 2.702, "step": 245800 }, { "epoch": 0.88, "learning_rate": 6.112953371574615e-06, "loss": 2.6909, "step": 245900 }, { "epoch": 0.88, "learning_rate": 6.095105853628936e-06, "loss": 2.6906, "step": 246000 }, { "epoch": 0.88, "learning_rate": 6.077258335683257e-06, "loss": 2.6964, "step": 246100 }, { "epoch": 0.88, "learning_rate": 6.059410817737577e-06, "loss": 2.7003, "step": 246200 }, { "epoch": 0.88, "learning_rate": 6.0415632997918986e-06, "loss": 2.6929, "step": 246300 }, { "epoch": 0.88, "learning_rate": 6.023715781846219e-06, "loss": 2.6888, "step": 246400 }, { "epoch": 0.88, "learning_rate": 6.005868263900539e-06, "loss": 2.701, "step": 246500 }, { "epoch": 0.88, "learning_rate": 5.9880207459548605e-06, "loss": 2.6993, "step": 246600 }, { "epoch": 0.88, "learning_rate": 5.970173228009181e-06, "loss": 2.6882, "step": 246700 }, { "epoch": 0.88, "learning_rate": 5.952325710063502e-06, "loss": 2.6969, "step": 246800 }, { "epoch": 0.88, "learning_rate": 5.9344781921178225e-06, "loss": 2.7008, "step": 246900 }, { "epoch": 0.88, "learning_rate": 5.916630674172143e-06, "loss": 2.6964, "step": 247000 }, { "epoch": 0.88, "learning_rate": 5.898783156226464e-06, "loss": 2.6921, "step": 247100 }, { "epoch": 0.88, "learning_rate": 5.8809356382807844e-06, "loss": 2.6875, "step": 247200 }, { "epoch": 0.88, "learning_rate": 5.863088120335105e-06, "loss": 2.6913, "step": 247300 }, { "epoch": 0.88, "learning_rate": 5.845240602389426e-06, "loss": 2.7004, "step": 247400 }, { "epoch": 0.88, "learning_rate": 5.827393084443747e-06, "loss": 2.6947, "step": 247500 }, { "epoch": 0.88, "learning_rate": 5.809545566498068e-06, "loss": 2.6968, "step": 247600 }, { "epoch": 0.88, "learning_rate": 5.791698048552388e-06, "loss": 2.6929, "step": 247700 }, { "epoch": 0.88, "learning_rate": 5.773850530606708e-06, "loss": 2.686, "step": 247800 }, { "epoch": 0.88, "learning_rate": 5.7560030126610295e-06, "loss": 2.6899, "step": 247900 }, { "epoch": 0.89, "learning_rate": 5.738155494715351e-06, "loss": 2.6949, "step": 248000 }, { "epoch": 0.89, "learning_rate": 5.720307976769671e-06, "loss": 2.6969, "step": 248100 }, { "epoch": 0.89, "learning_rate": 5.7024604588239915e-06, "loss": 2.691, "step": 248200 }, { "epoch": 0.89, "learning_rate": 5.684612940878313e-06, "loss": 2.6885, "step": 248300 }, { "epoch": 0.89, "learning_rate": 5.666765422932633e-06, "loss": 2.69, "step": 248400 }, { "epoch": 0.89, "learning_rate": 5.6489179049869534e-06, "loss": 2.6909, "step": 248500 }, { "epoch": 0.89, "learning_rate": 5.631070387041274e-06, "loss": 2.7055, "step": 248600 }, { "epoch": 0.89, "learning_rate": 5.613222869095595e-06, "loss": 2.6838, "step": 248700 }, { "epoch": 0.89, "learning_rate": 5.595375351149916e-06, "loss": 2.6962, "step": 248800 }, { "epoch": 0.89, "learning_rate": 5.577527833204237e-06, "loss": 2.6954, "step": 248900 }, { "epoch": 0.89, "learning_rate": 5.559680315258557e-06, "loss": 2.6933, "step": 249000 }, { "epoch": 0.89, "learning_rate": 5.541832797312878e-06, "loss": 2.6939, "step": 249100 }, { "epoch": 0.89, "learning_rate": 5.5239852793671986e-06, "loss": 2.6894, "step": 249200 }, { "epoch": 0.89, "learning_rate": 5.50613776142152e-06, "loss": 2.6872, "step": 249300 }, { "epoch": 0.89, "learning_rate": 5.48829024347584e-06, "loss": 2.6983, "step": 249400 }, { "epoch": 0.89, "learning_rate": 5.4704427255301605e-06, "loss": 2.6958, "step": 249500 }, { "epoch": 0.89, "learning_rate": 5.452595207584482e-06, "loss": 2.6866, "step": 249600 }, { "epoch": 0.89, "learning_rate": 5.434747689638802e-06, "loss": 2.6886, "step": 249700 }, { "epoch": 0.89, "learning_rate": 5.4169001716931225e-06, "loss": 2.6916, "step": 249800 }, { "epoch": 0.89, "learning_rate": 5.399052653747444e-06, "loss": 2.6907, "step": 249900 }, { "epoch": 0.89, "learning_rate": 5.381205135801765e-06, "loss": 2.6826, "step": 250000 }, { "epoch": 0.89, "learning_rate": 5.363357617856085e-06, "loss": 2.6888, "step": 250100 }, { "epoch": 0.89, "learning_rate": 5.345510099910406e-06, "loss": 2.6791, "step": 250200 }, { "epoch": 0.89, "learning_rate": 5.327662581964726e-06, "loss": 2.6788, "step": 250300 }, { "epoch": 0.89, "learning_rate": 5.309815064019047e-06, "loss": 2.7038, "step": 250400 }, { "epoch": 0.89, "learning_rate": 5.2919675460733676e-06, "loss": 2.6972, "step": 250500 }, { "epoch": 0.89, "learning_rate": 5.274120028127689e-06, "loss": 2.6859, "step": 250600 }, { "epoch": 0.89, "learning_rate": 5.256272510182009e-06, "loss": 2.6911, "step": 250700 }, { "epoch": 0.9, "learning_rate": 5.23842499223633e-06, "loss": 2.6862, "step": 250800 }, { "epoch": 0.9, "learning_rate": 5.220577474290651e-06, "loss": 2.6866, "step": 250900 }, { "epoch": 0.9, "learning_rate": 5.202729956344971e-06, "loss": 2.676, "step": 251000 }, { "epoch": 0.9, "learning_rate": 5.1848824383992915e-06, "loss": 2.6817, "step": 251100 }, { "epoch": 0.9, "learning_rate": 5.167034920453613e-06, "loss": 2.6998, "step": 251200 }, { "epoch": 0.9, "learning_rate": 5.149187402507934e-06, "loss": 2.6946, "step": 251300 }, { "epoch": 0.9, "learning_rate": 5.131339884562254e-06, "loss": 2.6848, "step": 251400 }, { "epoch": 0.9, "learning_rate": 5.113492366616575e-06, "loss": 2.7021, "step": 251500 }, { "epoch": 0.9, "learning_rate": 5.095644848670896e-06, "loss": 2.695, "step": 251600 }, { "epoch": 0.9, "learning_rate": 5.077797330725216e-06, "loss": 2.6968, "step": 251700 }, { "epoch": 0.9, "learning_rate": 5.059949812779537e-06, "loss": 2.6969, "step": 251800 }, { "epoch": 0.9, "learning_rate": 5.042102294833858e-06, "loss": 2.6867, "step": 251900 }, { "epoch": 0.9, "learning_rate": 5.024254776888178e-06, "loss": 2.6927, "step": 252000 }, { "epoch": 0.9, "learning_rate": 5.006407258942499e-06, "loss": 2.692, "step": 252100 }, { "epoch": 0.9, "learning_rate": 4.98855974099682e-06, "loss": 2.6792, "step": 252200 }, { "epoch": 0.9, "learning_rate": 4.97071222305114e-06, "loss": 2.6827, "step": 252300 }, { "epoch": 0.9, "learning_rate": 4.952864705105461e-06, "loss": 2.6947, "step": 252400 }, { "epoch": 0.9, "learning_rate": 4.9350171871597826e-06, "loss": 2.7035, "step": 252500 }, { "epoch": 0.9, "learning_rate": 4.917169669214103e-06, "loss": 2.6853, "step": 252600 }, { "epoch": 0.9, "learning_rate": 4.899322151268423e-06, "loss": 2.689, "step": 252700 }, { "epoch": 0.9, "learning_rate": 4.881474633322744e-06, "loss": 2.6917, "step": 252800 }, { "epoch": 0.9, "learning_rate": 4.863627115377065e-06, "loss": 2.6907, "step": 252900 }, { "epoch": 0.9, "learning_rate": 4.845779597431385e-06, "loss": 2.6849, "step": 253000 }, { "epoch": 0.9, "learning_rate": 4.8279320794857065e-06, "loss": 2.6868, "step": 253100 }, { "epoch": 0.9, "learning_rate": 4.810084561540027e-06, "loss": 2.6929, "step": 253200 }, { "epoch": 0.9, "learning_rate": 4.792237043594348e-06, "loss": 2.6968, "step": 253300 }, { "epoch": 0.9, "learning_rate": 4.774389525648668e-06, "loss": 2.683, "step": 253400 }, { "epoch": 0.9, "learning_rate": 4.756542007702989e-06, "loss": 2.6751, "step": 253500 }, { "epoch": 0.91, "learning_rate": 4.738694489757309e-06, "loss": 2.6998, "step": 253600 }, { "epoch": 0.91, "learning_rate": 4.72084697181163e-06, "loss": 2.7031, "step": 253700 }, { "epoch": 0.91, "learning_rate": 4.7029994538659516e-06, "loss": 2.6882, "step": 253800 }, { "epoch": 0.91, "learning_rate": 4.685151935920272e-06, "loss": 2.6885, "step": 253900 }, { "epoch": 0.91, "learning_rate": 4.667304417974592e-06, "loss": 2.6818, "step": 254000 }, { "epoch": 0.91, "learning_rate": 4.6494569000289135e-06, "loss": 2.6897, "step": 254100 }, { "epoch": 0.91, "learning_rate": 4.631609382083234e-06, "loss": 2.6905, "step": 254200 }, { "epoch": 0.91, "learning_rate": 4.613761864137554e-06, "loss": 2.6938, "step": 254300 }, { "epoch": 0.91, "learning_rate": 4.5959143461918755e-06, "loss": 2.6965, "step": 254400 }, { "epoch": 0.91, "learning_rate": 4.578066828246196e-06, "loss": 2.6999, "step": 254500 }, { "epoch": 0.91, "learning_rate": 4.560219310300517e-06, "loss": 2.6905, "step": 254600 }, { "epoch": 0.91, "learning_rate": 4.5423717923548374e-06, "loss": 2.6768, "step": 254700 }, { "epoch": 0.91, "learning_rate": 4.524524274409158e-06, "loss": 2.6879, "step": 254800 }, { "epoch": 0.91, "learning_rate": 4.506676756463479e-06, "loss": 2.6928, "step": 254900 }, { "epoch": 0.91, "learning_rate": 4.488829238517799e-06, "loss": 2.6959, "step": 255000 }, { "epoch": 0.91, "learning_rate": 4.470981720572121e-06, "loss": 2.6813, "step": 255100 }, { "epoch": 0.91, "learning_rate": 4.453134202626441e-06, "loss": 2.6957, "step": 255200 }, { "epoch": 0.91, "learning_rate": 4.435286684680761e-06, "loss": 2.6822, "step": 255300 }, { "epoch": 0.91, "learning_rate": 4.4174391667350825e-06, "loss": 2.695, "step": 255400 }, { "epoch": 0.91, "learning_rate": 4.399591648789403e-06, "loss": 2.6949, "step": 255500 }, { "epoch": 0.91, "learning_rate": 4.381744130843723e-06, "loss": 2.6821, "step": 255600 }, { "epoch": 0.91, "learning_rate": 4.3638966128980445e-06, "loss": 2.6904, "step": 255700 }, { "epoch": 0.91, "learning_rate": 4.346049094952366e-06, "loss": 2.6912, "step": 255800 }, { "epoch": 0.91, "learning_rate": 4.328201577006686e-06, "loss": 2.6931, "step": 255900 }, { "epoch": 0.91, "learning_rate": 4.3103540590610064e-06, "loss": 2.686, "step": 256000 }, { "epoch": 0.91, "learning_rate": 4.292506541115327e-06, "loss": 2.6957, "step": 256100 }, { "epoch": 0.91, "learning_rate": 4.274659023169648e-06, "loss": 2.6824, "step": 256200 }, { "epoch": 0.91, "learning_rate": 4.256811505223969e-06, "loss": 2.6841, "step": 256300 }, { "epoch": 0.92, "learning_rate": 4.23896398727829e-06, "loss": 2.6876, "step": 256400 }, { "epoch": 0.92, "learning_rate": 4.22111646933261e-06, "loss": 2.6932, "step": 256500 }, { "epoch": 0.92, "learning_rate": 4.203268951386931e-06, "loss": 2.6916, "step": 256600 }, { "epoch": 0.92, "learning_rate": 4.1854214334412516e-06, "loss": 2.6874, "step": 256700 }, { "epoch": 0.92, "learning_rate": 4.167573915495572e-06, "loss": 2.6823, "step": 256800 }, { "epoch": 0.92, "learning_rate": 4.149726397549892e-06, "loss": 2.687, "step": 256900 }, { "epoch": 0.92, "learning_rate": 4.1318788796042135e-06, "loss": 2.6951, "step": 257000 }, { "epoch": 0.92, "learning_rate": 4.114031361658535e-06, "loss": 2.6977, "step": 257100 }, { "epoch": 0.92, "learning_rate": 4.096183843712855e-06, "loss": 2.6904, "step": 257200 }, { "epoch": 0.92, "learning_rate": 4.0783363257671755e-06, "loss": 2.6816, "step": 257300 }, { "epoch": 0.92, "learning_rate": 4.060488807821497e-06, "loss": 2.6873, "step": 257400 }, { "epoch": 0.92, "learning_rate": 4.042641289875817e-06, "loss": 2.6894, "step": 257500 }, { "epoch": 0.92, "learning_rate": 4.024793771930138e-06, "loss": 2.6893, "step": 257600 }, { "epoch": 0.92, "learning_rate": 4.006946253984459e-06, "loss": 2.6965, "step": 257700 }, { "epoch": 0.92, "learning_rate": 3.989098736038779e-06, "loss": 2.685, "step": 257800 }, { "epoch": 0.92, "learning_rate": 3.9712512180931e-06, "loss": 2.6865, "step": 257900 }, { "epoch": 0.92, "learning_rate": 3.953403700147421e-06, "loss": 2.6911, "step": 258000 }, { "epoch": 0.92, "learning_rate": 3.935556182201741e-06, "loss": 2.6885, "step": 258100 }, { "epoch": 0.92, "learning_rate": 3.917708664256062e-06, "loss": 2.6951, "step": 258200 }, { "epoch": 0.92, "learning_rate": 3.899861146310383e-06, "loss": 2.6964, "step": 258300 }, { "epoch": 0.92, "learning_rate": 3.882013628364704e-06, "loss": 2.6863, "step": 258400 }, { "epoch": 0.92, "learning_rate": 3.864166110419024e-06, "loss": 2.6828, "step": 258500 }, { "epoch": 0.92, "learning_rate": 3.8463185924733445e-06, "loss": 2.6844, "step": 258600 }, { "epoch": 0.92, "learning_rate": 3.828471074527666e-06, "loss": 2.6863, "step": 258700 }, { "epoch": 0.92, "learning_rate": 3.810623556581986e-06, "loss": 2.6983, "step": 258800 }, { "epoch": 0.92, "learning_rate": 3.7927760386363073e-06, "loss": 2.6873, "step": 258900 }, { "epoch": 0.92, "learning_rate": 3.774928520690628e-06, "loss": 2.6917, "step": 259000 }, { "epoch": 0.92, "learning_rate": 3.7570810027449484e-06, "loss": 2.6926, "step": 259100 }, { "epoch": 0.93, "learning_rate": 3.7392334847992692e-06, "loss": 2.688, "step": 259200 }, { "epoch": 0.93, "learning_rate": 3.7213859668535896e-06, "loss": 2.6805, "step": 259300 }, { "epoch": 0.93, "learning_rate": 3.7035384489079104e-06, "loss": 2.6856, "step": 259400 }, { "epoch": 0.93, "learning_rate": 3.6856909309622316e-06, "loss": 2.6907, "step": 259500 }, { "epoch": 0.93, "learning_rate": 3.6678434130165524e-06, "loss": 2.6858, "step": 259600 }, { "epoch": 0.93, "learning_rate": 3.6499958950708728e-06, "loss": 2.6867, "step": 259700 }, { "epoch": 0.93, "learning_rate": 3.6321483771251936e-06, "loss": 2.688, "step": 259800 }, { "epoch": 0.93, "learning_rate": 3.614300859179514e-06, "loss": 2.689, "step": 259900 }, { "epoch": 0.93, "learning_rate": 3.5964533412338347e-06, "loss": 2.6853, "step": 260000 }, { "epoch": 0.93, "learning_rate": 3.578605823288155e-06, "loss": 2.6822, "step": 260100 }, { "epoch": 0.93, "learning_rate": 3.5607583053424763e-06, "loss": 2.6965, "step": 260200 }, { "epoch": 0.93, "learning_rate": 3.542910787396797e-06, "loss": 2.6966, "step": 260300 }, { "epoch": 0.93, "learning_rate": 3.5250632694511175e-06, "loss": 2.6908, "step": 260400 }, { "epoch": 0.93, "learning_rate": 3.5072157515054382e-06, "loss": 2.6828, "step": 260500 }, { "epoch": 0.93, "learning_rate": 3.489368233559759e-06, "loss": 2.6718, "step": 260600 }, { "epoch": 0.93, "learning_rate": 3.4715207156140794e-06, "loss": 2.686, "step": 260700 }, { "epoch": 0.93, "learning_rate": 3.4536731976684006e-06, "loss": 2.6885, "step": 260800 }, { "epoch": 0.93, "learning_rate": 3.4358256797227214e-06, "loss": 2.6732, "step": 260900 }, { "epoch": 0.93, "learning_rate": 3.4179781617770418e-06, "loss": 2.68, "step": 261000 }, { "epoch": 0.93, "learning_rate": 3.4001306438313626e-06, "loss": 2.6809, "step": 261100 }, { "epoch": 0.93, "learning_rate": 3.382283125885683e-06, "loss": 2.6972, "step": 261200 }, { "epoch": 0.93, "learning_rate": 3.3644356079400037e-06, "loss": 2.6898, "step": 261300 }, { "epoch": 0.93, "learning_rate": 3.346588089994325e-06, "loss": 2.6904, "step": 261400 }, { "epoch": 0.93, "learning_rate": 3.3287405720486457e-06, "loss": 2.6834, "step": 261500 }, { "epoch": 0.93, "learning_rate": 3.310893054102966e-06, "loss": 2.6889, "step": 261600 }, { "epoch": 0.93, "learning_rate": 3.293045536157287e-06, "loss": 2.6897, "step": 261700 }, { "epoch": 0.93, "learning_rate": 3.2751980182116073e-06, "loss": 2.6902, "step": 261800 }, { "epoch": 0.93, "learning_rate": 3.257350500265928e-06, "loss": 2.7015, "step": 261900 }, { "epoch": 0.94, "learning_rate": 3.2395029823202484e-06, "loss": 2.6892, "step": 262000 }, { "epoch": 0.94, "learning_rate": 3.22165546437457e-06, "loss": 2.6933, "step": 262100 }, { "epoch": 0.94, "learning_rate": 3.2038079464288904e-06, "loss": 2.6754, "step": 262200 }, { "epoch": 0.94, "learning_rate": 3.1859604284832112e-06, "loss": 2.6942, "step": 262300 }, { "epoch": 0.94, "learning_rate": 3.1681129105375316e-06, "loss": 2.6842, "step": 262400 }, { "epoch": 0.94, "learning_rate": 3.1502653925918524e-06, "loss": 2.676, "step": 262500 }, { "epoch": 0.94, "learning_rate": 3.1324178746461727e-06, "loss": 2.6829, "step": 262600 }, { "epoch": 0.94, "learning_rate": 3.114570356700494e-06, "loss": 2.6774, "step": 262700 }, { "epoch": 0.94, "learning_rate": 3.0967228387548143e-06, "loss": 2.6876, "step": 262800 }, { "epoch": 0.94, "learning_rate": 3.078875320809135e-06, "loss": 2.6815, "step": 262900 }, { "epoch": 0.94, "learning_rate": 3.061027802863456e-06, "loss": 2.6745, "step": 263000 }, { "epoch": 0.94, "learning_rate": 3.0431802849177767e-06, "loss": 2.6781, "step": 263100 }, { "epoch": 0.94, "learning_rate": 3.0253327669720975e-06, "loss": 2.6946, "step": 263200 }, { "epoch": 0.94, "learning_rate": 3.007485249026418e-06, "loss": 2.6873, "step": 263300 }, { "epoch": 0.94, "learning_rate": 2.9896377310807387e-06, "loss": 2.6945, "step": 263400 }, { "epoch": 0.94, "learning_rate": 2.9717902131350594e-06, "loss": 2.6972, "step": 263500 }, { "epoch": 0.94, "learning_rate": 2.9539426951893802e-06, "loss": 2.6925, "step": 263600 }, { "epoch": 0.94, "learning_rate": 2.9360951772437006e-06, "loss": 2.6725, "step": 263700 }, { "epoch": 0.94, "learning_rate": 2.918247659298022e-06, "loss": 2.6786, "step": 263800 }, { "epoch": 0.94, "learning_rate": 2.900400141352342e-06, "loss": 2.6985, "step": 263900 }, { "epoch": 0.94, "learning_rate": 2.882552623406663e-06, "loss": 2.6911, "step": 264000 }, { "epoch": 0.94, "learning_rate": 2.8647051054609833e-06, "loss": 2.6892, "step": 264100 }, { "epoch": 0.94, "learning_rate": 2.8468575875153046e-06, "loss": 2.6764, "step": 264200 }, { "epoch": 0.94, "learning_rate": 2.829010069569625e-06, "loss": 2.6838, "step": 264300 }, { "epoch": 0.94, "learning_rate": 2.8111625516239457e-06, "loss": 2.6867, "step": 264400 }, { "epoch": 0.94, "learning_rate": 2.7933150336782665e-06, "loss": 2.6797, "step": 264500 }, { "epoch": 0.94, "learning_rate": 2.7754675157325873e-06, "loss": 2.6744, "step": 264600 }, { "epoch": 0.94, "learning_rate": 2.7576199977869077e-06, "loss": 2.6841, "step": 264700 }, { "epoch": 0.95, "learning_rate": 2.739772479841229e-06, "loss": 2.6936, "step": 264800 }, { "epoch": 0.95, "learning_rate": 2.7219249618955493e-06, "loss": 2.6896, "step": 264900 }, { "epoch": 0.95, "learning_rate": 2.70407744394987e-06, "loss": 2.6804, "step": 265000 }, { "epoch": 0.95, "learning_rate": 2.686229926004191e-06, "loss": 2.6878, "step": 265100 }, { "epoch": 0.95, "learning_rate": 2.6683824080585116e-06, "loss": 2.6837, "step": 265200 }, { "epoch": 0.95, "learning_rate": 2.650534890112832e-06, "loss": 2.6931, "step": 265300 }, { "epoch": 0.95, "learning_rate": 2.6326873721671528e-06, "loss": 2.6771, "step": 265400 }, { "epoch": 0.95, "learning_rate": 2.6148398542214736e-06, "loss": 2.6959, "step": 265500 }, { "epoch": 0.95, "learning_rate": 2.5969923362757944e-06, "loss": 2.6821, "step": 265600 }, { "epoch": 0.95, "learning_rate": 2.5791448183301147e-06, "loss": 2.6785, "step": 265700 }, { "epoch": 0.95, "learning_rate": 2.5612973003844355e-06, "loss": 2.6831, "step": 265800 }, { "epoch": 0.95, "learning_rate": 2.5434497824387563e-06, "loss": 2.6949, "step": 265900 }, { "epoch": 0.95, "learning_rate": 2.525602264493077e-06, "loss": 2.7065, "step": 266000 }, { "epoch": 0.95, "learning_rate": 2.507754746547398e-06, "loss": 2.6873, "step": 266100 }, { "epoch": 0.95, "learning_rate": 2.4899072286017183e-06, "loss": 2.688, "step": 266200 }, { "epoch": 0.95, "learning_rate": 2.472059710656039e-06, "loss": 2.697, "step": 266300 }, { "epoch": 0.95, "learning_rate": 2.45421219271036e-06, "loss": 2.6867, "step": 266400 }, { "epoch": 0.95, "learning_rate": 2.4363646747646806e-06, "loss": 2.6913, "step": 266500 }, { "epoch": 0.95, "learning_rate": 2.418517156819001e-06, "loss": 2.6789, "step": 266600 }, { "epoch": 0.95, "learning_rate": 2.4006696388733222e-06, "loss": 2.6866, "step": 266700 }, { "epoch": 0.95, "learning_rate": 2.3828221209276426e-06, "loss": 2.6865, "step": 266800 }, { "epoch": 0.95, "learning_rate": 2.3649746029819634e-06, "loss": 2.6802, "step": 266900 }, { "epoch": 0.95, "learning_rate": 2.347127085036284e-06, "loss": 2.6881, "step": 267000 }, { "epoch": 0.95, "learning_rate": 2.329279567090605e-06, "loss": 2.6901, "step": 267100 }, { "epoch": 0.95, "learning_rate": 2.3114320491449253e-06, "loss": 2.6825, "step": 267200 }, { "epoch": 0.95, "learning_rate": 2.293584531199246e-06, "loss": 2.6923, "step": 267300 }, { "epoch": 0.95, "learning_rate": 2.275737013253567e-06, "loss": 2.6966, "step": 267400 }, { "epoch": 0.95, "learning_rate": 2.2578894953078877e-06, "loss": 2.6804, "step": 267500 }, { "epoch": 0.96, "learning_rate": 2.240041977362208e-06, "loss": 2.69, "step": 267600 }, { "epoch": 0.96, "learning_rate": 2.2221944594165293e-06, "loss": 2.6876, "step": 267700 }, { "epoch": 0.96, "learning_rate": 2.2043469414708497e-06, "loss": 2.6957, "step": 267800 }, { "epoch": 0.96, "learning_rate": 2.1864994235251705e-06, "loss": 2.6904, "step": 267900 }, { "epoch": 0.96, "learning_rate": 2.1686519055794912e-06, "loss": 2.6903, "step": 268000 }, { "epoch": 0.96, "learning_rate": 2.150804387633812e-06, "loss": 2.6769, "step": 268100 }, { "epoch": 0.96, "learning_rate": 2.1329568696881324e-06, "loss": 2.6971, "step": 268200 }, { "epoch": 0.96, "learning_rate": 2.115109351742453e-06, "loss": 2.6879, "step": 268300 }, { "epoch": 0.96, "learning_rate": 2.097261833796774e-06, "loss": 2.68, "step": 268400 }, { "epoch": 0.96, "learning_rate": 2.0794143158510948e-06, "loss": 2.6877, "step": 268500 }, { "epoch": 0.96, "learning_rate": 2.0615667979054156e-06, "loss": 2.686, "step": 268600 }, { "epoch": 0.96, "learning_rate": 2.043719279959736e-06, "loss": 2.6775, "step": 268700 }, { "epoch": 0.96, "learning_rate": 2.0258717620140567e-06, "loss": 2.6828, "step": 268800 }, { "epoch": 0.96, "learning_rate": 2.0080242440683775e-06, "loss": 2.6877, "step": 268900 }, { "epoch": 0.96, "learning_rate": 1.9901767261226983e-06, "loss": 2.6853, "step": 269000 }, { "epoch": 0.96, "learning_rate": 1.9723292081770187e-06, "loss": 2.6961, "step": 269100 }, { "epoch": 0.96, "learning_rate": 1.9544816902313395e-06, "loss": 2.6897, "step": 269200 }, { "epoch": 0.96, "learning_rate": 1.9366341722856603e-06, "loss": 2.7004, "step": 269300 }, { "epoch": 0.96, "learning_rate": 1.918786654339981e-06, "loss": 2.6924, "step": 269400 }, { "epoch": 0.96, "learning_rate": 1.9009391363943016e-06, "loss": 2.6848, "step": 269500 }, { "epoch": 0.96, "learning_rate": 1.8830916184486226e-06, "loss": 2.6926, "step": 269600 }, { "epoch": 0.96, "learning_rate": 1.8652441005029432e-06, "loss": 2.6867, "step": 269700 }, { "epoch": 0.96, "learning_rate": 1.8473965825572638e-06, "loss": 2.6938, "step": 269800 }, { "epoch": 0.96, "learning_rate": 1.8295490646115846e-06, "loss": 2.6975, "step": 269900 }, { "epoch": 0.96, "learning_rate": 1.8117015466659054e-06, "loss": 2.687, "step": 270000 }, { "epoch": 0.96, "learning_rate": 1.793854028720226e-06, "loss": 2.6801, "step": 270100 }, { "epoch": 0.96, "learning_rate": 1.7760065107745468e-06, "loss": 2.6872, "step": 270200 }, { "epoch": 0.96, "learning_rate": 1.7581589928288673e-06, "loss": 2.6981, "step": 270300 }, { "epoch": 0.97, "learning_rate": 1.740311474883188e-06, "loss": 2.6815, "step": 270400 }, { "epoch": 0.97, "learning_rate": 1.722463956937509e-06, "loss": 2.685, "step": 270500 }, { "epoch": 0.97, "learning_rate": 1.7046164389918295e-06, "loss": 2.6959, "step": 270600 }, { "epoch": 0.97, "learning_rate": 1.68676892104615e-06, "loss": 2.6866, "step": 270700 }, { "epoch": 0.97, "learning_rate": 1.6689214031004707e-06, "loss": 2.6828, "step": 270800 }, { "epoch": 0.97, "learning_rate": 1.6510738851547917e-06, "loss": 2.6883, "step": 270900 }, { "epoch": 0.97, "learning_rate": 1.6332263672091122e-06, "loss": 2.6851, "step": 271000 }, { "epoch": 0.97, "learning_rate": 1.6153788492634328e-06, "loss": 2.6877, "step": 271100 }, { "epoch": 0.97, "learning_rate": 1.5975313313177538e-06, "loss": 2.6836, "step": 271200 }, { "epoch": 0.97, "learning_rate": 1.5796838133720744e-06, "loss": 2.6864, "step": 271300 }, { "epoch": 0.97, "learning_rate": 1.5618362954263952e-06, "loss": 2.68, "step": 271400 }, { "epoch": 0.97, "learning_rate": 1.5439887774807158e-06, "loss": 2.6907, "step": 271500 }, { "epoch": 0.97, "learning_rate": 1.5261412595350366e-06, "loss": 2.6813, "step": 271600 }, { "epoch": 0.97, "learning_rate": 1.5082937415893574e-06, "loss": 2.682, "step": 271700 }, { "epoch": 0.97, "learning_rate": 1.490446223643678e-06, "loss": 2.6878, "step": 271800 }, { "epoch": 0.97, "learning_rate": 1.4725987056979987e-06, "loss": 2.6864, "step": 271900 }, { "epoch": 0.97, "learning_rate": 1.4547511877523193e-06, "loss": 2.6798, "step": 272000 }, { "epoch": 0.97, "learning_rate": 1.43690366980664e-06, "loss": 2.6884, "step": 272100 }, { "epoch": 0.97, "learning_rate": 1.4190561518609609e-06, "loss": 2.6844, "step": 272200 }, { "epoch": 0.97, "learning_rate": 1.4012086339152815e-06, "loss": 2.6828, "step": 272300 }, { "epoch": 0.97, "learning_rate": 1.3833611159696023e-06, "loss": 2.6909, "step": 272400 }, { "epoch": 0.97, "learning_rate": 1.365513598023923e-06, "loss": 2.6864, "step": 272500 }, { "epoch": 0.97, "learning_rate": 1.3476660800782436e-06, "loss": 2.6834, "step": 272600 }, { "epoch": 0.97, "learning_rate": 1.3298185621325644e-06, "loss": 2.6892, "step": 272700 }, { "epoch": 0.97, "learning_rate": 1.311971044186885e-06, "loss": 2.6802, "step": 272800 }, { "epoch": 0.97, "learning_rate": 1.2941235262412056e-06, "loss": 2.684, "step": 272900 }, { "epoch": 0.97, "learning_rate": 1.2762760082955264e-06, "loss": 2.6775, "step": 273000 }, { "epoch": 0.97, "learning_rate": 1.258428490349847e-06, "loss": 2.6922, "step": 273100 }, { "epoch": 0.98, "learning_rate": 1.2405809724041677e-06, "loss": 2.6803, "step": 273200 }, { "epoch": 0.98, "learning_rate": 1.2227334544584885e-06, "loss": 2.6858, "step": 273300 }, { "epoch": 0.98, "learning_rate": 1.2048859365128091e-06, "loss": 2.6886, "step": 273400 }, { "epoch": 0.98, "learning_rate": 1.18703841856713e-06, "loss": 2.6808, "step": 273500 }, { "epoch": 0.98, "learning_rate": 1.1691909006214505e-06, "loss": 2.6876, "step": 273600 }, { "epoch": 0.98, "learning_rate": 1.1513433826757713e-06, "loss": 2.6869, "step": 273700 }, { "epoch": 0.98, "learning_rate": 1.133495864730092e-06, "loss": 2.6812, "step": 273800 }, { "epoch": 0.98, "learning_rate": 1.1156483467844126e-06, "loss": 2.6768, "step": 273900 }, { "epoch": 0.98, "learning_rate": 1.0978008288387334e-06, "loss": 2.6939, "step": 274000 }, { "epoch": 0.98, "learning_rate": 1.0799533108930542e-06, "loss": 2.6957, "step": 274100 }, { "epoch": 0.98, "learning_rate": 1.0621057929473748e-06, "loss": 2.6826, "step": 274200 }, { "epoch": 0.98, "learning_rate": 1.0442582750016956e-06, "loss": 2.681, "step": 274300 }, { "epoch": 0.98, "learning_rate": 1.0264107570560162e-06, "loss": 2.6763, "step": 274400 }, { "epoch": 0.98, "learning_rate": 1.008563239110337e-06, "loss": 2.6828, "step": 274500 }, { "epoch": 0.98, "learning_rate": 9.907157211646578e-07, "loss": 2.6802, "step": 274600 }, { "epoch": 0.98, "learning_rate": 9.728682032189783e-07, "loss": 2.6733, "step": 274700 }, { "epoch": 0.98, "learning_rate": 9.550206852732991e-07, "loss": 2.6931, "step": 274800 }, { "epoch": 0.98, "learning_rate": 9.371731673276198e-07, "loss": 2.6864, "step": 274900 }, { "epoch": 0.98, "learning_rate": 9.193256493819405e-07, "loss": 2.6891, "step": 275000 }, { "epoch": 0.98, "learning_rate": 9.014781314362612e-07, "loss": 2.686, "step": 275100 }, { "epoch": 0.98, "learning_rate": 8.836306134905818e-07, "loss": 2.6796, "step": 275200 }, { "epoch": 0.98, "learning_rate": 8.657830955449026e-07, "loss": 2.6936, "step": 275300 }, { "epoch": 0.98, "learning_rate": 8.479355775992234e-07, "loss": 2.6868, "step": 275400 }, { "epoch": 0.98, "learning_rate": 8.300880596535439e-07, "loss": 2.6777, "step": 275500 }, { "epoch": 0.98, "learning_rate": 8.122405417078647e-07, "loss": 2.6859, "step": 275600 }, { "epoch": 0.98, "learning_rate": 7.943930237621855e-07, "loss": 2.6915, "step": 275700 }, { "epoch": 0.98, "learning_rate": 7.765455058165062e-07, "loss": 2.6764, "step": 275800 }, { "epoch": 0.98, "learning_rate": 7.586979878708269e-07, "loss": 2.6742, "step": 275900 }, { "epoch": 0.99, "learning_rate": 7.408504699251476e-07, "loss": 2.6871, "step": 276000 }, { "epoch": 0.99, "learning_rate": 7.230029519794683e-07, "loss": 2.6726, "step": 276100 }, { "epoch": 0.99, "learning_rate": 7.05155434033789e-07, "loss": 2.6765, "step": 276200 }, { "epoch": 0.99, "learning_rate": 6.873079160881096e-07, "loss": 2.6749, "step": 276300 }, { "epoch": 0.99, "learning_rate": 6.694603981424303e-07, "loss": 2.685, "step": 276400 }, { "epoch": 0.99, "learning_rate": 6.51612880196751e-07, "loss": 2.6819, "step": 276500 }, { "epoch": 0.99, "learning_rate": 6.337653622510718e-07, "loss": 2.6861, "step": 276600 }, { "epoch": 0.99, "learning_rate": 6.159178443053925e-07, "loss": 2.6946, "step": 276700 }, { "epoch": 0.99, "learning_rate": 5.980703263597132e-07, "loss": 2.6903, "step": 276800 }, { "epoch": 0.99, "learning_rate": 5.802228084140338e-07, "loss": 2.6886, "step": 276900 }, { "epoch": 0.99, "learning_rate": 5.623752904683546e-07, "loss": 2.6833, "step": 277000 }, { "epoch": 0.99, "learning_rate": 5.445277725226753e-07, "loss": 2.6829, "step": 277100 }, { "epoch": 0.99, "learning_rate": 5.26680254576996e-07, "loss": 2.69, "step": 277200 }, { "epoch": 0.99, "learning_rate": 5.088327366313167e-07, "loss": 2.6866, "step": 277300 }, { "epoch": 0.99, "learning_rate": 4.909852186856375e-07, "loss": 2.6755, "step": 277400 }, { "epoch": 0.99, "learning_rate": 4.731377007399581e-07, "loss": 2.6973, "step": 277500 }, { "epoch": 0.99, "learning_rate": 4.552901827942788e-07, "loss": 2.6872, "step": 277600 }, { "epoch": 0.99, "learning_rate": 4.374426648485995e-07, "loss": 2.6886, "step": 277700 }, { "epoch": 0.99, "learning_rate": 4.195951469029203e-07, "loss": 2.697, "step": 277800 }, { "epoch": 0.99, "learning_rate": 4.0174762895724097e-07, "loss": 2.6859, "step": 277900 }, { "epoch": 0.99, "learning_rate": 3.8390011101156165e-07, "loss": 2.7017, "step": 278000 }, { "epoch": 0.99, "learning_rate": 3.6605259306588234e-07, "loss": 2.6748, "step": 278100 }, { "epoch": 0.99, "learning_rate": 3.48205075120203e-07, "loss": 2.6919, "step": 278200 }, { "epoch": 0.99, "learning_rate": 3.3035755717452376e-07, "loss": 2.6853, "step": 278300 }, { "epoch": 0.99, "learning_rate": 3.1251003922884445e-07, "loss": 2.6843, "step": 278400 }, { "epoch": 0.99, "learning_rate": 2.946625212831652e-07, "loss": 2.6747, "step": 278500 }, { "epoch": 0.99, "learning_rate": 2.7681500333748587e-07, "loss": 2.6838, "step": 278600 }, { "epoch": 0.99, "learning_rate": 2.589674853918066e-07, "loss": 2.6912, "step": 278700 }, { "epoch": 1.0, "learning_rate": 2.411199674461273e-07, "loss": 2.6809, "step": 278800 }, { "epoch": 1.0, "learning_rate": 2.23272449500448e-07, "loss": 2.6763, "step": 278900 }, { "epoch": 1.0, "learning_rate": 2.0542493155476867e-07, "loss": 2.6928, "step": 279000 }, { "epoch": 1.0, "learning_rate": 1.8757741360908938e-07, "loss": 2.6839, "step": 279100 }, { "epoch": 1.0, "learning_rate": 1.697298956634101e-07, "loss": 2.6872, "step": 279200 }, { "epoch": 1.0, "learning_rate": 1.518823777177308e-07, "loss": 2.6973, "step": 279300 }, { "epoch": 1.0, "learning_rate": 1.340348597720515e-07, "loss": 2.706, "step": 279400 }, { "epoch": 1.0, "learning_rate": 1.1618734182637223e-07, "loss": 2.6818, "step": 279500 }, { "epoch": 1.0, "learning_rate": 9.833982388069292e-08, "loss": 2.6777, "step": 279600 }, { "epoch": 1.0, "learning_rate": 8.049230593501362e-08, "loss": 2.692, "step": 279700 }, { "epoch": 1.0, "learning_rate": 6.264478798933432e-08, "loss": 2.6734, "step": 279800 }, { "epoch": 1.0, "learning_rate": 4.4797270043655027e-08, "loss": 2.6892, "step": 279900 }, { "epoch": 1.0, "learning_rate": 2.6949752097975735e-08, "loss": 2.6823, "step": 280000 } ], "max_steps": 280151, "num_train_epochs": 1, "total_flos": 7.58820243932971e+18, "trial_name": null, "trial_params": null }