{ "best_metric": null, "best_model_checkpoint": null, "epoch": 3.023207577828543, "eval_steps": 1000000, "global_step": 316291, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 5.000000000000001e-07, "loss": 10.1142, "step": 500 }, { "epoch": 0.01, "learning_rate": 1.0000000000000002e-06, "loss": 9.5511, "step": 1000 }, { "epoch": 0.01, "learning_rate": 1.5e-06, "loss": 9.2328, "step": 1500 }, { "epoch": 0.02, "learning_rate": 2.0000000000000003e-06, "loss": 8.8678, "step": 2000 }, { "epoch": 0.02, "learning_rate": 2.5e-06, "loss": 8.4212, "step": 2500 }, { "epoch": 0.03, "learning_rate": 3e-06, "loss": 7.9655, "step": 3000 }, { "epoch": 0.03, "learning_rate": 3.5e-06, "loss": 7.5665, "step": 3500 }, { "epoch": 0.04, "learning_rate": 4.000000000000001e-06, "loss": 7.29, "step": 4000 }, { "epoch": 0.04, "learning_rate": 4.5e-06, "loss": 7.1147, "step": 4500 }, { "epoch": 0.05, "learning_rate": 5e-06, "loss": 7.0175, "step": 5000 }, { "epoch": 0.05, "learning_rate": 5.500000000000001e-06, "loss": 6.9471, "step": 5500 }, { "epoch": 0.06, "learning_rate": 6e-06, "loss": 6.8963, "step": 6000 }, { "epoch": 0.06, "learning_rate": 6.5000000000000004e-06, "loss": 6.8524, "step": 6500 }, { "epoch": 0.07, "learning_rate": 7e-06, "loss": 6.8028, "step": 7000 }, { "epoch": 0.07, "learning_rate": 7.500000000000001e-06, "loss": 6.7756, "step": 7500 }, { "epoch": 0.08, "learning_rate": 8.000000000000001e-06, "loss": 6.7456, "step": 8000 }, { "epoch": 0.08, "learning_rate": 8.5e-06, "loss": 6.7231, "step": 8500 }, { "epoch": 0.09, "learning_rate": 9e-06, "loss": 6.7018, "step": 9000 }, { "epoch": 0.09, "learning_rate": 9.5e-06, "loss": 6.6652, "step": 9500 }, { "epoch": 0.1, "learning_rate": 1e-05, "loss": 6.6464, "step": 10000 }, { "epoch": 0.1, "learning_rate": 9.999521627232806e-06, "loss": 6.6252, "step": 10500 }, { "epoch": 0.11, "learning_rate": 9.99904325446561e-06, "loss": 6.6039, "step": 11000 }, { "epoch": 0.11, "learning_rate": 9.998564881698415e-06, "loss": 6.5831, "step": 11500 }, { "epoch": 0.11, "learning_rate": 9.99808650893122e-06, "loss": 6.5715, "step": 12000 }, { "epoch": 0.12, "learning_rate": 9.997608136164025e-06, "loss": 6.5571, "step": 12500 }, { "epoch": 0.12, "learning_rate": 9.99712976339683e-06, "loss": 6.5398, "step": 13000 }, { "epoch": 0.13, "learning_rate": 9.996651390629635e-06, "loss": 6.5275, "step": 13500 }, { "epoch": 0.13, "learning_rate": 9.99617301786244e-06, "loss": 6.5166, "step": 14000 }, { "epoch": 0.14, "learning_rate": 9.995694645095244e-06, "loss": 6.5021, "step": 14500 }, { "epoch": 0.14, "learning_rate": 9.99521627232805e-06, "loss": 6.4914, "step": 15000 }, { "epoch": 0.15, "learning_rate": 9.994737899560854e-06, "loss": 6.4798, "step": 15500 }, { "epoch": 0.15, "learning_rate": 9.994259526793659e-06, "loss": 6.4637, "step": 16000 }, { "epoch": 0.16, "learning_rate": 9.993781154026464e-06, "loss": 6.4586, "step": 16500 }, { "epoch": 0.16, "learning_rate": 9.99330278125927e-06, "loss": 6.4408, "step": 17000 }, { "epoch": 0.17, "learning_rate": 9.992824408492075e-06, "loss": 6.4345, "step": 17500 }, { "epoch": 0.17, "learning_rate": 9.992346035724878e-06, "loss": 6.4254, "step": 18000 }, { "epoch": 0.18, "learning_rate": 9.991867662957683e-06, "loss": 6.4096, "step": 18500 }, { "epoch": 0.18, "learning_rate": 9.991389290190488e-06, "loss": 6.4029, "step": 19000 }, { "epoch": 0.19, "learning_rate": 9.990910917423293e-06, "loss": 6.4015, "step": 19500 }, { "epoch": 0.19, "learning_rate": 9.9904325446561e-06, "loss": 6.3942, "step": 20000 }, { "epoch": 0.2, "learning_rate": 9.989954171888904e-06, "loss": 6.3887, "step": 20500 }, { "epoch": 0.2, "learning_rate": 9.989475799121709e-06, "loss": 6.3738, "step": 21000 }, { "epoch": 0.21, "learning_rate": 9.988997426354512e-06, "loss": 6.3742, "step": 21500 }, { "epoch": 0.21, "learning_rate": 9.988519053587317e-06, "loss": 6.367, "step": 22000 }, { "epoch": 0.22, "learning_rate": 9.988040680820124e-06, "loss": 6.3631, "step": 22500 }, { "epoch": 0.22, "learning_rate": 9.987562308052929e-06, "loss": 6.3489, "step": 23000 }, { "epoch": 0.22, "learning_rate": 9.987083935285733e-06, "loss": 6.3379, "step": 23500 }, { "epoch": 0.23, "learning_rate": 9.986605562518538e-06, "loss": 6.3388, "step": 24000 }, { "epoch": 0.23, "learning_rate": 9.986127189751343e-06, "loss": 6.3334, "step": 24500 }, { "epoch": 0.24, "learning_rate": 9.985648816984146e-06, "loss": 6.3269, "step": 25000 }, { "epoch": 0.24, "learning_rate": 9.985170444216953e-06, "loss": 6.3212, "step": 25500 }, { "epoch": 0.25, "learning_rate": 9.984692071449758e-06, "loss": 6.3185, "step": 26000 }, { "epoch": 0.25, "learning_rate": 9.984213698682562e-06, "loss": 6.3145, "step": 26500 }, { "epoch": 0.26, "learning_rate": 9.983735325915367e-06, "loss": 6.3026, "step": 27000 }, { "epoch": 0.26, "learning_rate": 9.983256953148172e-06, "loss": 6.3089, "step": 27500 }, { "epoch": 0.27, "learning_rate": 9.982778580380977e-06, "loss": 6.2963, "step": 28000 }, { "epoch": 0.27, "learning_rate": 9.982300207613782e-06, "loss": 6.2851, "step": 28500 }, { "epoch": 0.28, "learning_rate": 9.981821834846587e-06, "loss": 6.2834, "step": 29000 }, { "epoch": 0.28, "learning_rate": 9.981343462079391e-06, "loss": 6.2829, "step": 29500 }, { "epoch": 0.29, "learning_rate": 9.980865089312196e-06, "loss": 6.2815, "step": 30000 }, { "epoch": 0.29, "learning_rate": 9.980386716545001e-06, "loss": 6.2791, "step": 30500 }, { "epoch": 0.3, "learning_rate": 9.979908343777806e-06, "loss": 6.2612, "step": 31000 }, { "epoch": 0.3, "learning_rate": 9.979429971010611e-06, "loss": 6.2633, "step": 31500 }, { "epoch": 0.31, "learning_rate": 9.978951598243416e-06, "loss": 6.269, "step": 32000 }, { "epoch": 0.31, "learning_rate": 9.97847322547622e-06, "loss": 6.2536, "step": 32500 }, { "epoch": 0.32, "learning_rate": 9.977994852709025e-06, "loss": 6.2565, "step": 33000 }, { "epoch": 0.32, "learning_rate": 9.977516479941832e-06, "loss": 6.2539, "step": 33500 }, { "epoch": 0.32, "learning_rate": 9.977038107174635e-06, "loss": 6.2455, "step": 34000 }, { "epoch": 0.33, "learning_rate": 9.97655973440744e-06, "loss": 6.2427, "step": 34500 }, { "epoch": 0.33, "learning_rate": 9.976081361640245e-06, "loss": 6.245, "step": 35000 }, { "epoch": 0.34, "learning_rate": 9.97560298887305e-06, "loss": 6.2428, "step": 35500 }, { "epoch": 0.34, "learning_rate": 9.975124616105856e-06, "loss": 6.2307, "step": 36000 }, { "epoch": 0.35, "learning_rate": 9.974646243338661e-06, "loss": 6.2386, "step": 36500 }, { "epoch": 0.35, "learning_rate": 9.974167870571466e-06, "loss": 6.2326, "step": 37000 }, { "epoch": 0.36, "learning_rate": 9.973689497804269e-06, "loss": 6.2253, "step": 37500 }, { "epoch": 0.36, "learning_rate": 9.973211125037074e-06, "loss": 6.2274, "step": 38000 }, { "epoch": 0.37, "learning_rate": 9.972732752269879e-06, "loss": 6.223, "step": 38500 }, { "epoch": 0.37, "learning_rate": 9.972254379502685e-06, "loss": 6.2102, "step": 39000 }, { "epoch": 0.38, "learning_rate": 9.97177600673549e-06, "loss": 6.2161, "step": 39500 }, { "epoch": 0.38, "learning_rate": 9.971297633968295e-06, "loss": 6.2132, "step": 40000 }, { "epoch": 0.39, "learning_rate": 9.9708192612011e-06, "loss": 6.2075, "step": 40500 }, { "epoch": 0.39, "learning_rate": 9.970340888433903e-06, "loss": 6.2137, "step": 41000 }, { "epoch": 0.4, "learning_rate": 9.96986251566671e-06, "loss": 6.2015, "step": 41500 }, { "epoch": 0.4, "learning_rate": 9.969384142899514e-06, "loss": 6.2033, "step": 42000 }, { "epoch": 0.41, "learning_rate": 9.96890577013232e-06, "loss": 6.2043, "step": 42500 }, { "epoch": 0.41, "learning_rate": 9.968427397365124e-06, "loss": 6.1973, "step": 43000 }, { "epoch": 0.42, "learning_rate": 9.967949024597929e-06, "loss": 6.1969, "step": 43500 }, { "epoch": 0.42, "learning_rate": 9.967470651830734e-06, "loss": 6.1928, "step": 44000 }, { "epoch": 0.43, "learning_rate": 9.966992279063539e-06, "loss": 6.1808, "step": 44500 }, { "epoch": 0.43, "learning_rate": 9.966513906296343e-06, "loss": 6.1866, "step": 45000 }, { "epoch": 0.43, "learning_rate": 9.966035533529148e-06, "loss": 6.1835, "step": 45500 }, { "epoch": 0.44, "learning_rate": 9.965557160761953e-06, "loss": 6.181, "step": 46000 }, { "epoch": 0.44, "learning_rate": 9.965078787994758e-06, "loss": 6.1803, "step": 46500 }, { "epoch": 0.45, "learning_rate": 9.964600415227563e-06, "loss": 6.1733, "step": 47000 }, { "epoch": 0.45, "learning_rate": 9.964122042460368e-06, "loss": 6.1784, "step": 47500 }, { "epoch": 0.46, "learning_rate": 9.963643669693172e-06, "loss": 6.1749, "step": 48000 }, { "epoch": 0.46, "learning_rate": 9.963165296925977e-06, "loss": 6.1726, "step": 48500 }, { "epoch": 0.47, "learning_rate": 9.962686924158782e-06, "loss": 6.1698, "step": 49000 }, { "epoch": 0.47, "learning_rate": 9.962208551391587e-06, "loss": 6.1635, "step": 49500 }, { "epoch": 0.48, "learning_rate": 9.961730178624392e-06, "loss": 6.159, "step": 50000 }, { "epoch": 0.48, "learning_rate": 9.961251805857197e-06, "loss": 6.1591, "step": 50500 }, { "epoch": 0.49, "learning_rate": 9.960773433090002e-06, "loss": 6.1599, "step": 51000 }, { "epoch": 0.49, "learning_rate": 9.960295060322806e-06, "loss": 6.1562, "step": 51500 }, { "epoch": 0.5, "learning_rate": 9.959816687555611e-06, "loss": 6.1611, "step": 52000 }, { "epoch": 0.5, "learning_rate": 9.959338314788418e-06, "loss": 6.1564, "step": 52500 }, { "epoch": 0.51, "learning_rate": 9.958859942021221e-06, "loss": 6.1521, "step": 53000 }, { "epoch": 0.51, "learning_rate": 9.958381569254026e-06, "loss": 6.149, "step": 53500 }, { "epoch": 0.52, "learning_rate": 9.95790319648683e-06, "loss": 6.1447, "step": 54000 }, { "epoch": 0.52, "learning_rate": 9.957424823719635e-06, "loss": 6.1425, "step": 54500 }, { "epoch": 0.53, "learning_rate": 9.95694645095244e-06, "loss": 6.1485, "step": 55000 }, { "epoch": 0.53, "learning_rate": 9.956468078185247e-06, "loss": 6.1379, "step": 55500 }, { "epoch": 0.54, "learning_rate": 9.955989705418052e-06, "loss": 6.1448, "step": 56000 }, { "epoch": 0.54, "learning_rate": 9.955511332650855e-06, "loss": 6.1442, "step": 56500 }, { "epoch": 0.54, "learning_rate": 9.95503295988366e-06, "loss": 6.1347, "step": 57000 }, { "epoch": 0.55, "learning_rate": 9.954554587116465e-06, "loss": 6.1384, "step": 57500 }, { "epoch": 0.55, "learning_rate": 9.954076214349271e-06, "loss": 6.135, "step": 58000 }, { "epoch": 0.56, "learning_rate": 9.953597841582076e-06, "loss": 6.1303, "step": 58500 }, { "epoch": 0.56, "learning_rate": 9.95311946881488e-06, "loss": 6.1298, "step": 59000 }, { "epoch": 0.57, "learning_rate": 9.952641096047686e-06, "loss": 6.1275, "step": 59500 }, { "epoch": 0.57, "learning_rate": 9.952162723280489e-06, "loss": 6.1239, "step": 60000 }, { "epoch": 0.58, "learning_rate": 9.951684350513294e-06, "loss": 6.1284, "step": 60500 }, { "epoch": 0.58, "learning_rate": 9.9512059777461e-06, "loss": 6.1326, "step": 61000 }, { "epoch": 0.59, "learning_rate": 9.950727604978905e-06, "loss": 6.1208, "step": 61500 }, { "epoch": 0.59, "learning_rate": 9.95024923221171e-06, "loss": 6.124, "step": 62000 }, { "epoch": 0.6, "learning_rate": 9.949770859444515e-06, "loss": 6.1214, "step": 62500 }, { "epoch": 0.6, "learning_rate": 9.94929248667732e-06, "loss": 6.1169, "step": 63000 }, { "epoch": 0.61, "learning_rate": 9.948814113910124e-06, "loss": 6.1203, "step": 63500 }, { "epoch": 0.61, "learning_rate": 9.94833574114293e-06, "loss": 6.1184, "step": 64000 }, { "epoch": 0.62, "learning_rate": 9.947857368375734e-06, "loss": 6.1153, "step": 64500 }, { "epoch": 0.62, "learning_rate": 9.947378995608539e-06, "loss": 6.1117, "step": 65000 }, { "epoch": 0.63, "learning_rate": 9.946900622841344e-06, "loss": 6.1216, "step": 65500 }, { "epoch": 0.63, "learning_rate": 9.946422250074149e-06, "loss": 6.1097, "step": 66000 }, { "epoch": 0.64, "learning_rate": 9.945943877306953e-06, "loss": 6.11, "step": 66500 }, { "epoch": 0.64, "learning_rate": 9.945465504539758e-06, "loss": 6.1052, "step": 67000 }, { "epoch": 0.65, "learning_rate": 9.944987131772563e-06, "loss": 6.1073, "step": 67500 }, { "epoch": 0.65, "learning_rate": 9.944508759005368e-06, "loss": 6.1028, "step": 68000 }, { "epoch": 0.65, "learning_rate": 9.944030386238173e-06, "loss": 6.1119, "step": 68500 }, { "epoch": 0.66, "learning_rate": 9.943552013470978e-06, "loss": 6.1071, "step": 69000 }, { "epoch": 0.66, "learning_rate": 9.943073640703783e-06, "loss": 6.0998, "step": 69500 }, { "epoch": 0.67, "learning_rate": 9.942595267936587e-06, "loss": 6.0989, "step": 70000 }, { "epoch": 0.67, "learning_rate": 9.942116895169392e-06, "loss": 6.1023, "step": 70500 }, { "epoch": 0.68, "learning_rate": 9.941638522402197e-06, "loss": 6.0954, "step": 71000 }, { "epoch": 0.68, "learning_rate": 9.941160149635002e-06, "loss": 6.096, "step": 71500 }, { "epoch": 0.69, "learning_rate": 9.940681776867808e-06, "loss": 6.0966, "step": 72000 }, { "epoch": 0.69, "learning_rate": 9.940203404100612e-06, "loss": 6.0901, "step": 72500 }, { "epoch": 0.7, "learning_rate": 9.939725031333416e-06, "loss": 6.0914, "step": 73000 }, { "epoch": 0.7, "learning_rate": 9.939246658566221e-06, "loss": 6.0908, "step": 73500 }, { "epoch": 0.71, "learning_rate": 9.938768285799026e-06, "loss": 6.0869, "step": 74000 }, { "epoch": 0.71, "learning_rate": 9.938289913031833e-06, "loss": 6.0892, "step": 74500 }, { "epoch": 0.72, "learning_rate": 9.937811540264638e-06, "loss": 6.089, "step": 75000 }, { "epoch": 0.72, "learning_rate": 9.937333167497442e-06, "loss": 6.0836, "step": 75500 }, { "epoch": 0.73, "learning_rate": 9.936854794730246e-06, "loss": 6.0819, "step": 76000 }, { "epoch": 0.73, "learning_rate": 9.93637642196305e-06, "loss": 6.0752, "step": 76500 }, { "epoch": 0.74, "learning_rate": 9.935898049195855e-06, "loss": 6.072, "step": 77000 }, { "epoch": 0.74, "learning_rate": 9.935419676428662e-06, "loss": 6.0704, "step": 77500 }, { "epoch": 0.75, "learning_rate": 9.934941303661467e-06, "loss": 6.0523, "step": 78000 }, { "epoch": 0.75, "learning_rate": 9.934462930894271e-06, "loss": 6.0462, "step": 78500 }, { "epoch": 0.76, "learning_rate": 9.933984558127076e-06, "loss": 6.0286, "step": 79000 }, { "epoch": 0.76, "learning_rate": 9.93350618535988e-06, "loss": 6.0251, "step": 79500 }, { "epoch": 0.76, "learning_rate": 9.933027812592686e-06, "loss": 6.0155, "step": 80000 }, { "epoch": 0.77, "learning_rate": 9.932549439825491e-06, "loss": 5.9971, "step": 80500 }, { "epoch": 0.77, "learning_rate": 9.932071067058296e-06, "loss": 5.9849, "step": 81000 }, { "epoch": 0.78, "learning_rate": 9.9315926942911e-06, "loss": 5.9732, "step": 81500 }, { "epoch": 0.78, "learning_rate": 9.931114321523905e-06, "loss": 5.9741, "step": 82000 }, { "epoch": 0.79, "learning_rate": 9.93063594875671e-06, "loss": 5.9598, "step": 82500 }, { "epoch": 0.79, "learning_rate": 9.930157575989515e-06, "loss": 5.9568, "step": 83000 }, { "epoch": 0.8, "learning_rate": 9.92967920322232e-06, "loss": 5.9456, "step": 83500 }, { "epoch": 0.8, "learning_rate": 9.929200830455125e-06, "loss": 5.9409, "step": 84000 }, { "epoch": 0.81, "learning_rate": 9.92872245768793e-06, "loss": 5.9377, "step": 84500 }, { "epoch": 0.81, "learning_rate": 9.928244084920734e-06, "loss": 5.9289, "step": 85000 }, { "epoch": 0.82, "learning_rate": 9.92776571215354e-06, "loss": 5.9214, "step": 85500 }, { "epoch": 0.82, "learning_rate": 9.927287339386344e-06, "loss": 5.9178, "step": 86000 }, { "epoch": 0.83, "learning_rate": 9.926808966619149e-06, "loss": 5.91, "step": 86500 }, { "epoch": 0.83, "learning_rate": 9.926330593851954e-06, "loss": 5.9013, "step": 87000 }, { "epoch": 0.84, "learning_rate": 9.925852221084759e-06, "loss": 5.9006, "step": 87500 }, { "epoch": 0.84, "learning_rate": 9.925373848317564e-06, "loss": 5.8968, "step": 88000 }, { "epoch": 0.85, "learning_rate": 9.924895475550368e-06, "loss": 5.8952, "step": 88500 }, { "epoch": 0.85, "learning_rate": 9.924417102783173e-06, "loss": 5.8875, "step": 89000 }, { "epoch": 0.86, "learning_rate": 9.923938730015978e-06, "loss": 5.8821, "step": 89500 }, { "epoch": 0.86, "learning_rate": 9.923460357248783e-06, "loss": 5.8739, "step": 90000 }, { "epoch": 0.87, "learning_rate": 9.922981984481588e-06, "loss": 5.8687, "step": 90500 }, { "epoch": 0.87, "learning_rate": 9.922503611714394e-06, "loss": 5.8709, "step": 91000 }, { "epoch": 0.87, "learning_rate": 9.922025238947197e-06, "loss": 5.861, "step": 91500 }, { "epoch": 0.88, "learning_rate": 9.921546866180002e-06, "loss": 5.8628, "step": 92000 }, { "epoch": 0.88, "learning_rate": 9.921068493412807e-06, "loss": 5.8534, "step": 92500 }, { "epoch": 0.89, "learning_rate": 9.920590120645612e-06, "loss": 5.8476, "step": 93000 }, { "epoch": 0.89, "learning_rate": 9.920111747878419e-06, "loss": 5.8483, "step": 93500 }, { "epoch": 0.9, "learning_rate": 9.919633375111223e-06, "loss": 5.8383, "step": 94000 }, { "epoch": 0.9, "learning_rate": 9.919155002344028e-06, "loss": 5.8349, "step": 94500 }, { "epoch": 0.91, "learning_rate": 9.918676629576831e-06, "loss": 5.8298, "step": 95000 }, { "epoch": 0.91, "learning_rate": 9.918198256809636e-06, "loss": 5.8276, "step": 95500 }, { "epoch": 0.92, "learning_rate": 9.917719884042441e-06, "loss": 5.8258, "step": 96000 }, { "epoch": 0.92, "learning_rate": 9.917241511275248e-06, "loss": 5.8154, "step": 96500 }, { "epoch": 0.93, "learning_rate": 9.916763138508052e-06, "loss": 5.8143, "step": 97000 }, { "epoch": 0.93, "learning_rate": 9.916284765740857e-06, "loss": 5.8075, "step": 97500 }, { "epoch": 0.94, "learning_rate": 9.915806392973662e-06, "loss": 5.8021, "step": 98000 }, { "epoch": 0.94, "learning_rate": 9.915328020206465e-06, "loss": 5.8027, "step": 98500 }, { "epoch": 0.95, "learning_rate": 9.914849647439272e-06, "loss": 5.795, "step": 99000 }, { "epoch": 0.95, "learning_rate": 9.914371274672077e-06, "loss": 5.7893, "step": 99500 }, { "epoch": 0.96, "learning_rate": 9.913892901904882e-06, "loss": 5.7924, "step": 100000 }, { "epoch": 0.96, "learning_rate": 9.913414529137686e-06, "loss": 5.781, "step": 100500 }, { "epoch": 0.97, "learning_rate": 9.912936156370491e-06, "loss": 5.7797, "step": 101000 }, { "epoch": 0.97, "learning_rate": 9.912457783603296e-06, "loss": 5.7714, "step": 101500 }, { "epoch": 0.97, "learning_rate": 9.911979410836101e-06, "loss": 5.7704, "step": 102000 }, { "epoch": 0.98, "learning_rate": 9.911501038068906e-06, "loss": 5.7623, "step": 102500 }, { "epoch": 0.98, "learning_rate": 9.91102266530171e-06, "loss": 5.7579, "step": 103000 }, { "epoch": 0.99, "learning_rate": 9.910544292534515e-06, "loss": 5.7494, "step": 103500 }, { "epoch": 0.99, "learning_rate": 9.91006591976732e-06, "loss": 5.7505, "step": 104000 }, { "epoch": 1.0, "learning_rate": 9.909587547000125e-06, "loss": 5.7471, "step": 104500 }, { "epoch": 1.0, "learning_rate": 9.90910917423293e-06, "loss": 5.7368, "step": 105000 }, { "epoch": 1.01, "learning_rate": 9.908630801465735e-06, "loss": 5.7301, "step": 105500 }, { "epoch": 1.01, "learning_rate": 9.90815242869854e-06, "loss": 5.7361, "step": 106000 }, { "epoch": 1.02, "learning_rate": 9.907674055931345e-06, "loss": 5.728, "step": 106500 }, { "epoch": 1.02, "learning_rate": 9.90719568316415e-06, "loss": 5.7297, "step": 107000 }, { "epoch": 1.03, "learning_rate": 9.906717310396954e-06, "loss": 5.7172, "step": 107500 }, { "epoch": 1.03, "learning_rate": 9.906238937629759e-06, "loss": 5.7187, "step": 108000 }, { "epoch": 1.04, "learning_rate": 9.905760564862564e-06, "loss": 5.7126, "step": 108500 }, { "epoch": 1.04, "learning_rate": 9.905282192095369e-06, "loss": 5.7042, "step": 109000 }, { "epoch": 1.05, "learning_rate": 9.904803819328174e-06, "loss": 5.6989, "step": 109500 }, { "epoch": 1.05, "learning_rate": 9.90432544656098e-06, "loss": 5.6969, "step": 110000 }, { "epoch": 1.06, "learning_rate": 9.903847073793785e-06, "loss": 5.6875, "step": 110500 }, { "epoch": 1.06, "learning_rate": 9.903368701026588e-06, "loss": 5.6885, "step": 111000 }, { "epoch": 1.07, "learning_rate": 9.902890328259393e-06, "loss": 5.6797, "step": 111500 }, { "epoch": 1.07, "learning_rate": 9.902411955492198e-06, "loss": 5.6754, "step": 112000 }, { "epoch": 1.08, "learning_rate": 9.901933582725003e-06, "loss": 5.6757, "step": 112500 }, { "epoch": 1.08, "learning_rate": 9.90145520995781e-06, "loss": 5.6665, "step": 113000 }, { "epoch": 1.08, "learning_rate": 9.900976837190614e-06, "loss": 5.6657, "step": 113500 }, { "epoch": 1.09, "learning_rate": 9.900498464423419e-06, "loss": 5.6578, "step": 114000 }, { "epoch": 1.09, "learning_rate": 9.900020091656222e-06, "loss": 5.6503, "step": 114500 }, { "epoch": 1.1, "learning_rate": 9.899541718889027e-06, "loss": 5.6515, "step": 115000 }, { "epoch": 1.1, "learning_rate": 9.899063346121833e-06, "loss": 5.6431, "step": 115500 }, { "epoch": 1.11, "learning_rate": 9.898584973354638e-06, "loss": 5.6348, "step": 116000 }, { "epoch": 1.11, "learning_rate": 9.898106600587443e-06, "loss": 5.6287, "step": 116500 }, { "epoch": 1.12, "learning_rate": 9.897628227820248e-06, "loss": 5.6251, "step": 117000 }, { "epoch": 1.12, "learning_rate": 9.897149855053053e-06, "loss": 5.6212, "step": 117500 }, { "epoch": 1.13, "learning_rate": 9.896671482285856e-06, "loss": 5.6165, "step": 118000 }, { "epoch": 1.13, "learning_rate": 9.896193109518663e-06, "loss": 5.6058, "step": 118500 }, { "epoch": 1.14, "learning_rate": 9.895714736751467e-06, "loss": 5.6112, "step": 119000 }, { "epoch": 1.14, "learning_rate": 9.895236363984272e-06, "loss": 5.6013, "step": 119500 }, { "epoch": 1.15, "learning_rate": 9.894757991217077e-06, "loss": 5.5953, "step": 120000 }, { "epoch": 1.15, "learning_rate": 9.894279618449882e-06, "loss": 5.5951, "step": 120500 }, { "epoch": 1.16, "learning_rate": 9.893801245682687e-06, "loss": 5.5826, "step": 121000 }, { "epoch": 1.16, "learning_rate": 9.893322872915492e-06, "loss": 5.5844, "step": 121500 }, { "epoch": 1.17, "learning_rate": 9.892844500148296e-06, "loss": 5.5728, "step": 122000 }, { "epoch": 1.17, "learning_rate": 9.892366127381101e-06, "loss": 5.576, "step": 122500 }, { "epoch": 1.18, "learning_rate": 9.891887754613906e-06, "loss": 5.5647, "step": 123000 }, { "epoch": 1.18, "learning_rate": 9.891409381846711e-06, "loss": 5.5574, "step": 123500 }, { "epoch": 1.19, "learning_rate": 9.890931009079516e-06, "loss": 5.5552, "step": 124000 }, { "epoch": 1.19, "learning_rate": 9.89045263631232e-06, "loss": 5.5494, "step": 124500 }, { "epoch": 1.19, "learning_rate": 9.889974263545126e-06, "loss": 5.547, "step": 125000 }, { "epoch": 1.2, "learning_rate": 9.88949589077793e-06, "loss": 5.5358, "step": 125500 }, { "epoch": 1.2, "learning_rate": 9.889017518010735e-06, "loss": 5.5293, "step": 126000 }, { "epoch": 1.21, "learning_rate": 9.88853914524354e-06, "loss": 5.5243, "step": 126500 }, { "epoch": 1.21, "learning_rate": 9.888060772476345e-06, "loss": 5.5261, "step": 127000 }, { "epoch": 1.22, "learning_rate": 9.88758239970915e-06, "loss": 5.522, "step": 127500 }, { "epoch": 1.22, "learning_rate": 9.887104026941955e-06, "loss": 5.5088, "step": 128000 }, { "epoch": 1.23, "learning_rate": 9.88662565417476e-06, "loss": 5.5062, "step": 128500 }, { "epoch": 1.23, "learning_rate": 9.886147281407564e-06, "loss": 5.5005, "step": 129000 }, { "epoch": 1.24, "learning_rate": 9.88566890864037e-06, "loss": 5.489, "step": 129500 }, { "epoch": 1.24, "learning_rate": 9.885190535873174e-06, "loss": 5.4907, "step": 130000 }, { "epoch": 1.25, "learning_rate": 9.884712163105979e-06, "loss": 5.479, "step": 130500 }, { "epoch": 1.25, "learning_rate": 9.884233790338784e-06, "loss": 5.478, "step": 131000 }, { "epoch": 1.26, "learning_rate": 9.883755417571589e-06, "loss": 5.4671, "step": 131500 }, { "epoch": 1.26, "learning_rate": 9.883277044804395e-06, "loss": 5.4688, "step": 132000 }, { "epoch": 1.27, "learning_rate": 9.8827986720372e-06, "loss": 5.4587, "step": 132500 }, { "epoch": 1.27, "learning_rate": 9.882320299270005e-06, "loss": 5.4518, "step": 133000 }, { "epoch": 1.28, "learning_rate": 9.881841926502808e-06, "loss": 5.4391, "step": 133500 }, { "epoch": 1.28, "learning_rate": 9.881363553735613e-06, "loss": 5.4432, "step": 134000 }, { "epoch": 1.29, "learning_rate": 9.880885180968418e-06, "loss": 5.4323, "step": 134500 }, { "epoch": 1.29, "learning_rate": 9.880406808201224e-06, "loss": 5.4307, "step": 135000 }, { "epoch": 1.3, "learning_rate": 9.879928435434029e-06, "loss": 5.4226, "step": 135500 }, { "epoch": 1.3, "learning_rate": 9.879450062666834e-06, "loss": 5.4058, "step": 136000 }, { "epoch": 1.3, "learning_rate": 9.878971689899639e-06, "loss": 5.4063, "step": 136500 }, { "epoch": 1.31, "learning_rate": 9.878493317132442e-06, "loss": 5.4044, "step": 137000 }, { "epoch": 1.31, "learning_rate": 9.878014944365248e-06, "loss": 5.4, "step": 137500 }, { "epoch": 1.32, "learning_rate": 9.877536571598053e-06, "loss": 5.3789, "step": 138000 }, { "epoch": 1.32, "learning_rate": 9.877058198830858e-06, "loss": 5.371, "step": 138500 }, { "epoch": 1.33, "learning_rate": 9.876579826063663e-06, "loss": 5.3734, "step": 139000 }, { "epoch": 1.33, "learning_rate": 9.876101453296468e-06, "loss": 5.3623, "step": 139500 }, { "epoch": 1.34, "learning_rate": 9.875623080529273e-06, "loss": 5.3542, "step": 140000 }, { "epoch": 1.34, "learning_rate": 9.875144707762077e-06, "loss": 5.34, "step": 140500 }, { "epoch": 1.35, "learning_rate": 9.874666334994882e-06, "loss": 5.3289, "step": 141000 }, { "epoch": 1.35, "learning_rate": 9.874187962227687e-06, "loss": 5.3153, "step": 141500 }, { "epoch": 1.36, "learning_rate": 9.873709589460492e-06, "loss": 5.2933, "step": 142000 }, { "epoch": 1.36, "learning_rate": 9.873231216693297e-06, "loss": 5.2792, "step": 142500 }, { "epoch": 1.37, "learning_rate": 9.872752843926102e-06, "loss": 5.2455, "step": 143000 }, { "epoch": 1.37, "learning_rate": 9.872274471158907e-06, "loss": 5.2316, "step": 143500 }, { "epoch": 1.38, "learning_rate": 9.871796098391711e-06, "loss": 5.2126, "step": 144000 }, { "epoch": 1.38, "learning_rate": 9.871317725624516e-06, "loss": 5.1829, "step": 144500 }, { "epoch": 1.39, "learning_rate": 9.870839352857321e-06, "loss": 5.173, "step": 145000 }, { "epoch": 1.39, "learning_rate": 9.870360980090128e-06, "loss": 5.1552, "step": 145500 }, { "epoch": 1.4, "learning_rate": 9.86988260732293e-06, "loss": 5.1391, "step": 146000 }, { "epoch": 1.4, "learning_rate": 9.869404234555736e-06, "loss": 5.1257, "step": 146500 }, { "epoch": 1.41, "learning_rate": 9.86892586178854e-06, "loss": 5.1034, "step": 147000 }, { "epoch": 1.41, "learning_rate": 9.868447489021345e-06, "loss": 5.0826, "step": 147500 }, { "epoch": 1.41, "learning_rate": 9.86796911625415e-06, "loss": 5.0633, "step": 148000 }, { "epoch": 1.42, "learning_rate": 9.867490743486957e-06, "loss": 5.0564, "step": 148500 }, { "epoch": 1.42, "learning_rate": 9.867012370719761e-06, "loss": 5.0306, "step": 149000 }, { "epoch": 1.43, "learning_rate": 9.866533997952565e-06, "loss": 5.0163, "step": 149500 }, { "epoch": 1.43, "learning_rate": 9.86605562518537e-06, "loss": 4.9982, "step": 150000 }, { "epoch": 1.44, "learning_rate": 9.865577252418174e-06, "loss": 4.9884, "step": 150500 }, { "epoch": 1.44, "learning_rate": 9.865098879650981e-06, "loss": 4.9626, "step": 151000 }, { "epoch": 1.45, "learning_rate": 9.864620506883786e-06, "loss": 4.9384, "step": 151500 }, { "epoch": 1.45, "learning_rate": 9.86414213411659e-06, "loss": 4.9251, "step": 152000 }, { "epoch": 1.46, "learning_rate": 9.863663761349395e-06, "loss": 4.9083, "step": 152500 }, { "epoch": 1.46, "learning_rate": 9.863185388582199e-06, "loss": 4.8933, "step": 153000 }, { "epoch": 1.47, "learning_rate": 9.862707015815003e-06, "loss": 4.8731, "step": 153500 }, { "epoch": 1.47, "learning_rate": 9.86222864304781e-06, "loss": 4.8558, "step": 154000 }, { "epoch": 1.48, "learning_rate": 9.861750270280615e-06, "loss": 4.8415, "step": 154500 }, { "epoch": 1.48, "learning_rate": 9.86127189751342e-06, "loss": 4.8268, "step": 155000 }, { "epoch": 1.49, "learning_rate": 9.860793524746224e-06, "loss": 4.8135, "step": 155500 }, { "epoch": 1.49, "learning_rate": 9.86031515197903e-06, "loss": 4.7925, "step": 156000 }, { "epoch": 1.5, "learning_rate": 9.859836779211834e-06, "loss": 4.7784, "step": 156500 }, { "epoch": 1.5, "learning_rate": 9.859358406444639e-06, "loss": 4.7704, "step": 157000 }, { "epoch": 1.51, "learning_rate": 9.858880033677444e-06, "loss": 4.7592, "step": 157500 }, { "epoch": 1.51, "learning_rate": 9.858401660910249e-06, "loss": 4.7386, "step": 158000 }, { "epoch": 1.51, "learning_rate": 9.857923288143054e-06, "loss": 4.7279, "step": 158500 }, { "epoch": 1.52, "learning_rate": 9.857444915375858e-06, "loss": 4.7177, "step": 159000 }, { "epoch": 1.52, "learning_rate": 9.856966542608663e-06, "loss": 4.7034, "step": 159500 }, { "epoch": 1.53, "learning_rate": 9.856488169841468e-06, "loss": 4.6922, "step": 160000 }, { "epoch": 1.53, "learning_rate": 9.856009797074273e-06, "loss": 4.679, "step": 160500 }, { "epoch": 1.54, "learning_rate": 9.855531424307078e-06, "loss": 4.6716, "step": 161000 }, { "epoch": 1.54, "learning_rate": 9.855053051539883e-06, "loss": 4.6645, "step": 161500 }, { "epoch": 1.55, "learning_rate": 9.854574678772687e-06, "loss": 4.6504, "step": 162000 }, { "epoch": 1.55, "learning_rate": 9.854096306005492e-06, "loss": 4.6389, "step": 162500 }, { "epoch": 1.56, "learning_rate": 9.853617933238297e-06, "loss": 4.6373, "step": 163000 }, { "epoch": 1.56, "learning_rate": 9.853139560471102e-06, "loss": 4.6163, "step": 163500 }, { "epoch": 1.57, "learning_rate": 9.852661187703907e-06, "loss": 4.602, "step": 164000 }, { "epoch": 1.57, "learning_rate": 9.852182814936712e-06, "loss": 4.5961, "step": 164500 }, { "epoch": 1.58, "learning_rate": 9.851704442169517e-06, "loss": 4.5813, "step": 165000 }, { "epoch": 1.58, "learning_rate": 9.851226069402321e-06, "loss": 4.567, "step": 165500 }, { "epoch": 1.59, "learning_rate": 9.850747696635126e-06, "loss": 4.5621, "step": 166000 }, { "epoch": 1.59, "learning_rate": 9.850269323867931e-06, "loss": 4.5456, "step": 166500 }, { "epoch": 1.6, "learning_rate": 9.849790951100736e-06, "loss": 4.5373, "step": 167000 }, { "epoch": 1.6, "learning_rate": 9.849312578333542e-06, "loss": 4.5184, "step": 167500 }, { "epoch": 1.61, "learning_rate": 9.848834205566347e-06, "loss": 4.498, "step": 168000 }, { "epoch": 1.61, "learning_rate": 9.84835583279915e-06, "loss": 4.4701, "step": 168500 }, { "epoch": 1.62, "learning_rate": 9.847877460031955e-06, "loss": 4.4465, "step": 169000 }, { "epoch": 1.62, "learning_rate": 9.84739908726476e-06, "loss": 4.4072, "step": 169500 }, { "epoch": 1.62, "learning_rate": 9.846920714497565e-06, "loss": 4.3902, "step": 170000 }, { "epoch": 1.63, "learning_rate": 9.846442341730372e-06, "loss": 4.3738, "step": 170500 }, { "epoch": 1.63, "learning_rate": 9.845963968963176e-06, "loss": 4.3583, "step": 171000 }, { "epoch": 1.64, "learning_rate": 9.845485596195981e-06, "loss": 4.3435, "step": 171500 }, { "epoch": 1.64, "learning_rate": 9.845007223428784e-06, "loss": 4.322, "step": 172000 }, { "epoch": 1.65, "learning_rate": 9.84452885066159e-06, "loss": 4.3095, "step": 172500 }, { "epoch": 1.65, "learning_rate": 9.844050477894396e-06, "loss": 4.2901, "step": 173000 }, { "epoch": 1.66, "learning_rate": 9.8435721051272e-06, "loss": 4.273, "step": 173500 }, { "epoch": 1.66, "learning_rate": 9.843093732360005e-06, "loss": 4.2556, "step": 174000 }, { "epoch": 1.67, "learning_rate": 9.84261535959281e-06, "loss": 4.2453, "step": 174500 }, { "epoch": 1.67, "learning_rate": 9.842136986825615e-06, "loss": 4.2254, "step": 175000 }, { "epoch": 1.68, "learning_rate": 9.841658614058418e-06, "loss": 4.2093, "step": 175500 }, { "epoch": 1.68, "learning_rate": 9.841180241291225e-06, "loss": 4.2046, "step": 176000 }, { "epoch": 1.69, "learning_rate": 9.84070186852403e-06, "loss": 4.1857, "step": 176500 }, { "epoch": 1.69, "learning_rate": 9.840223495756835e-06, "loss": 4.1714, "step": 177000 }, { "epoch": 1.7, "learning_rate": 9.83974512298964e-06, "loss": 4.1522, "step": 177500 }, { "epoch": 1.7, "learning_rate": 9.839266750222444e-06, "loss": 4.1511, "step": 178000 }, { "epoch": 1.71, "learning_rate": 9.838788377455249e-06, "loss": 4.1253, "step": 178500 }, { "epoch": 1.71, "learning_rate": 9.838310004688054e-06, "loss": 4.1197, "step": 179000 }, { "epoch": 1.72, "learning_rate": 9.837831631920859e-06, "loss": 4.1028, "step": 179500 }, { "epoch": 1.72, "learning_rate": 9.837353259153664e-06, "loss": 4.0855, "step": 180000 }, { "epoch": 1.73, "learning_rate": 9.836874886386468e-06, "loss": 4.0793, "step": 180500 }, { "epoch": 1.73, "learning_rate": 9.836396513619273e-06, "loss": 4.0672, "step": 181000 }, { "epoch": 1.73, "learning_rate": 9.835918140852078e-06, "loss": 4.0536, "step": 181500 }, { "epoch": 1.74, "learning_rate": 9.835439768084883e-06, "loss": 4.0404, "step": 182000 }, { "epoch": 1.74, "learning_rate": 9.834961395317688e-06, "loss": 4.0236, "step": 182500 }, { "epoch": 1.75, "learning_rate": 9.834483022550493e-06, "loss": 4.0184, "step": 183000 }, { "epoch": 1.75, "learning_rate": 9.834004649783298e-06, "loss": 4.0002, "step": 183500 }, { "epoch": 1.76, "learning_rate": 9.833526277016104e-06, "loss": 3.9823, "step": 184000 }, { "epoch": 1.76, "learning_rate": 9.833047904248907e-06, "loss": 3.9778, "step": 184500 }, { "epoch": 1.77, "learning_rate": 9.832569531481712e-06, "loss": 3.96, "step": 185000 }, { "epoch": 1.77, "learning_rate": 9.832091158714517e-06, "loss": 3.9406, "step": 185500 }, { "epoch": 1.78, "learning_rate": 9.831612785947322e-06, "loss": 3.9228, "step": 186000 }, { "epoch": 1.78, "learning_rate": 9.831134413180127e-06, "loss": 3.9115, "step": 186500 }, { "epoch": 1.79, "learning_rate": 9.830656040412933e-06, "loss": 3.8995, "step": 187000 }, { "epoch": 1.79, "learning_rate": 9.830177667645738e-06, "loss": 3.8848, "step": 187500 }, { "epoch": 1.8, "learning_rate": 9.829699294878541e-06, "loss": 3.8814, "step": 188000 }, { "epoch": 1.8, "learning_rate": 9.829220922111346e-06, "loss": 3.8569, "step": 188500 }, { "epoch": 1.81, "learning_rate": 9.828742549344151e-06, "loss": 3.8503, "step": 189000 }, { "epoch": 1.81, "learning_rate": 9.828264176576957e-06, "loss": 3.8414, "step": 189500 }, { "epoch": 1.82, "learning_rate": 9.827785803809762e-06, "loss": 3.8223, "step": 190000 }, { "epoch": 1.82, "learning_rate": 9.827307431042567e-06, "loss": 3.8167, "step": 190500 }, { "epoch": 1.83, "learning_rate": 9.826829058275372e-06, "loss": 3.7987, "step": 191000 }, { "epoch": 1.83, "learning_rate": 9.826350685508175e-06, "loss": 3.7885, "step": 191500 }, { "epoch": 1.84, "learning_rate": 9.825872312740982e-06, "loss": 3.7784, "step": 192000 }, { "epoch": 1.84, "learning_rate": 9.825393939973786e-06, "loss": 3.7693, "step": 192500 }, { "epoch": 1.84, "learning_rate": 9.824915567206591e-06, "loss": 3.7579, "step": 193000 }, { "epoch": 1.85, "learning_rate": 9.824437194439396e-06, "loss": 3.749, "step": 193500 }, { "epoch": 1.85, "learning_rate": 9.823958821672201e-06, "loss": 3.7342, "step": 194000 }, { "epoch": 1.86, "learning_rate": 9.823480448905006e-06, "loss": 3.7268, "step": 194500 }, { "epoch": 1.86, "learning_rate": 9.82300207613781e-06, "loss": 3.7171, "step": 195000 }, { "epoch": 1.87, "learning_rate": 9.822523703370616e-06, "loss": 3.7119, "step": 195500 }, { "epoch": 1.87, "learning_rate": 9.82204533060342e-06, "loss": 3.6949, "step": 196000 }, { "epoch": 1.88, "learning_rate": 9.821566957836225e-06, "loss": 3.6864, "step": 196500 }, { "epoch": 1.88, "learning_rate": 9.82108858506903e-06, "loss": 3.6783, "step": 197000 }, { "epoch": 1.89, "learning_rate": 9.820610212301835e-06, "loss": 3.6633, "step": 197500 }, { "epoch": 1.89, "learning_rate": 9.82013183953464e-06, "loss": 3.6593, "step": 198000 }, { "epoch": 1.9, "learning_rate": 9.819653466767445e-06, "loss": 3.6514, "step": 198500 }, { "epoch": 1.9, "learning_rate": 9.81917509400025e-06, "loss": 3.6486, "step": 199000 }, { "epoch": 1.91, "learning_rate": 9.818696721233054e-06, "loss": 3.6347, "step": 199500 }, { "epoch": 1.91, "learning_rate": 9.81821834846586e-06, "loss": 3.6218, "step": 200000 }, { "epoch": 1.92, "learning_rate": 9.817739975698664e-06, "loss": 3.6248, "step": 200500 }, { "epoch": 1.92, "learning_rate": 9.817261602931469e-06, "loss": 3.6132, "step": 201000 }, { "epoch": 1.93, "learning_rate": 9.816783230164274e-06, "loss": 3.609, "step": 201500 }, { "epoch": 1.93, "learning_rate": 9.816304857397079e-06, "loss": 3.6019, "step": 202000 }, { "epoch": 1.94, "learning_rate": 9.815826484629883e-06, "loss": 3.5913, "step": 202500 }, { "epoch": 1.94, "learning_rate": 9.81534811186269e-06, "loss": 3.5817, "step": 203000 }, { "epoch": 1.95, "learning_rate": 9.814869739095493e-06, "loss": 3.573, "step": 203500 }, { "epoch": 1.95, "learning_rate": 9.814391366328298e-06, "loss": 3.5647, "step": 204000 }, { "epoch": 1.95, "learning_rate": 9.813912993561103e-06, "loss": 3.5592, "step": 204500 }, { "epoch": 1.96, "learning_rate": 9.813434620793908e-06, "loss": 3.5478, "step": 205000 }, { "epoch": 1.96, "learning_rate": 9.812956248026712e-06, "loss": 3.5439, "step": 205500 }, { "epoch": 1.97, "learning_rate": 9.812477875259519e-06, "loss": 3.5353, "step": 206000 }, { "epoch": 1.97, "learning_rate": 9.811999502492324e-06, "loss": 3.5265, "step": 206500 }, { "epoch": 1.98, "learning_rate": 9.811521129725127e-06, "loss": 3.5218, "step": 207000 }, { "epoch": 1.98, "learning_rate": 9.811042756957932e-06, "loss": 3.525, "step": 207500 }, { "epoch": 1.99, "learning_rate": 9.810564384190737e-06, "loss": 3.51, "step": 208000 }, { "epoch": 1.99, "learning_rate": 9.810086011423543e-06, "loss": 3.5128, "step": 208500 }, { "epoch": 2.0, "learning_rate": 9.809607638656348e-06, "loss": 3.5048, "step": 209000 }, { "epoch": 2.0, "learning_rate": 9.809129265889153e-06, "loss": 3.4891, "step": 209500 }, { "epoch": 2.01, "learning_rate": 9.808650893121958e-06, "loss": 3.488, "step": 210000 }, { "epoch": 2.01, "learning_rate": 9.808172520354761e-06, "loss": 3.4773, "step": 210500 }, { "epoch": 2.02, "learning_rate": 9.807694147587566e-06, "loss": 3.4804, "step": 211000 }, { "epoch": 2.02, "learning_rate": 9.807215774820372e-06, "loss": 3.4672, "step": 211500 }, { "epoch": 2.03, "learning_rate": 9.806737402053177e-06, "loss": 3.4676, "step": 212000 }, { "epoch": 2.03, "learning_rate": 9.806259029285982e-06, "loss": 3.4596, "step": 212500 }, { "epoch": 2.04, "learning_rate": 9.805780656518787e-06, "loss": 3.4557, "step": 213000 }, { "epoch": 2.04, "learning_rate": 9.805302283751592e-06, "loss": 3.4492, "step": 213500 }, { "epoch": 2.05, "learning_rate": 9.804823910984397e-06, "loss": 3.4386, "step": 214000 }, { "epoch": 2.05, "learning_rate": 9.804345538217201e-06, "loss": 3.4286, "step": 214500 }, { "epoch": 2.06, "learning_rate": 9.803867165450006e-06, "loss": 3.4339, "step": 215000 }, { "epoch": 2.06, "learning_rate": 9.803388792682811e-06, "loss": 3.4243, "step": 215500 }, { "epoch": 2.06, "learning_rate": 9.802910419915616e-06, "loss": 3.4195, "step": 216000 }, { "epoch": 2.07, "learning_rate": 9.80243204714842e-06, "loss": 3.4109, "step": 216500 }, { "epoch": 2.07, "learning_rate": 9.801953674381226e-06, "loss": 3.4151, "step": 217000 }, { "epoch": 2.08, "learning_rate": 9.80147530161403e-06, "loss": 3.3984, "step": 217500 }, { "epoch": 2.08, "learning_rate": 9.800996928846835e-06, "loss": 3.3939, "step": 218000 }, { "epoch": 2.09, "learning_rate": 9.80051855607964e-06, "loss": 3.3947, "step": 218500 }, { "epoch": 2.09, "learning_rate": 9.800040183312445e-06, "loss": 3.3851, "step": 219000 }, { "epoch": 2.1, "learning_rate": 9.79956181054525e-06, "loss": 3.3826, "step": 219500 }, { "epoch": 2.1, "learning_rate": 9.799083437778055e-06, "loss": 3.3843, "step": 220000 }, { "epoch": 2.11, "learning_rate": 9.79860506501086e-06, "loss": 3.3843, "step": 220500 }, { "epoch": 2.11, "learning_rate": 9.798126692243664e-06, "loss": 3.3693, "step": 221000 }, { "epoch": 2.12, "learning_rate": 9.79764831947647e-06, "loss": 3.3606, "step": 221500 }, { "epoch": 2.12, "learning_rate": 9.797169946709274e-06, "loss": 3.3608, "step": 222000 }, { "epoch": 2.13, "learning_rate": 9.79669157394208e-06, "loss": 3.3499, "step": 222500 }, { "epoch": 2.13, "learning_rate": 9.796213201174884e-06, "loss": 3.3608, "step": 223000 }, { "epoch": 2.14, "learning_rate": 9.795734828407689e-06, "loss": 3.3523, "step": 223500 }, { "epoch": 2.14, "learning_rate": 9.795256455640493e-06, "loss": 3.3466, "step": 224000 }, { "epoch": 2.15, "learning_rate": 9.794778082873298e-06, "loss": 3.3418, "step": 224500 }, { "epoch": 2.15, "learning_rate": 9.794299710106105e-06, "loss": 3.3397, "step": 225000 }, { "epoch": 2.16, "learning_rate": 9.79382133733891e-06, "loss": 3.3338, "step": 225500 }, { "epoch": 2.16, "learning_rate": 9.793342964571715e-06, "loss": 3.3287, "step": 226000 }, { "epoch": 2.16, "learning_rate": 9.792864591804518e-06, "loss": 3.3269, "step": 226500 }, { "epoch": 2.17, "learning_rate": 9.792386219037323e-06, "loss": 3.3226, "step": 227000 }, { "epoch": 2.17, "learning_rate": 9.791907846270127e-06, "loss": 3.3219, "step": 227500 }, { "epoch": 2.18, "learning_rate": 9.791429473502934e-06, "loss": 3.3137, "step": 228000 }, { "epoch": 2.18, "learning_rate": 9.790951100735739e-06, "loss": 3.317, "step": 228500 }, { "epoch": 2.19, "learning_rate": 9.790472727968544e-06, "loss": 3.3051, "step": 229000 }, { "epoch": 2.19, "learning_rate": 9.789994355201348e-06, "loss": 3.3009, "step": 229500 }, { "epoch": 2.2, "learning_rate": 9.789515982434152e-06, "loss": 3.3025, "step": 230000 }, { "epoch": 2.2, "learning_rate": 9.789037609666958e-06, "loss": 3.2909, "step": 230500 }, { "epoch": 2.21, "learning_rate": 9.788559236899763e-06, "loss": 3.2882, "step": 231000 }, { "epoch": 2.21, "learning_rate": 9.788080864132568e-06, "loss": 3.3115, "step": 231500 }, { "epoch": 2.22, "learning_rate": 9.787602491365373e-06, "loss": 3.2912, "step": 232000 }, { "epoch": 2.22, "learning_rate": 9.787124118598178e-06, "loss": 3.2843, "step": 232500 }, { "epoch": 2.23, "learning_rate": 9.786645745830982e-06, "loss": 3.2775, "step": 233000 }, { "epoch": 2.23, "learning_rate": 9.786167373063787e-06, "loss": 3.2703, "step": 233500 }, { "epoch": 2.24, "learning_rate": 9.785689000296592e-06, "loss": 3.266, "step": 234000 }, { "epoch": 2.24, "learning_rate": 9.785210627529397e-06, "loss": 3.2663, "step": 234500 }, { "epoch": 2.25, "learning_rate": 9.784732254762202e-06, "loss": 3.2643, "step": 235000 }, { "epoch": 2.25, "learning_rate": 9.784253881995007e-06, "loss": 3.2597, "step": 235500 }, { "epoch": 2.26, "learning_rate": 9.783775509227811e-06, "loss": 3.2553, "step": 236000 }, { "epoch": 2.26, "learning_rate": 9.783297136460616e-06, "loss": 3.2501, "step": 236500 }, { "epoch": 2.27, "learning_rate": 9.782818763693421e-06, "loss": 3.248, "step": 237000 }, { "epoch": 2.27, "learning_rate": 9.782340390926226e-06, "loss": 3.244, "step": 237500 }, { "epoch": 2.27, "learning_rate": 9.78186201815903e-06, "loss": 3.2516, "step": 238000 }, { "epoch": 2.28, "learning_rate": 9.781383645391836e-06, "loss": 3.2435, "step": 238500 }, { "epoch": 2.28, "learning_rate": 9.78090527262464e-06, "loss": 3.2346, "step": 239000 }, { "epoch": 2.29, "learning_rate": 9.780426899857445e-06, "loss": 3.2347, "step": 239500 }, { "epoch": 2.29, "learning_rate": 9.77994852709025e-06, "loss": 3.2241, "step": 240000 }, { "epoch": 2.3, "learning_rate": 9.779470154323055e-06, "loss": 3.2258, "step": 240500 }, { "epoch": 2.3, "learning_rate": 9.77899178155586e-06, "loss": 3.2314, "step": 241000 }, { "epoch": 2.31, "learning_rate": 9.778513408788666e-06, "loss": 3.2194, "step": 241500 }, { "epoch": 2.31, "learning_rate": 9.77803503602147e-06, "loss": 3.2142, "step": 242000 }, { "epoch": 2.32, "learning_rate": 9.777556663254274e-06, "loss": 3.2074, "step": 242500 }, { "epoch": 2.32, "learning_rate": 9.77707829048708e-06, "loss": 3.2095, "step": 243000 }, { "epoch": 2.33, "learning_rate": 9.776599917719884e-06, "loss": 3.2066, "step": 243500 }, { "epoch": 2.33, "learning_rate": 9.77612154495269e-06, "loss": 3.2065, "step": 244000 }, { "epoch": 2.34, "learning_rate": 9.775643172185496e-06, "loss": 3.2076, "step": 244500 }, { "epoch": 2.34, "learning_rate": 9.7751647994183e-06, "loss": 3.2041, "step": 245000 }, { "epoch": 2.35, "learning_rate": 9.774686426651104e-06, "loss": 3.2041, "step": 245500 }, { "epoch": 2.35, "learning_rate": 9.774208053883908e-06, "loss": 3.1935, "step": 246000 }, { "epoch": 2.36, "learning_rate": 9.773729681116713e-06, "loss": 3.1916, "step": 246500 }, { "epoch": 2.36, "learning_rate": 9.77325130834952e-06, "loss": 3.1816, "step": 247000 }, { "epoch": 2.37, "learning_rate": 9.772772935582325e-06, "loss": 3.1831, "step": 247500 }, { "epoch": 2.37, "learning_rate": 9.77229456281513e-06, "loss": 3.1843, "step": 248000 }, { "epoch": 2.38, "learning_rate": 9.771816190047934e-06, "loss": 3.1754, "step": 248500 }, { "epoch": 2.38, "learning_rate": 9.771337817280737e-06, "loss": 3.1695, "step": 249000 }, { "epoch": 2.38, "learning_rate": 9.770859444513544e-06, "loss": 3.1748, "step": 249500 }, { "epoch": 2.39, "learning_rate": 9.770381071746349e-06, "loss": 3.1695, "step": 250000 }, { "epoch": 2.39, "learning_rate": 9.769902698979154e-06, "loss": 3.1735, "step": 250500 }, { "epoch": 2.4, "learning_rate": 9.769424326211959e-06, "loss": 3.169, "step": 251000 }, { "epoch": 2.4, "learning_rate": 9.768945953444763e-06, "loss": 3.1636, "step": 251500 }, { "epoch": 2.41, "learning_rate": 9.768467580677568e-06, "loss": 3.1627, "step": 252000 }, { "epoch": 2.41, "learning_rate": 9.767989207910373e-06, "loss": 3.1511, "step": 252500 }, { "epoch": 2.42, "learning_rate": 9.767510835143178e-06, "loss": 3.1578, "step": 253000 }, { "epoch": 2.42, "learning_rate": 9.767032462375983e-06, "loss": 3.147, "step": 253500 }, { "epoch": 2.43, "learning_rate": 9.766554089608788e-06, "loss": 3.1474, "step": 254000 }, { "epoch": 2.43, "learning_rate": 9.766075716841592e-06, "loss": 3.1468, "step": 254500 }, { "epoch": 2.44, "learning_rate": 9.765597344074397e-06, "loss": 3.1418, "step": 255000 }, { "epoch": 2.44, "learning_rate": 9.765118971307202e-06, "loss": 3.1387, "step": 255500 }, { "epoch": 2.45, "learning_rate": 9.764640598540007e-06, "loss": 3.1375, "step": 256000 }, { "epoch": 2.45, "learning_rate": 9.764162225772812e-06, "loss": 3.1357, "step": 256500 }, { "epoch": 2.46, "learning_rate": 9.763683853005617e-06, "loss": 3.1372, "step": 257000 }, { "epoch": 2.46, "learning_rate": 9.763205480238422e-06, "loss": 3.1248, "step": 257500 }, { "epoch": 2.47, "learning_rate": 9.762727107471226e-06, "loss": 3.1282, "step": 258000 }, { "epoch": 2.47, "learning_rate": 9.762248734704031e-06, "loss": 3.1229, "step": 258500 }, { "epoch": 2.48, "learning_rate": 9.761770361936836e-06, "loss": 3.1246, "step": 259000 }, { "epoch": 2.48, "learning_rate": 9.761291989169641e-06, "loss": 3.125, "step": 259500 }, { "epoch": 2.49, "learning_rate": 9.760813616402446e-06, "loss": 3.1186, "step": 260000 }, { "epoch": 2.49, "learning_rate": 9.760335243635252e-06, "loss": 3.117, "step": 260500 }, { "epoch": 2.49, "learning_rate": 9.759856870868057e-06, "loss": 3.1153, "step": 261000 }, { "epoch": 2.5, "learning_rate": 9.75937849810086e-06, "loss": 3.1086, "step": 261500 }, { "epoch": 2.5, "learning_rate": 9.758900125333665e-06, "loss": 3.1088, "step": 262000 }, { "epoch": 2.51, "learning_rate": 9.75842175256647e-06, "loss": 3.1068, "step": 262500 }, { "epoch": 2.51, "learning_rate": 9.757943379799275e-06, "loss": 3.1043, "step": 263000 }, { "epoch": 2.52, "learning_rate": 9.757465007032081e-06, "loss": 3.0966, "step": 263500 }, { "epoch": 2.52, "learning_rate": 9.756986634264886e-06, "loss": 3.0912, "step": 264000 }, { "epoch": 2.53, "learning_rate": 9.756508261497691e-06, "loss": 3.1003, "step": 264500 }, { "epoch": 2.53, "learning_rate": 9.756029888730494e-06, "loss": 3.1014, "step": 265000 }, { "epoch": 2.54, "learning_rate": 9.755551515963299e-06, "loss": 3.0923, "step": 265500 }, { "epoch": 2.54, "learning_rate": 9.755073143196106e-06, "loss": 3.0817, "step": 266000 }, { "epoch": 2.55, "learning_rate": 9.75459477042891e-06, "loss": 3.0916, "step": 266500 }, { "epoch": 2.55, "learning_rate": 9.754116397661715e-06, "loss": 3.0818, "step": 267000 }, { "epoch": 2.56, "learning_rate": 9.75363802489452e-06, "loss": 3.0799, "step": 267500 }, { "epoch": 2.56, "learning_rate": 9.753159652127325e-06, "loss": 3.0801, "step": 268000 }, { "epoch": 2.57, "learning_rate": 9.752681279360128e-06, "loss": 3.0798, "step": 268500 }, { "epoch": 2.57, "learning_rate": 9.752202906592935e-06, "loss": 3.0771, "step": 269000 }, { "epoch": 2.58, "learning_rate": 9.75172453382574e-06, "loss": 3.0693, "step": 269500 }, { "epoch": 2.58, "learning_rate": 9.751246161058544e-06, "loss": 3.0729, "step": 270000 }, { "epoch": 2.59, "learning_rate": 9.75076778829135e-06, "loss": 3.0695, "step": 270500 }, { "epoch": 2.59, "learning_rate": 9.750289415524154e-06, "loss": 3.0696, "step": 271000 }, { "epoch": 2.6, "learning_rate": 9.749811042756959e-06, "loss": 3.0669, "step": 271500 }, { "epoch": 2.6, "learning_rate": 9.749332669989764e-06, "loss": 3.0662, "step": 272000 }, { "epoch": 2.6, "learning_rate": 9.748854297222569e-06, "loss": 3.0653, "step": 272500 }, { "epoch": 2.61, "learning_rate": 9.748375924455373e-06, "loss": 3.0693, "step": 273000 }, { "epoch": 2.61, "learning_rate": 9.747897551688178e-06, "loss": 3.0703, "step": 273500 }, { "epoch": 2.62, "learning_rate": 9.747419178920983e-06, "loss": 3.0682, "step": 274000 }, { "epoch": 2.62, "learning_rate": 9.746940806153788e-06, "loss": 3.0575, "step": 274500 }, { "epoch": 2.63, "learning_rate": 9.746462433386593e-06, "loss": 3.069, "step": 275000 }, { "epoch": 2.63, "learning_rate": 9.745984060619398e-06, "loss": 3.0589, "step": 275500 }, { "epoch": 2.64, "learning_rate": 9.745505687852202e-06, "loss": 3.0558, "step": 276000 }, { "epoch": 2.64, "learning_rate": 9.745027315085007e-06, "loss": 3.0538, "step": 276500 }, { "epoch": 2.65, "learning_rate": 9.744548942317812e-06, "loss": 3.0519, "step": 277000 }, { "epoch": 2.65, "learning_rate": 9.744070569550617e-06, "loss": 3.0332, "step": 277500 }, { "epoch": 2.66, "learning_rate": 9.743592196783422e-06, "loss": 3.0355, "step": 278000 }, { "epoch": 2.66, "learning_rate": 9.743113824016227e-06, "loss": 3.0314, "step": 278500 }, { "epoch": 2.67, "learning_rate": 9.742635451249032e-06, "loss": 3.0327, "step": 279000 }, { "epoch": 2.67, "learning_rate": 9.742157078481836e-06, "loss": 3.0353, "step": 279500 }, { "epoch": 2.68, "learning_rate": 9.741678705714643e-06, "loss": 3.0259, "step": 280000 }, { "epoch": 2.68, "learning_rate": 9.741200332947446e-06, "loss": 3.0278, "step": 280500 }, { "epoch": 2.69, "learning_rate": 9.740721960180251e-06, "loss": 3.0294, "step": 281000 }, { "epoch": 2.69, "learning_rate": 9.740243587413056e-06, "loss": 3.0198, "step": 281500 }, { "epoch": 2.7, "learning_rate": 9.73976521464586e-06, "loss": 3.0255, "step": 282000 }, { "epoch": 2.7, "learning_rate": 9.739286841878667e-06, "loss": 3.0226, "step": 282500 }, { "epoch": 2.71, "learning_rate": 9.738808469111472e-06, "loss": 3.0167, "step": 283000 }, { "epoch": 2.71, "learning_rate": 9.738330096344277e-06, "loss": 3.0161, "step": 283500 }, { "epoch": 2.71, "learning_rate": 9.73785172357708e-06, "loss": 3.0121, "step": 284000 }, { "epoch": 2.72, "learning_rate": 9.737373350809885e-06, "loss": 3.0108, "step": 284500 }, { "epoch": 2.72, "learning_rate": 9.73689497804269e-06, "loss": 3.0089, "step": 285000 }, { "epoch": 2.73, "learning_rate": 9.736416605275496e-06, "loss": 3.0062, "step": 285500 }, { "epoch": 2.73, "learning_rate": 9.735938232508301e-06, "loss": 3.0031, "step": 286000 }, { "epoch": 2.74, "learning_rate": 9.735459859741106e-06, "loss": 3.0016, "step": 286500 }, { "epoch": 2.74, "learning_rate": 9.73498148697391e-06, "loss": 2.9938, "step": 287000 }, { "epoch": 2.75, "learning_rate": 9.734503114206714e-06, "loss": 3.0036, "step": 287500 }, { "epoch": 2.75, "learning_rate": 9.73402474143952e-06, "loss": 2.9998, "step": 288000 }, { "epoch": 2.76, "learning_rate": 9.733546368672325e-06, "loss": 2.9984, "step": 288500 }, { "epoch": 2.76, "learning_rate": 9.73306799590513e-06, "loss": 2.9847, "step": 289000 }, { "epoch": 2.77, "learning_rate": 9.732589623137935e-06, "loss": 2.99, "step": 289500 }, { "epoch": 2.77, "learning_rate": 9.73211125037074e-06, "loss": 2.9908, "step": 290000 }, { "epoch": 2.78, "learning_rate": 9.731632877603545e-06, "loss": 2.9885, "step": 290500 }, { "epoch": 2.78, "learning_rate": 9.73115450483635e-06, "loss": 2.9847, "step": 291000 }, { "epoch": 2.79, "learning_rate": 9.730676132069154e-06, "loss": 2.981, "step": 291500 }, { "epoch": 2.79, "learning_rate": 9.73019775930196e-06, "loss": 2.982, "step": 292000 }, { "epoch": 2.8, "learning_rate": 9.729719386534764e-06, "loss": 2.9808, "step": 292500 }, { "epoch": 2.8, "learning_rate": 9.729241013767569e-06, "loss": 2.9759, "step": 293000 }, { "epoch": 2.81, "learning_rate": 9.728762641000374e-06, "loss": 2.9751, "step": 293500 }, { "epoch": 2.81, "learning_rate": 9.728284268233179e-06, "loss": 2.9722, "step": 294000 }, { "epoch": 2.81, "learning_rate": 9.727805895465983e-06, "loss": 2.9685, "step": 294500 }, { "epoch": 2.82, "learning_rate": 9.727327522698788e-06, "loss": 2.9729, "step": 295000 }, { "epoch": 2.82, "learning_rate": 9.726849149931593e-06, "loss": 2.9698, "step": 295500 }, { "epoch": 2.83, "learning_rate": 9.7263707771644e-06, "loss": 2.9677, "step": 296000 }, { "epoch": 2.83, "learning_rate": 9.725892404397203e-06, "loss": 2.9629, "step": 296500 }, { "epoch": 2.84, "learning_rate": 9.725414031630008e-06, "loss": 2.9602, "step": 297000 }, { "epoch": 2.84, "learning_rate": 9.724935658862813e-06, "loss": 2.966, "step": 297500 }, { "epoch": 2.85, "learning_rate": 9.724457286095617e-06, "loss": 2.9528, "step": 298000 }, { "epoch": 2.85, "learning_rate": 9.723978913328422e-06, "loss": 2.9593, "step": 298500 }, { "epoch": 2.86, "learning_rate": 9.723500540561229e-06, "loss": 2.9592, "step": 299000 }, { "epoch": 2.86, "learning_rate": 9.723022167794034e-06, "loss": 2.9523, "step": 299500 }, { "epoch": 2.87, "learning_rate": 9.722543795026837e-06, "loss": 2.9496, "step": 300000 }, { "epoch": 2.87, "learning_rate": 9.722065422259642e-06, "loss": 2.9462, "step": 300500 }, { "epoch": 2.88, "learning_rate": 9.721587049492446e-06, "loss": 2.9482, "step": 301000 }, { "epoch": 2.88, "learning_rate": 9.721108676725253e-06, "loss": 2.9416, "step": 301500 }, { "epoch": 2.89, "learning_rate": 9.720630303958058e-06, "loss": 2.9457, "step": 302000 }, { "epoch": 2.89, "learning_rate": 9.720151931190863e-06, "loss": 2.9406, "step": 302500 }, { "epoch": 2.9, "learning_rate": 9.719673558423668e-06, "loss": 2.9373, "step": 303000 }, { "epoch": 2.9, "learning_rate": 9.71919518565647e-06, "loss": 2.9441, "step": 303500 }, { "epoch": 2.91, "learning_rate": 9.718716812889276e-06, "loss": 2.933, "step": 304000 }, { "epoch": 2.91, "learning_rate": 9.718238440122082e-06, "loss": 2.9306, "step": 304500 }, { "epoch": 2.92, "learning_rate": 9.717760067354887e-06, "loss": 2.9301, "step": 305000 }, { "epoch": 2.92, "learning_rate": 9.717281694587692e-06, "loss": 2.9353, "step": 305500 }, { "epoch": 2.92, "learning_rate": 9.716803321820497e-06, "loss": 2.936, "step": 306000 }, { "epoch": 2.93, "learning_rate": 9.716324949053301e-06, "loss": 2.9199, "step": 306500 }, { "epoch": 2.93, "learning_rate": 9.715846576286106e-06, "loss": 2.9242, "step": 307000 }, { "epoch": 2.94, "learning_rate": 9.715368203518911e-06, "loss": 2.9255, "step": 307500 }, { "epoch": 2.94, "learning_rate": 9.714889830751716e-06, "loss": 2.9275, "step": 308000 }, { "epoch": 2.95, "learning_rate": 9.714411457984521e-06, "loss": 2.921, "step": 308500 }, { "epoch": 2.95, "learning_rate": 9.713933085217326e-06, "loss": 2.9191, "step": 309000 }, { "epoch": 2.96, "learning_rate": 9.71345471245013e-06, "loss": 2.919, "step": 309500 }, { "epoch": 2.96, "learning_rate": 9.712976339682935e-06, "loss": 2.9155, "step": 310000 }, { "epoch": 2.97, "learning_rate": 9.71249796691574e-06, "loss": 2.9097, "step": 310500 }, { "epoch": 2.97, "learning_rate": 9.712019594148545e-06, "loss": 2.912, "step": 311000 }, { "epoch": 2.98, "learning_rate": 9.71154122138135e-06, "loss": 2.9149, "step": 311500 }, { "epoch": 2.98, "learning_rate": 9.711062848614155e-06, "loss": 2.9121, "step": 312000 }, { "epoch": 2.99, "learning_rate": 9.71058447584696e-06, "loss": 2.908, "step": 312500 }, { "epoch": 2.99, "learning_rate": 9.710106103079764e-06, "loss": 2.9095, "step": 313000 }, { "epoch": 3.0, "learning_rate": 9.70962773031257e-06, "loss": 2.9017, "step": 313500 }, { "epoch": 3.0, "learning_rate": 9.709149357545374e-06, "loss": 2.9028, "step": 314000 }, { "epoch": 3.01, "learning_rate": 9.708670984778179e-06, "loss": 2.9047, "step": 314500 }, { "epoch": 3.01, "learning_rate": 9.708192612010984e-06, "loss": 2.9059, "step": 315000 }, { "epoch": 3.02, "learning_rate": 9.707714239243789e-06, "loss": 2.8937, "step": 315500 }, { "epoch": 3.02, "learning_rate": 9.707235866476594e-06, "loss": 2.8916, "step": 316000 }, { "epoch": 3.02, "step": 316291, "total_flos": 2.335326550817243e+18, "train_loss": 4.694392849819643, "train_runtime": 197999.2569, "train_samples_per_second": 2958.98, "train_steps_per_second": 52.839 } ], "logging_steps": 500, "max_steps": 10462100, "num_train_epochs": 100, "save_steps": 1000000, "total_flos": 2.335326550817243e+18, "trial_name": null, "trial_params": null }