{ "best_metric": 0.33534759283065796, "best_model_checkpoint": "vit-base-patch16-224-in21k-bridgedefectVIT\\checkpoint-16605", "epoch": 2.0, "eval_steps": 500, "global_step": 16605, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 1.2045290291496025e-08, "loss": 1.618, "step": 10 }, { "epoch": 0.0, "learning_rate": 2.409058058299205e-08, "loss": 1.643, "step": 20 }, { "epoch": 0.0, "learning_rate": 3.6135870874488076e-08, "loss": 1.6249, "step": 30 }, { "epoch": 0.0, "learning_rate": 4.81811611659841e-08, "loss": 1.6316, "step": 40 }, { "epoch": 0.01, "learning_rate": 6.022645145748012e-08, "loss": 1.6223, "step": 50 }, { "epoch": 0.01, "learning_rate": 7.227174174897615e-08, "loss": 1.6194, "step": 60 }, { "epoch": 0.01, "learning_rate": 8.431703204047217e-08, "loss": 1.6223, "step": 70 }, { "epoch": 0.01, "learning_rate": 9.63623223319682e-08, "loss": 1.6369, "step": 80 }, { "epoch": 0.01, "learning_rate": 1.0840761262346423e-07, "loss": 1.6251, "step": 90 }, { "epoch": 0.01, "learning_rate": 1.2045290291496024e-07, "loss": 1.6344, "step": 100 }, { "epoch": 0.01, "learning_rate": 1.3249819320645627e-07, "loss": 1.6342, "step": 110 }, { "epoch": 0.01, "learning_rate": 1.445434834979523e-07, "loss": 1.6207, "step": 120 }, { "epoch": 0.02, "learning_rate": 1.5658877378944834e-07, "loss": 1.6193, "step": 130 }, { "epoch": 0.02, "learning_rate": 1.6863406408094434e-07, "loss": 1.6204, "step": 140 }, { "epoch": 0.02, "learning_rate": 1.8067935437244037e-07, "loss": 1.613, "step": 150 }, { "epoch": 0.02, "learning_rate": 1.927246446639364e-07, "loss": 1.612, "step": 160 }, { "epoch": 0.02, "learning_rate": 2.0476993495543244e-07, "loss": 1.6379, "step": 170 }, { "epoch": 0.02, "learning_rate": 2.1681522524692847e-07, "loss": 1.6225, "step": 180 }, { "epoch": 0.02, "learning_rate": 2.288605155384245e-07, "loss": 1.6313, "step": 190 }, { "epoch": 0.02, "learning_rate": 2.409058058299205e-07, "loss": 1.6287, "step": 200 }, { "epoch": 0.03, "learning_rate": 2.529510961214165e-07, "loss": 1.6071, "step": 210 }, { "epoch": 0.03, "learning_rate": 2.6499638641291254e-07, "loss": 1.6256, "step": 220 }, { "epoch": 0.03, "learning_rate": 2.770416767044086e-07, "loss": 1.6243, "step": 230 }, { "epoch": 0.03, "learning_rate": 2.890869669959046e-07, "loss": 1.6133, "step": 240 }, { "epoch": 0.03, "learning_rate": 3.0113225728740064e-07, "loss": 1.6262, "step": 250 }, { "epoch": 0.03, "learning_rate": 3.1317754757889667e-07, "loss": 1.6096, "step": 260 }, { "epoch": 0.03, "learning_rate": 3.252228378703927e-07, "loss": 1.6154, "step": 270 }, { "epoch": 0.03, "learning_rate": 3.372681281618887e-07, "loss": 1.628, "step": 280 }, { "epoch": 0.03, "learning_rate": 3.493134184533847e-07, "loss": 1.6236, "step": 290 }, { "epoch": 0.04, "learning_rate": 3.6135870874488075e-07, "loss": 1.5968, "step": 300 }, { "epoch": 0.04, "learning_rate": 3.734039990363768e-07, "loss": 1.6199, "step": 310 }, { "epoch": 0.04, "learning_rate": 3.854492893278728e-07, "loss": 1.6218, "step": 320 }, { "epoch": 0.04, "learning_rate": 3.974945796193688e-07, "loss": 1.6117, "step": 330 }, { "epoch": 0.04, "learning_rate": 4.095398699108649e-07, "loss": 1.6132, "step": 340 }, { "epoch": 0.04, "learning_rate": 4.2158516020236085e-07, "loss": 1.624, "step": 350 }, { "epoch": 0.04, "learning_rate": 4.3363045049385694e-07, "loss": 1.609, "step": 360 }, { "epoch": 0.04, "learning_rate": 4.456757407853529e-07, "loss": 1.614, "step": 370 }, { "epoch": 0.05, "learning_rate": 4.57721031076849e-07, "loss": 1.5852, "step": 380 }, { "epoch": 0.05, "learning_rate": 4.69766321368345e-07, "loss": 1.6097, "step": 390 }, { "epoch": 0.05, "learning_rate": 4.81811611659841e-07, "loss": 1.5951, "step": 400 }, { "epoch": 0.05, "learning_rate": 4.938569019513371e-07, "loss": 1.6238, "step": 410 }, { "epoch": 0.05, "learning_rate": 5.05902192242833e-07, "loss": 1.6078, "step": 420 }, { "epoch": 0.05, "learning_rate": 5.179474825343292e-07, "loss": 1.6084, "step": 430 }, { "epoch": 0.05, "learning_rate": 5.299927728258251e-07, "loss": 1.6042, "step": 440 }, { "epoch": 0.05, "learning_rate": 5.420380631173212e-07, "loss": 1.5997, "step": 450 }, { "epoch": 0.06, "learning_rate": 5.540833534088172e-07, "loss": 1.6163, "step": 460 }, { "epoch": 0.06, "learning_rate": 5.661286437003132e-07, "loss": 1.6056, "step": 470 }, { "epoch": 0.06, "learning_rate": 5.781739339918092e-07, "loss": 1.621, "step": 480 }, { "epoch": 0.06, "learning_rate": 5.902192242833052e-07, "loss": 1.6047, "step": 490 }, { "epoch": 0.06, "learning_rate": 6.022645145748013e-07, "loss": 1.6021, "step": 500 }, { "epoch": 0.06, "learning_rate": 6.143098048662973e-07, "loss": 1.6011, "step": 510 }, { "epoch": 0.06, "learning_rate": 6.263550951577933e-07, "loss": 1.6059, "step": 520 }, { "epoch": 0.06, "learning_rate": 6.384003854492894e-07, "loss": 1.6037, "step": 530 }, { "epoch": 0.07, "learning_rate": 6.504456757407854e-07, "loss": 1.6248, "step": 540 }, { "epoch": 0.07, "learning_rate": 6.624909660322814e-07, "loss": 1.6103, "step": 550 }, { "epoch": 0.07, "learning_rate": 6.745362563237774e-07, "loss": 1.6127, "step": 560 }, { "epoch": 0.07, "learning_rate": 6.865815466152735e-07, "loss": 1.602, "step": 570 }, { "epoch": 0.07, "learning_rate": 6.986268369067694e-07, "loss": 1.5854, "step": 580 }, { "epoch": 0.07, "learning_rate": 7.106721271982656e-07, "loss": 1.5853, "step": 590 }, { "epoch": 0.07, "learning_rate": 7.227174174897615e-07, "loss": 1.6071, "step": 600 }, { "epoch": 0.07, "learning_rate": 7.347627077812576e-07, "loss": 1.5947, "step": 610 }, { "epoch": 0.07, "learning_rate": 7.468079980727536e-07, "loss": 1.5873, "step": 620 }, { "epoch": 0.08, "learning_rate": 7.588532883642496e-07, "loss": 1.5872, "step": 630 }, { "epoch": 0.08, "learning_rate": 7.708985786557456e-07, "loss": 1.5932, "step": 640 }, { "epoch": 0.08, "learning_rate": 7.829438689472417e-07, "loss": 1.605, "step": 650 }, { "epoch": 0.08, "learning_rate": 7.949891592387376e-07, "loss": 1.5767, "step": 660 }, { "epoch": 0.08, "learning_rate": 8.070344495302338e-07, "loss": 1.5965, "step": 670 }, { "epoch": 0.08, "learning_rate": 8.190797398217297e-07, "loss": 1.5714, "step": 680 }, { "epoch": 0.08, "learning_rate": 8.311250301132258e-07, "loss": 1.5772, "step": 690 }, { "epoch": 0.08, "learning_rate": 8.431703204047217e-07, "loss": 1.5935, "step": 700 }, { "epoch": 0.09, "learning_rate": 8.552156106962177e-07, "loss": 1.5768, "step": 710 }, { "epoch": 0.09, "learning_rate": 8.672609009877139e-07, "loss": 1.5776, "step": 720 }, { "epoch": 0.09, "learning_rate": 8.793061912792099e-07, "loss": 1.5785, "step": 730 }, { "epoch": 0.09, "learning_rate": 8.913514815707058e-07, "loss": 1.5848, "step": 740 }, { "epoch": 0.09, "learning_rate": 9.033967718622019e-07, "loss": 1.5893, "step": 750 }, { "epoch": 0.09, "learning_rate": 9.15442062153698e-07, "loss": 1.5738, "step": 760 }, { "epoch": 0.09, "learning_rate": 9.27487352445194e-07, "loss": 1.5785, "step": 770 }, { "epoch": 0.09, "learning_rate": 9.3953264273669e-07, "loss": 1.5555, "step": 780 }, { "epoch": 0.1, "learning_rate": 9.51577933028186e-07, "loss": 1.566, "step": 790 }, { "epoch": 0.1, "learning_rate": 9.63623223319682e-07, "loss": 1.5961, "step": 800 }, { "epoch": 0.1, "learning_rate": 9.756685136111782e-07, "loss": 1.5703, "step": 810 }, { "epoch": 0.1, "learning_rate": 9.877138039026742e-07, "loss": 1.589, "step": 820 }, { "epoch": 0.1, "learning_rate": 9.9975909419417e-07, "loss": 1.5582, "step": 830 }, { "epoch": 0.1, "learning_rate": 1.011804384485666e-06, "loss": 1.5772, "step": 840 }, { "epoch": 0.1, "learning_rate": 1.0238496747771623e-06, "loss": 1.5611, "step": 850 }, { "epoch": 0.1, "learning_rate": 1.0358949650686583e-06, "loss": 1.5549, "step": 860 }, { "epoch": 0.1, "learning_rate": 1.0479402553601541e-06, "loss": 1.5564, "step": 870 }, { "epoch": 0.11, "learning_rate": 1.0599855456516502e-06, "loss": 1.5875, "step": 880 }, { "epoch": 0.11, "learning_rate": 1.0720308359431462e-06, "loss": 1.5517, "step": 890 }, { "epoch": 0.11, "learning_rate": 1.0840761262346425e-06, "loss": 1.5553, "step": 900 }, { "epoch": 0.11, "learning_rate": 1.0961214165261383e-06, "loss": 1.5471, "step": 910 }, { "epoch": 0.11, "learning_rate": 1.1081667068176343e-06, "loss": 1.5581, "step": 920 }, { "epoch": 0.11, "learning_rate": 1.1202119971091303e-06, "loss": 1.5512, "step": 930 }, { "epoch": 0.11, "learning_rate": 1.1322572874006264e-06, "loss": 1.5387, "step": 940 }, { "epoch": 0.11, "learning_rate": 1.1443025776921224e-06, "loss": 1.5541, "step": 950 }, { "epoch": 0.12, "learning_rate": 1.1563478679836184e-06, "loss": 1.5618, "step": 960 }, { "epoch": 0.12, "learning_rate": 1.1683931582751145e-06, "loss": 1.5539, "step": 970 }, { "epoch": 0.12, "learning_rate": 1.1804384485666105e-06, "loss": 1.5651, "step": 980 }, { "epoch": 0.12, "learning_rate": 1.1924837388581065e-06, "loss": 1.5555, "step": 990 }, { "epoch": 0.12, "learning_rate": 1.2045290291496026e-06, "loss": 1.5345, "step": 1000 }, { "epoch": 0.12, "learning_rate": 1.2165743194410986e-06, "loss": 1.5386, "step": 1010 }, { "epoch": 0.12, "learning_rate": 1.2286196097325946e-06, "loss": 1.5398, "step": 1020 }, { "epoch": 0.12, "learning_rate": 1.2406649000240907e-06, "loss": 1.5354, "step": 1030 }, { "epoch": 0.13, "learning_rate": 1.2527101903155867e-06, "loss": 1.5375, "step": 1040 }, { "epoch": 0.13, "learning_rate": 1.2647554806070827e-06, "loss": 1.5325, "step": 1050 }, { "epoch": 0.13, "learning_rate": 1.2768007708985788e-06, "loss": 1.5159, "step": 1060 }, { "epoch": 0.13, "learning_rate": 1.2888460611900748e-06, "loss": 1.5329, "step": 1070 }, { "epoch": 0.13, "learning_rate": 1.3008913514815708e-06, "loss": 1.5263, "step": 1080 }, { "epoch": 0.13, "learning_rate": 1.3129366417730668e-06, "loss": 1.5456, "step": 1090 }, { "epoch": 0.13, "learning_rate": 1.3249819320645629e-06, "loss": 1.5279, "step": 1100 }, { "epoch": 0.13, "learning_rate": 1.337027222356059e-06, "loss": 1.5281, "step": 1110 }, { "epoch": 0.13, "learning_rate": 1.3490725126475547e-06, "loss": 1.5259, "step": 1120 }, { "epoch": 0.14, "learning_rate": 1.361117802939051e-06, "loss": 1.509, "step": 1130 }, { "epoch": 0.14, "learning_rate": 1.373163093230547e-06, "loss": 1.5127, "step": 1140 }, { "epoch": 0.14, "learning_rate": 1.385208383522043e-06, "loss": 1.5164, "step": 1150 }, { "epoch": 0.14, "learning_rate": 1.3972536738135389e-06, "loss": 1.5139, "step": 1160 }, { "epoch": 0.14, "learning_rate": 1.409298964105035e-06, "loss": 1.5099, "step": 1170 }, { "epoch": 0.14, "learning_rate": 1.4213442543965311e-06, "loss": 1.5127, "step": 1180 }, { "epoch": 0.14, "learning_rate": 1.433389544688027e-06, "loss": 1.5041, "step": 1190 }, { "epoch": 0.14, "learning_rate": 1.445434834979523e-06, "loss": 1.5022, "step": 1200 }, { "epoch": 0.15, "learning_rate": 1.457480125271019e-06, "loss": 1.4873, "step": 1210 }, { "epoch": 0.15, "learning_rate": 1.4695254155625153e-06, "loss": 1.4851, "step": 1220 }, { "epoch": 0.15, "learning_rate": 1.481570705854011e-06, "loss": 1.4827, "step": 1230 }, { "epoch": 0.15, "learning_rate": 1.4936159961455071e-06, "loss": 1.5158, "step": 1240 }, { "epoch": 0.15, "learning_rate": 1.5056612864370031e-06, "loss": 1.4789, "step": 1250 }, { "epoch": 0.15, "learning_rate": 1.5177065767284992e-06, "loss": 1.459, "step": 1260 }, { "epoch": 0.15, "learning_rate": 1.5297518670199952e-06, "loss": 1.4856, "step": 1270 }, { "epoch": 0.15, "learning_rate": 1.5417971573114912e-06, "loss": 1.4825, "step": 1280 }, { "epoch": 0.16, "learning_rate": 1.5538424476029873e-06, "loss": 1.4407, "step": 1290 }, { "epoch": 0.16, "learning_rate": 1.5658877378944833e-06, "loss": 1.4795, "step": 1300 }, { "epoch": 0.16, "learning_rate": 1.5779330281859793e-06, "loss": 1.4671, "step": 1310 }, { "epoch": 0.16, "learning_rate": 1.5899783184774752e-06, "loss": 1.4437, "step": 1320 }, { "epoch": 0.16, "learning_rate": 1.6020236087689714e-06, "loss": 1.451, "step": 1330 }, { "epoch": 0.16, "learning_rate": 1.6140688990604676e-06, "loss": 1.489, "step": 1340 }, { "epoch": 0.16, "learning_rate": 1.6261141893519635e-06, "loss": 1.4714, "step": 1350 }, { "epoch": 0.16, "learning_rate": 1.6381594796434595e-06, "loss": 1.4717, "step": 1360 }, { "epoch": 0.17, "learning_rate": 1.6502047699349553e-06, "loss": 1.4306, "step": 1370 }, { "epoch": 0.17, "learning_rate": 1.6622500602264516e-06, "loss": 1.4715, "step": 1380 }, { "epoch": 0.17, "learning_rate": 1.6742953505179476e-06, "loss": 1.4339, "step": 1390 }, { "epoch": 0.17, "learning_rate": 1.6863406408094434e-06, "loss": 1.4685, "step": 1400 }, { "epoch": 0.17, "learning_rate": 1.6983859311009397e-06, "loss": 1.414, "step": 1410 }, { "epoch": 0.17, "learning_rate": 1.7104312213924355e-06, "loss": 1.4217, "step": 1420 }, { "epoch": 0.17, "learning_rate": 1.7224765116839317e-06, "loss": 1.4231, "step": 1430 }, { "epoch": 0.17, "learning_rate": 1.7345218019754278e-06, "loss": 1.42, "step": 1440 }, { "epoch": 0.17, "learning_rate": 1.7465670922669236e-06, "loss": 1.3899, "step": 1450 }, { "epoch": 0.18, "learning_rate": 1.7586123825584198e-06, "loss": 1.3898, "step": 1460 }, { "epoch": 0.18, "learning_rate": 1.7706576728499159e-06, "loss": 1.3781, "step": 1470 }, { "epoch": 0.18, "learning_rate": 1.7827029631414117e-06, "loss": 1.4362, "step": 1480 }, { "epoch": 0.18, "learning_rate": 1.794748253432908e-06, "loss": 1.4133, "step": 1490 }, { "epoch": 0.18, "learning_rate": 1.8067935437244037e-06, "loss": 1.3684, "step": 1500 }, { "epoch": 0.18, "learning_rate": 1.8188388340159e-06, "loss": 1.4252, "step": 1510 }, { "epoch": 0.18, "learning_rate": 1.830884124307396e-06, "loss": 1.4306, "step": 1520 }, { "epoch": 0.18, "learning_rate": 1.8429294145988918e-06, "loss": 1.3647, "step": 1530 }, { "epoch": 0.19, "learning_rate": 1.854974704890388e-06, "loss": 1.3758, "step": 1540 }, { "epoch": 0.19, "learning_rate": 1.867019995181884e-06, "loss": 1.3738, "step": 1550 }, { "epoch": 0.19, "learning_rate": 1.87906528547338e-06, "loss": 1.3461, "step": 1560 }, { "epoch": 0.19, "learning_rate": 1.8911105757648762e-06, "loss": 1.3683, "step": 1570 }, { "epoch": 0.19, "learning_rate": 1.903155866056372e-06, "loss": 1.3433, "step": 1580 }, { "epoch": 0.19, "learning_rate": 1.9152011563478682e-06, "loss": 1.3271, "step": 1590 }, { "epoch": 0.19, "learning_rate": 1.927246446639364e-06, "loss": 1.3713, "step": 1600 }, { "epoch": 0.19, "learning_rate": 1.9392917369308603e-06, "loss": 1.3267, "step": 1610 }, { "epoch": 0.2, "learning_rate": 1.9513370272223563e-06, "loss": 1.3437, "step": 1620 }, { "epoch": 0.2, "learning_rate": 1.963382317513852e-06, "loss": 1.3406, "step": 1630 }, { "epoch": 0.2, "learning_rate": 1.9754276078053484e-06, "loss": 1.3392, "step": 1640 }, { "epoch": 0.2, "learning_rate": 1.987472898096844e-06, "loss": 1.3235, "step": 1650 }, { "epoch": 0.2, "learning_rate": 1.99951818838834e-06, "loss": 1.3319, "step": 1660 }, { "epoch": 0.2, "learning_rate": 2.0115634786798365e-06, "loss": 1.3191, "step": 1670 }, { "epoch": 0.2, "learning_rate": 2.023608768971332e-06, "loss": 1.3068, "step": 1680 }, { "epoch": 0.2, "learning_rate": 2.0356540592628286e-06, "loss": 1.3093, "step": 1690 }, { "epoch": 0.2, "learning_rate": 2.0476993495543246e-06, "loss": 1.2408, "step": 1700 }, { "epoch": 0.21, "learning_rate": 2.05974463984582e-06, "loss": 1.264, "step": 1710 }, { "epoch": 0.21, "learning_rate": 2.0717899301373166e-06, "loss": 1.2897, "step": 1720 }, { "epoch": 0.21, "learning_rate": 2.0838352204288123e-06, "loss": 1.2984, "step": 1730 }, { "epoch": 0.21, "learning_rate": 2.0958805107203083e-06, "loss": 1.2369, "step": 1740 }, { "epoch": 0.21, "learning_rate": 2.1079258010118047e-06, "loss": 1.208, "step": 1750 }, { "epoch": 0.21, "learning_rate": 2.1199710913033004e-06, "loss": 1.3214, "step": 1760 }, { "epoch": 0.21, "learning_rate": 2.132016381594797e-06, "loss": 1.2363, "step": 1770 }, { "epoch": 0.21, "learning_rate": 2.1440616718862924e-06, "loss": 1.2309, "step": 1780 }, { "epoch": 0.22, "learning_rate": 2.1561069621777884e-06, "loss": 1.2572, "step": 1790 }, { "epoch": 0.22, "learning_rate": 2.168152252469285e-06, "loss": 1.2339, "step": 1800 }, { "epoch": 0.22, "learning_rate": 2.1801975427607805e-06, "loss": 1.2175, "step": 1810 }, { "epoch": 0.22, "learning_rate": 2.1922428330522765e-06, "loss": 1.2457, "step": 1820 }, { "epoch": 0.22, "learning_rate": 2.2042881233437726e-06, "loss": 1.2167, "step": 1830 }, { "epoch": 0.22, "learning_rate": 2.2163334136352686e-06, "loss": 1.1986, "step": 1840 }, { "epoch": 0.22, "learning_rate": 2.228378703926765e-06, "loss": 1.2868, "step": 1850 }, { "epoch": 0.22, "learning_rate": 2.2404239942182607e-06, "loss": 1.2269, "step": 1860 }, { "epoch": 0.23, "learning_rate": 2.2524692845097567e-06, "loss": 1.1896, "step": 1870 }, { "epoch": 0.23, "learning_rate": 2.2645145748012527e-06, "loss": 1.1821, "step": 1880 }, { "epoch": 0.23, "learning_rate": 2.2765598650927488e-06, "loss": 1.1892, "step": 1890 }, { "epoch": 0.23, "learning_rate": 2.288605155384245e-06, "loss": 1.1755, "step": 1900 }, { "epoch": 0.23, "learning_rate": 2.300650445675741e-06, "loss": 1.1762, "step": 1910 }, { "epoch": 0.23, "learning_rate": 2.312695735967237e-06, "loss": 1.2152, "step": 1920 }, { "epoch": 0.23, "learning_rate": 2.324741026258733e-06, "loss": 1.1999, "step": 1930 }, { "epoch": 0.23, "learning_rate": 2.336786316550229e-06, "loss": 1.2199, "step": 1940 }, { "epoch": 0.23, "learning_rate": 2.348831606841725e-06, "loss": 1.1848, "step": 1950 }, { "epoch": 0.24, "learning_rate": 2.360876897133221e-06, "loss": 1.151, "step": 1960 }, { "epoch": 0.24, "learning_rate": 2.372922187424717e-06, "loss": 1.2209, "step": 1970 }, { "epoch": 0.24, "learning_rate": 2.384967477716213e-06, "loss": 1.088, "step": 1980 }, { "epoch": 0.24, "learning_rate": 2.397012768007709e-06, "loss": 1.1774, "step": 1990 }, { "epoch": 0.24, "learning_rate": 2.409058058299205e-06, "loss": 1.1574, "step": 2000 }, { "epoch": 0.24, "learning_rate": 2.421103348590701e-06, "loss": 1.1444, "step": 2010 }, { "epoch": 0.24, "learning_rate": 2.433148638882197e-06, "loss": 1.1154, "step": 2020 }, { "epoch": 0.24, "learning_rate": 2.4451939291736932e-06, "loss": 1.0801, "step": 2030 }, { "epoch": 0.25, "learning_rate": 2.4572392194651892e-06, "loss": 1.1164, "step": 2040 }, { "epoch": 0.25, "learning_rate": 2.4692845097566853e-06, "loss": 1.0906, "step": 2050 }, { "epoch": 0.25, "learning_rate": 2.4813298000481813e-06, "loss": 1.0527, "step": 2060 }, { "epoch": 0.25, "learning_rate": 2.4933750903396773e-06, "loss": 1.0492, "step": 2070 }, { "epoch": 0.25, "learning_rate": 2.5054203806311734e-06, "loss": 1.2051, "step": 2080 }, { "epoch": 0.25, "learning_rate": 2.5174656709226694e-06, "loss": 1.0355, "step": 2090 }, { "epoch": 0.25, "learning_rate": 2.5295109612141654e-06, "loss": 1.1023, "step": 2100 }, { "epoch": 0.25, "learning_rate": 2.5415562515056615e-06, "loss": 1.0717, "step": 2110 }, { "epoch": 0.26, "learning_rate": 2.5536015417971575e-06, "loss": 1.0252, "step": 2120 }, { "epoch": 0.26, "learning_rate": 2.5656468320886535e-06, "loss": 1.0674, "step": 2130 }, { "epoch": 0.26, "learning_rate": 2.5776921223801496e-06, "loss": 1.0206, "step": 2140 }, { "epoch": 0.26, "learning_rate": 2.5897374126716456e-06, "loss": 1.1487, "step": 2150 }, { "epoch": 0.26, "learning_rate": 2.6017827029631416e-06, "loss": 1.032, "step": 2160 }, { "epoch": 0.26, "learning_rate": 2.6138279932546377e-06, "loss": 1.1007, "step": 2170 }, { "epoch": 0.26, "learning_rate": 2.6258732835461337e-06, "loss": 1.103, "step": 2180 }, { "epoch": 0.26, "learning_rate": 2.6379185738376293e-06, "loss": 1.1403, "step": 2190 }, { "epoch": 0.26, "learning_rate": 2.6499638641291258e-06, "loss": 1.0392, "step": 2200 }, { "epoch": 0.27, "learning_rate": 2.662009154420622e-06, "loss": 1.013, "step": 2210 }, { "epoch": 0.27, "learning_rate": 2.674054444712118e-06, "loss": 1.0057, "step": 2220 }, { "epoch": 0.27, "learning_rate": 2.686099735003614e-06, "loss": 1.0906, "step": 2230 }, { "epoch": 0.27, "learning_rate": 2.6981450252951095e-06, "loss": 0.9256, "step": 2240 }, { "epoch": 0.27, "learning_rate": 2.710190315586606e-06, "loss": 1.0398, "step": 2250 }, { "epoch": 0.27, "learning_rate": 2.722235605878102e-06, "loss": 0.99, "step": 2260 }, { "epoch": 0.27, "learning_rate": 2.7342808961695976e-06, "loss": 0.9632, "step": 2270 }, { "epoch": 0.27, "learning_rate": 2.746326186461094e-06, "loss": 0.8818, "step": 2280 }, { "epoch": 0.28, "learning_rate": 2.7583714767525896e-06, "loss": 0.8905, "step": 2290 }, { "epoch": 0.28, "learning_rate": 2.770416767044086e-06, "loss": 0.9031, "step": 2300 }, { "epoch": 0.28, "learning_rate": 2.782462057335582e-06, "loss": 1.0191, "step": 2310 }, { "epoch": 0.28, "learning_rate": 2.7945073476270777e-06, "loss": 0.9683, "step": 2320 }, { "epoch": 0.28, "learning_rate": 2.806552637918574e-06, "loss": 0.9268, "step": 2330 }, { "epoch": 0.28, "learning_rate": 2.81859792821007e-06, "loss": 0.8958, "step": 2340 }, { "epoch": 0.28, "learning_rate": 2.830643218501566e-06, "loss": 0.9567, "step": 2350 }, { "epoch": 0.28, "learning_rate": 2.8426885087930623e-06, "loss": 0.9084, "step": 2360 }, { "epoch": 0.29, "learning_rate": 2.854733799084558e-06, "loss": 0.9237, "step": 2370 }, { "epoch": 0.29, "learning_rate": 2.866779089376054e-06, "loss": 0.8828, "step": 2380 }, { "epoch": 0.29, "learning_rate": 2.8788243796675504e-06, "loss": 0.9673, "step": 2390 }, { "epoch": 0.29, "learning_rate": 2.890869669959046e-06, "loss": 0.9653, "step": 2400 }, { "epoch": 0.29, "learning_rate": 2.9029149602505424e-06, "loss": 0.7813, "step": 2410 }, { "epoch": 0.29, "learning_rate": 2.914960250542038e-06, "loss": 0.9837, "step": 2420 }, { "epoch": 0.29, "learning_rate": 2.927005540833534e-06, "loss": 0.883, "step": 2430 }, { "epoch": 0.29, "learning_rate": 2.9390508311250305e-06, "loss": 0.8632, "step": 2440 }, { "epoch": 0.3, "learning_rate": 2.951096121416526e-06, "loss": 1.018, "step": 2450 }, { "epoch": 0.3, "learning_rate": 2.963141411708022e-06, "loss": 0.8925, "step": 2460 }, { "epoch": 0.3, "learning_rate": 2.975186701999518e-06, "loss": 0.9369, "step": 2470 }, { "epoch": 0.3, "learning_rate": 2.9872319922910142e-06, "loss": 0.8835, "step": 2480 }, { "epoch": 0.3, "learning_rate": 2.9992772825825107e-06, "loss": 0.7873, "step": 2490 }, { "epoch": 0.3, "learning_rate": 3.0113225728740063e-06, "loss": 0.8504, "step": 2500 }, { "epoch": 0.3, "learning_rate": 3.0233678631655023e-06, "loss": 0.8252, "step": 2510 }, { "epoch": 0.3, "learning_rate": 3.0354131534569984e-06, "loss": 0.8785, "step": 2520 }, { "epoch": 0.3, "learning_rate": 3.0474584437484944e-06, "loss": 0.8333, "step": 2530 }, { "epoch": 0.31, "learning_rate": 3.0595037340399904e-06, "loss": 0.8386, "step": 2540 }, { "epoch": 0.31, "learning_rate": 3.0715490243314865e-06, "loss": 1.0213, "step": 2550 }, { "epoch": 0.31, "learning_rate": 3.0835943146229825e-06, "loss": 1.0045, "step": 2560 }, { "epoch": 0.31, "learning_rate": 3.095639604914479e-06, "loss": 0.8696, "step": 2570 }, { "epoch": 0.31, "learning_rate": 3.1076848952059746e-06, "loss": 0.8633, "step": 2580 }, { "epoch": 0.31, "learning_rate": 3.1197301854974706e-06, "loss": 0.7537, "step": 2590 }, { "epoch": 0.31, "learning_rate": 3.1317754757889666e-06, "loss": 0.9511, "step": 2600 }, { "epoch": 0.31, "learning_rate": 3.1438207660804622e-06, "loss": 0.8493, "step": 2610 }, { "epoch": 0.32, "learning_rate": 3.1558660563719587e-06, "loss": 0.8316, "step": 2620 }, { "epoch": 0.32, "learning_rate": 3.1679113466634547e-06, "loss": 0.9185, "step": 2630 }, { "epoch": 0.32, "learning_rate": 3.1799566369549503e-06, "loss": 0.825, "step": 2640 }, { "epoch": 0.32, "learning_rate": 3.192001927246447e-06, "loss": 0.8185, "step": 2650 }, { "epoch": 0.32, "learning_rate": 3.204047217537943e-06, "loss": 0.8666, "step": 2660 }, { "epoch": 0.32, "learning_rate": 3.216092507829439e-06, "loss": 0.6777, "step": 2670 }, { "epoch": 0.32, "learning_rate": 3.2281377981209353e-06, "loss": 0.8687, "step": 2680 }, { "epoch": 0.32, "learning_rate": 3.240183088412431e-06, "loss": 0.7363, "step": 2690 }, { "epoch": 0.33, "learning_rate": 3.252228378703927e-06, "loss": 0.738, "step": 2700 }, { "epoch": 0.33, "learning_rate": 3.2642736689954234e-06, "loss": 0.8507, "step": 2710 }, { "epoch": 0.33, "learning_rate": 3.276318959286919e-06, "loss": 0.8058, "step": 2720 }, { "epoch": 0.33, "learning_rate": 3.288364249578415e-06, "loss": 0.8257, "step": 2730 }, { "epoch": 0.33, "learning_rate": 3.3004095398699106e-06, "loss": 0.7094, "step": 2740 }, { "epoch": 0.33, "learning_rate": 3.312454830161407e-06, "loss": 0.8156, "step": 2750 }, { "epoch": 0.33, "learning_rate": 3.324500120452903e-06, "loss": 0.807, "step": 2760 }, { "epoch": 0.33, "learning_rate": 3.3365454107443987e-06, "loss": 0.924, "step": 2770 }, { "epoch": 0.33, "learning_rate": 3.348590701035895e-06, "loss": 0.6514, "step": 2780 }, { "epoch": 0.34, "learning_rate": 3.3606359913273912e-06, "loss": 0.8851, "step": 2790 }, { "epoch": 0.34, "learning_rate": 3.372681281618887e-06, "loss": 0.7191, "step": 2800 }, { "epoch": 0.34, "learning_rate": 3.3847265719103837e-06, "loss": 0.8087, "step": 2810 }, { "epoch": 0.34, "learning_rate": 3.3967718622018793e-06, "loss": 0.7068, "step": 2820 }, { "epoch": 0.34, "learning_rate": 3.408817152493375e-06, "loss": 0.879, "step": 2830 }, { "epoch": 0.34, "learning_rate": 3.420862442784871e-06, "loss": 0.7523, "step": 2840 }, { "epoch": 0.34, "learning_rate": 3.4329077330763674e-06, "loss": 0.6441, "step": 2850 }, { "epoch": 0.34, "learning_rate": 3.4449530233678634e-06, "loss": 0.8603, "step": 2860 }, { "epoch": 0.35, "learning_rate": 3.456998313659359e-06, "loss": 0.7534, "step": 2870 }, { "epoch": 0.35, "learning_rate": 3.4690436039508555e-06, "loss": 0.6736, "step": 2880 }, { "epoch": 0.35, "learning_rate": 3.4810888942423515e-06, "loss": 0.6928, "step": 2890 }, { "epoch": 0.35, "learning_rate": 3.493134184533847e-06, "loss": 0.7565, "step": 2900 }, { "epoch": 0.35, "learning_rate": 3.5051794748253436e-06, "loss": 0.7842, "step": 2910 }, { "epoch": 0.35, "learning_rate": 3.5172247651168396e-06, "loss": 0.7485, "step": 2920 }, { "epoch": 0.35, "learning_rate": 3.5292700554083352e-06, "loss": 0.6787, "step": 2930 }, { "epoch": 0.35, "learning_rate": 3.5413153456998317e-06, "loss": 0.6819, "step": 2940 }, { "epoch": 0.36, "learning_rate": 3.5533606359913277e-06, "loss": 0.8492, "step": 2950 }, { "epoch": 0.36, "learning_rate": 3.5654059262828233e-06, "loss": 0.7842, "step": 2960 }, { "epoch": 0.36, "learning_rate": 3.5774512165743194e-06, "loss": 0.7891, "step": 2970 }, { "epoch": 0.36, "learning_rate": 3.589496506865816e-06, "loss": 0.852, "step": 2980 }, { "epoch": 0.36, "learning_rate": 3.6015417971573114e-06, "loss": 0.806, "step": 2990 }, { "epoch": 0.36, "learning_rate": 3.6135870874488075e-06, "loss": 0.6639, "step": 3000 }, { "epoch": 0.36, "learning_rate": 3.625632377740304e-06, "loss": 0.8274, "step": 3010 }, { "epoch": 0.36, "learning_rate": 3.6376776680318e-06, "loss": 0.7737, "step": 3020 }, { "epoch": 0.36, "learning_rate": 3.6497229583232956e-06, "loss": 0.7151, "step": 3030 }, { "epoch": 0.37, "learning_rate": 3.661768248614792e-06, "loss": 0.8096, "step": 3040 }, { "epoch": 0.37, "learning_rate": 3.673813538906288e-06, "loss": 0.6899, "step": 3050 }, { "epoch": 0.37, "learning_rate": 3.6858588291977837e-06, "loss": 0.6569, "step": 3060 }, { "epoch": 0.37, "learning_rate": 3.6979041194892797e-06, "loss": 0.7375, "step": 3070 }, { "epoch": 0.37, "learning_rate": 3.709949409780776e-06, "loss": 0.7564, "step": 3080 }, { "epoch": 0.37, "learning_rate": 3.7219947000722718e-06, "loss": 0.7394, "step": 3090 }, { "epoch": 0.37, "learning_rate": 3.734039990363768e-06, "loss": 0.7201, "step": 3100 }, { "epoch": 0.37, "learning_rate": 3.7460852806552642e-06, "loss": 0.7294, "step": 3110 }, { "epoch": 0.38, "learning_rate": 3.75813057094676e-06, "loss": 0.6986, "step": 3120 }, { "epoch": 0.38, "learning_rate": 3.770175861238256e-06, "loss": 0.6387, "step": 3130 }, { "epoch": 0.38, "learning_rate": 3.7822211515297523e-06, "loss": 0.6523, "step": 3140 }, { "epoch": 0.38, "learning_rate": 3.794266441821248e-06, "loss": 0.7157, "step": 3150 }, { "epoch": 0.38, "learning_rate": 3.806311732112744e-06, "loss": 0.6255, "step": 3160 }, { "epoch": 0.38, "learning_rate": 3.8183570224042404e-06, "loss": 0.6307, "step": 3170 }, { "epoch": 0.38, "learning_rate": 3.8304023126957365e-06, "loss": 0.7539, "step": 3180 }, { "epoch": 0.38, "learning_rate": 3.8424476029872325e-06, "loss": 0.761, "step": 3190 }, { "epoch": 0.39, "learning_rate": 3.854492893278728e-06, "loss": 0.7775, "step": 3200 }, { "epoch": 0.39, "learning_rate": 3.8665381835702246e-06, "loss": 0.6311, "step": 3210 }, { "epoch": 0.39, "learning_rate": 3.878583473861721e-06, "loss": 0.7344, "step": 3220 }, { "epoch": 0.39, "learning_rate": 3.890628764153216e-06, "loss": 0.6123, "step": 3230 }, { "epoch": 0.39, "learning_rate": 3.902674054444713e-06, "loss": 0.7842, "step": 3240 }, { "epoch": 0.39, "learning_rate": 3.914719344736209e-06, "loss": 0.6592, "step": 3250 }, { "epoch": 0.39, "learning_rate": 3.926764635027704e-06, "loss": 0.6351, "step": 3260 }, { "epoch": 0.39, "learning_rate": 3.938809925319201e-06, "loss": 0.6838, "step": 3270 }, { "epoch": 0.4, "learning_rate": 3.950855215610697e-06, "loss": 0.6787, "step": 3280 }, { "epoch": 0.4, "learning_rate": 3.962900505902192e-06, "loss": 0.6171, "step": 3290 }, { "epoch": 0.4, "learning_rate": 3.974945796193688e-06, "loss": 0.6119, "step": 3300 }, { "epoch": 0.4, "learning_rate": 3.986991086485185e-06, "loss": 0.6064, "step": 3310 }, { "epoch": 0.4, "learning_rate": 3.99903637677668e-06, "loss": 0.8103, "step": 3320 }, { "epoch": 0.4, "learning_rate": 4.011081667068176e-06, "loss": 0.7416, "step": 3330 }, { "epoch": 0.4, "learning_rate": 4.023126957359673e-06, "loss": 0.6061, "step": 3340 }, { "epoch": 0.4, "learning_rate": 4.035172247651168e-06, "loss": 0.6428, "step": 3350 }, { "epoch": 0.4, "learning_rate": 4.047217537942664e-06, "loss": 0.611, "step": 3360 }, { "epoch": 0.41, "learning_rate": 4.059262828234161e-06, "loss": 0.6872, "step": 3370 }, { "epoch": 0.41, "learning_rate": 4.071308118525657e-06, "loss": 0.7002, "step": 3380 }, { "epoch": 0.41, "learning_rate": 4.083353408817152e-06, "loss": 0.5585, "step": 3390 }, { "epoch": 0.41, "learning_rate": 4.095398699108649e-06, "loss": 0.7227, "step": 3400 }, { "epoch": 0.41, "learning_rate": 4.107443989400145e-06, "loss": 0.7608, "step": 3410 }, { "epoch": 0.41, "learning_rate": 4.11948927969164e-06, "loss": 0.7097, "step": 3420 }, { "epoch": 0.41, "learning_rate": 4.131534569983136e-06, "loss": 0.5039, "step": 3430 }, { "epoch": 0.41, "learning_rate": 4.143579860274633e-06, "loss": 0.8008, "step": 3440 }, { "epoch": 0.42, "learning_rate": 4.1556251505661285e-06, "loss": 0.8033, "step": 3450 }, { "epoch": 0.42, "learning_rate": 4.1676704408576245e-06, "loss": 0.6777, "step": 3460 }, { "epoch": 0.42, "learning_rate": 4.179715731149121e-06, "loss": 0.5464, "step": 3470 }, { "epoch": 0.42, "learning_rate": 4.191761021440617e-06, "loss": 0.602, "step": 3480 }, { "epoch": 0.42, "learning_rate": 4.203806311732113e-06, "loss": 0.6271, "step": 3490 }, { "epoch": 0.42, "learning_rate": 4.2158516020236095e-06, "loss": 0.5394, "step": 3500 }, { "epoch": 0.42, "learning_rate": 4.227896892315105e-06, "loss": 0.5083, "step": 3510 }, { "epoch": 0.42, "learning_rate": 4.239942182606601e-06, "loss": 0.7406, "step": 3520 }, { "epoch": 0.43, "learning_rate": 4.251987472898097e-06, "loss": 0.5001, "step": 3530 }, { "epoch": 0.43, "learning_rate": 4.264032763189594e-06, "loss": 0.581, "step": 3540 }, { "epoch": 0.43, "learning_rate": 4.276078053481089e-06, "loss": 0.5438, "step": 3550 }, { "epoch": 0.43, "learning_rate": 4.288123343772585e-06, "loss": 0.6164, "step": 3560 }, { "epoch": 0.43, "learning_rate": 4.300168634064082e-06, "loss": 0.5689, "step": 3570 }, { "epoch": 0.43, "learning_rate": 4.312213924355577e-06, "loss": 0.6068, "step": 3580 }, { "epoch": 0.43, "learning_rate": 4.324259214647073e-06, "loss": 0.78, "step": 3590 }, { "epoch": 0.43, "learning_rate": 4.33630450493857e-06, "loss": 0.5941, "step": 3600 }, { "epoch": 0.43, "learning_rate": 4.348349795230065e-06, "loss": 0.5974, "step": 3610 }, { "epoch": 0.44, "learning_rate": 4.360395085521561e-06, "loss": 0.6201, "step": 3620 }, { "epoch": 0.44, "learning_rate": 4.372440375813058e-06, "loss": 0.6388, "step": 3630 }, { "epoch": 0.44, "learning_rate": 4.384485666104553e-06, "loss": 0.6887, "step": 3640 }, { "epoch": 0.44, "learning_rate": 4.396530956396049e-06, "loss": 0.4635, "step": 3650 }, { "epoch": 0.44, "learning_rate": 4.408576246687545e-06, "loss": 0.5648, "step": 3660 }, { "epoch": 0.44, "learning_rate": 4.420621536979041e-06, "loss": 0.5182, "step": 3670 }, { "epoch": 0.44, "learning_rate": 4.432666827270537e-06, "loss": 0.5923, "step": 3680 }, { "epoch": 0.44, "learning_rate": 4.444712117562033e-06, "loss": 0.7088, "step": 3690 }, { "epoch": 0.45, "learning_rate": 4.45675740785353e-06, "loss": 0.4766, "step": 3700 }, { "epoch": 0.45, "learning_rate": 4.468802698145025e-06, "loss": 0.6942, "step": 3710 }, { "epoch": 0.45, "learning_rate": 4.480847988436521e-06, "loss": 0.5993, "step": 3720 }, { "epoch": 0.45, "learning_rate": 4.492893278728018e-06, "loss": 0.6405, "step": 3730 }, { "epoch": 0.45, "learning_rate": 4.504938569019513e-06, "loss": 0.5261, "step": 3740 }, { "epoch": 0.45, "learning_rate": 4.5169838593110094e-06, "loss": 0.7343, "step": 3750 }, { "epoch": 0.45, "learning_rate": 4.5290291496025055e-06, "loss": 0.5309, "step": 3760 }, { "epoch": 0.45, "learning_rate": 4.5410744398940015e-06, "loss": 0.8686, "step": 3770 }, { "epoch": 0.46, "learning_rate": 4.5531197301854975e-06, "loss": 0.4225, "step": 3780 }, { "epoch": 0.46, "learning_rate": 4.5651650204769936e-06, "loss": 0.6005, "step": 3790 }, { "epoch": 0.46, "learning_rate": 4.57721031076849e-06, "loss": 0.5354, "step": 3800 }, { "epoch": 0.46, "learning_rate": 4.589255601059986e-06, "loss": 0.6548, "step": 3810 }, { "epoch": 0.46, "learning_rate": 4.601300891351482e-06, "loss": 0.6499, "step": 3820 }, { "epoch": 0.46, "learning_rate": 4.613346181642978e-06, "loss": 0.7827, "step": 3830 }, { "epoch": 0.46, "learning_rate": 4.625391471934474e-06, "loss": 0.4766, "step": 3840 }, { "epoch": 0.46, "learning_rate": 4.63743676222597e-06, "loss": 0.485, "step": 3850 }, { "epoch": 0.46, "learning_rate": 4.649482052517466e-06, "loss": 0.6619, "step": 3860 }, { "epoch": 0.47, "learning_rate": 4.661527342808962e-06, "loss": 0.5379, "step": 3870 }, { "epoch": 0.47, "learning_rate": 4.673572633100458e-06, "loss": 0.5998, "step": 3880 }, { "epoch": 0.47, "learning_rate": 4.685617923391954e-06, "loss": 0.6462, "step": 3890 }, { "epoch": 0.47, "learning_rate": 4.69766321368345e-06, "loss": 0.5092, "step": 3900 }, { "epoch": 0.47, "learning_rate": 4.709708503974946e-06, "loss": 0.5216, "step": 3910 }, { "epoch": 0.47, "learning_rate": 4.721753794266442e-06, "loss": 0.5176, "step": 3920 }, { "epoch": 0.47, "learning_rate": 4.733799084557938e-06, "loss": 0.5415, "step": 3930 }, { "epoch": 0.47, "learning_rate": 4.745844374849434e-06, "loss": 0.4709, "step": 3940 }, { "epoch": 0.48, "learning_rate": 4.75788966514093e-06, "loss": 0.5058, "step": 3950 }, { "epoch": 0.48, "learning_rate": 4.769934955432426e-06, "loss": 0.4911, "step": 3960 }, { "epoch": 0.48, "learning_rate": 4.781980245723922e-06, "loss": 0.4754, "step": 3970 }, { "epoch": 0.48, "learning_rate": 4.794025536015418e-06, "loss": 0.655, "step": 3980 }, { "epoch": 0.48, "learning_rate": 4.806070826306914e-06, "loss": 0.6335, "step": 3990 }, { "epoch": 0.48, "learning_rate": 4.81811611659841e-06, "loss": 0.7247, "step": 4000 }, { "epoch": 0.48, "learning_rate": 4.830161406889906e-06, "loss": 0.589, "step": 4010 }, { "epoch": 0.48, "learning_rate": 4.842206697181402e-06, "loss": 0.6547, "step": 4020 }, { "epoch": 0.49, "learning_rate": 4.854251987472898e-06, "loss": 0.5324, "step": 4030 }, { "epoch": 0.49, "learning_rate": 4.866297277764394e-06, "loss": 0.5176, "step": 4040 }, { "epoch": 0.49, "learning_rate": 4.87834256805589e-06, "loss": 0.4329, "step": 4050 }, { "epoch": 0.49, "learning_rate": 4.8903878583473864e-06, "loss": 0.6436, "step": 4060 }, { "epoch": 0.49, "learning_rate": 4.9024331486388825e-06, "loss": 0.7, "step": 4070 }, { "epoch": 0.49, "learning_rate": 4.9144784389303785e-06, "loss": 0.6959, "step": 4080 }, { "epoch": 0.49, "learning_rate": 4.9265237292218745e-06, "loss": 0.5458, "step": 4090 }, { "epoch": 0.49, "learning_rate": 4.9385690195133706e-06, "loss": 0.5097, "step": 4100 }, { "epoch": 0.5, "learning_rate": 4.950614309804867e-06, "loss": 0.5525, "step": 4110 }, { "epoch": 0.5, "learning_rate": 4.962659600096363e-06, "loss": 0.545, "step": 4120 }, { "epoch": 0.5, "learning_rate": 4.974704890387859e-06, "loss": 0.6471, "step": 4130 }, { "epoch": 0.5, "learning_rate": 4.986750180679355e-06, "loss": 0.7276, "step": 4140 }, { "epoch": 0.5, "learning_rate": 4.998795470970851e-06, "loss": 0.5004, "step": 4150 }, { "epoch": 0.5, "learning_rate": 5.010840761262347e-06, "loss": 0.6134, "step": 4160 }, { "epoch": 0.5, "learning_rate": 5.022886051553843e-06, "loss": 0.4791, "step": 4170 }, { "epoch": 0.5, "learning_rate": 5.034931341845339e-06, "loss": 0.6029, "step": 4180 }, { "epoch": 0.5, "learning_rate": 5.046976632136835e-06, "loss": 0.5072, "step": 4190 }, { "epoch": 0.51, "learning_rate": 5.059021922428331e-06, "loss": 0.6683, "step": 4200 }, { "epoch": 0.51, "learning_rate": 5.071067212719827e-06, "loss": 0.4545, "step": 4210 }, { "epoch": 0.51, "learning_rate": 5.083112503011323e-06, "loss": 0.4516, "step": 4220 }, { "epoch": 0.51, "learning_rate": 5.095157793302819e-06, "loss": 0.6589, "step": 4230 }, { "epoch": 0.51, "learning_rate": 5.107203083594315e-06, "loss": 0.5102, "step": 4240 }, { "epoch": 0.51, "learning_rate": 5.11924837388581e-06, "loss": 0.5088, "step": 4250 }, { "epoch": 0.51, "learning_rate": 5.131293664177307e-06, "loss": 0.481, "step": 4260 }, { "epoch": 0.51, "learning_rate": 5.143338954468803e-06, "loss": 0.3794, "step": 4270 }, { "epoch": 0.52, "learning_rate": 5.155384244760299e-06, "loss": 0.4362, "step": 4280 }, { "epoch": 0.52, "learning_rate": 5.167429535051795e-06, "loss": 0.5157, "step": 4290 }, { "epoch": 0.52, "learning_rate": 5.179474825343291e-06, "loss": 0.4668, "step": 4300 }, { "epoch": 0.52, "learning_rate": 5.191520115634787e-06, "loss": 0.4992, "step": 4310 }, { "epoch": 0.52, "learning_rate": 5.203565405926283e-06, "loss": 0.5478, "step": 4320 }, { "epoch": 0.52, "learning_rate": 5.215610696217779e-06, "loss": 0.5219, "step": 4330 }, { "epoch": 0.52, "learning_rate": 5.227655986509275e-06, "loss": 0.5858, "step": 4340 }, { "epoch": 0.52, "learning_rate": 5.2397012768007705e-06, "loss": 0.6463, "step": 4350 }, { "epoch": 0.53, "learning_rate": 5.251746567092267e-06, "loss": 0.5957, "step": 4360 }, { "epoch": 0.53, "learning_rate": 5.2637918573837634e-06, "loss": 0.4581, "step": 4370 }, { "epoch": 0.53, "learning_rate": 5.275837147675259e-06, "loss": 0.5995, "step": 4380 }, { "epoch": 0.53, "learning_rate": 5.2878824379667555e-06, "loss": 0.4992, "step": 4390 }, { "epoch": 0.53, "learning_rate": 5.2999277282582515e-06, "loss": 0.5782, "step": 4400 }, { "epoch": 0.53, "learning_rate": 5.311973018549747e-06, "loss": 0.4754, "step": 4410 }, { "epoch": 0.53, "learning_rate": 5.324018308841244e-06, "loss": 0.5505, "step": 4420 }, { "epoch": 0.53, "learning_rate": 5.33606359913274e-06, "loss": 0.5724, "step": 4430 }, { "epoch": 0.53, "learning_rate": 5.348108889424236e-06, "loss": 0.5318, "step": 4440 }, { "epoch": 0.54, "learning_rate": 5.360154179715732e-06, "loss": 0.6368, "step": 4450 }, { "epoch": 0.54, "learning_rate": 5.372199470007228e-06, "loss": 0.6284, "step": 4460 }, { "epoch": 0.54, "learning_rate": 5.384244760298724e-06, "loss": 0.5533, "step": 4470 }, { "epoch": 0.54, "learning_rate": 5.396290050590219e-06, "loss": 0.4426, "step": 4480 }, { "epoch": 0.54, "learning_rate": 5.408335340881716e-06, "loss": 0.4107, "step": 4490 }, { "epoch": 0.54, "learning_rate": 5.420380631173212e-06, "loss": 0.6605, "step": 4500 }, { "epoch": 0.54, "learning_rate": 5.432425921464707e-06, "loss": 0.4764, "step": 4510 }, { "epoch": 0.54, "learning_rate": 5.444471211756204e-06, "loss": 0.4938, "step": 4520 }, { "epoch": 0.55, "learning_rate": 5.4565165020477e-06, "loss": 0.5035, "step": 4530 }, { "epoch": 0.55, "learning_rate": 5.468561792339195e-06, "loss": 0.4689, "step": 4540 }, { "epoch": 0.55, "learning_rate": 5.480607082630692e-06, "loss": 0.5122, "step": 4550 }, { "epoch": 0.55, "learning_rate": 5.492652372922188e-06, "loss": 0.4704, "step": 4560 }, { "epoch": 0.55, "learning_rate": 5.504697663213683e-06, "loss": 0.5194, "step": 4570 }, { "epoch": 0.55, "learning_rate": 5.516742953505179e-06, "loss": 0.6027, "step": 4580 }, { "epoch": 0.55, "learning_rate": 5.528788243796676e-06, "loss": 0.4998, "step": 4590 }, { "epoch": 0.55, "learning_rate": 5.540833534088172e-06, "loss": 0.5504, "step": 4600 }, { "epoch": 0.56, "learning_rate": 5.552878824379667e-06, "loss": 0.6342, "step": 4610 }, { "epoch": 0.56, "learning_rate": 5.564924114671164e-06, "loss": 0.5564, "step": 4620 }, { "epoch": 0.56, "learning_rate": 5.57696940496266e-06, "loss": 0.4296, "step": 4630 }, { "epoch": 0.56, "learning_rate": 5.5890146952541554e-06, "loss": 0.4676, "step": 4640 }, { "epoch": 0.56, "learning_rate": 5.601059985545652e-06, "loss": 0.7282, "step": 4650 }, { "epoch": 0.56, "learning_rate": 5.613105275837148e-06, "loss": 0.4806, "step": 4660 }, { "epoch": 0.56, "learning_rate": 5.6251505661286435e-06, "loss": 0.7047, "step": 4670 }, { "epoch": 0.56, "learning_rate": 5.63719585642014e-06, "loss": 0.3619, "step": 4680 }, { "epoch": 0.56, "learning_rate": 5.6492411467116364e-06, "loss": 0.5778, "step": 4690 }, { "epoch": 0.57, "learning_rate": 5.661286437003132e-06, "loss": 0.6242, "step": 4700 }, { "epoch": 0.57, "learning_rate": 5.673331727294628e-06, "loss": 0.4773, "step": 4710 }, { "epoch": 0.57, "learning_rate": 5.6853770175861245e-06, "loss": 0.5888, "step": 4720 }, { "epoch": 0.57, "learning_rate": 5.69742230787762e-06, "loss": 0.4765, "step": 4730 }, { "epoch": 0.57, "learning_rate": 5.709467598169116e-06, "loss": 0.39, "step": 4740 }, { "epoch": 0.57, "learning_rate": 5.721512888460613e-06, "loss": 0.4885, "step": 4750 }, { "epoch": 0.57, "learning_rate": 5.733558178752108e-06, "loss": 0.538, "step": 4760 }, { "epoch": 0.57, "learning_rate": 5.745603469043604e-06, "loss": 0.4018, "step": 4770 }, { "epoch": 0.58, "learning_rate": 5.757648759335101e-06, "loss": 0.6024, "step": 4780 }, { "epoch": 0.58, "learning_rate": 5.769694049626597e-06, "loss": 0.5773, "step": 4790 }, { "epoch": 0.58, "learning_rate": 5.781739339918092e-06, "loss": 0.4602, "step": 4800 }, { "epoch": 0.58, "learning_rate": 5.793784630209588e-06, "loss": 0.6262, "step": 4810 }, { "epoch": 0.58, "learning_rate": 5.805829920501085e-06, "loss": 0.4825, "step": 4820 }, { "epoch": 0.58, "learning_rate": 5.81787521079258e-06, "loss": 0.5615, "step": 4830 }, { "epoch": 0.58, "learning_rate": 5.829920501084076e-06, "loss": 0.612, "step": 4840 }, { "epoch": 0.58, "learning_rate": 5.841965791375573e-06, "loss": 0.2971, "step": 4850 }, { "epoch": 0.59, "learning_rate": 5.854011081667068e-06, "loss": 0.5283, "step": 4860 }, { "epoch": 0.59, "learning_rate": 5.866056371958564e-06, "loss": 0.4795, "step": 4870 }, { "epoch": 0.59, "learning_rate": 5.878101662250061e-06, "loss": 0.7299, "step": 4880 }, { "epoch": 0.59, "learning_rate": 5.890146952541556e-06, "loss": 0.6781, "step": 4890 }, { "epoch": 0.59, "learning_rate": 5.902192242833052e-06, "loss": 0.4355, "step": 4900 }, { "epoch": 0.59, "learning_rate": 5.914237533124549e-06, "loss": 0.602, "step": 4910 }, { "epoch": 0.59, "learning_rate": 5.926282823416044e-06, "loss": 0.5103, "step": 4920 }, { "epoch": 0.59, "learning_rate": 5.93832811370754e-06, "loss": 0.3305, "step": 4930 }, { "epoch": 0.6, "learning_rate": 5.950373403999036e-06, "loss": 0.5849, "step": 4940 }, { "epoch": 0.6, "learning_rate": 5.962418694290533e-06, "loss": 0.4632, "step": 4950 }, { "epoch": 0.6, "learning_rate": 5.9744639845820285e-06, "loss": 0.504, "step": 4960 }, { "epoch": 0.6, "learning_rate": 5.9865092748735245e-06, "loss": 0.4671, "step": 4970 }, { "epoch": 0.6, "learning_rate": 5.998554565165021e-06, "loss": 0.5343, "step": 4980 }, { "epoch": 0.6, "learning_rate": 6.0105998554565166e-06, "loss": 0.4728, "step": 4990 }, { "epoch": 0.6, "learning_rate": 6.022645145748013e-06, "loss": 0.4928, "step": 5000 }, { "epoch": 0.6, "learning_rate": 6.0346904360395095e-06, "loss": 0.4513, "step": 5010 }, { "epoch": 0.6, "learning_rate": 6.046735726331005e-06, "loss": 0.5332, "step": 5020 }, { "epoch": 0.61, "learning_rate": 6.058781016622501e-06, "loss": 0.5641, "step": 5030 }, { "epoch": 0.61, "learning_rate": 6.070826306913997e-06, "loss": 0.4369, "step": 5040 }, { "epoch": 0.61, "learning_rate": 6.082871597205493e-06, "loss": 0.3886, "step": 5050 }, { "epoch": 0.61, "learning_rate": 6.094916887496989e-06, "loss": 0.3751, "step": 5060 }, { "epoch": 0.61, "learning_rate": 6.106962177788485e-06, "loss": 0.519, "step": 5070 }, { "epoch": 0.61, "learning_rate": 6.119007468079981e-06, "loss": 0.459, "step": 5080 }, { "epoch": 0.61, "learning_rate": 6.131052758371477e-06, "loss": 0.4521, "step": 5090 }, { "epoch": 0.61, "learning_rate": 6.143098048662973e-06, "loss": 0.4481, "step": 5100 }, { "epoch": 0.62, "learning_rate": 6.15514333895447e-06, "loss": 0.4205, "step": 5110 }, { "epoch": 0.62, "learning_rate": 6.167188629245965e-06, "loss": 0.421, "step": 5120 }, { "epoch": 0.62, "learning_rate": 6.179233919537461e-06, "loss": 0.4428, "step": 5130 }, { "epoch": 0.62, "learning_rate": 6.191279209828958e-06, "loss": 0.5141, "step": 5140 }, { "epoch": 0.62, "learning_rate": 6.203324500120453e-06, "loss": 0.3254, "step": 5150 }, { "epoch": 0.62, "learning_rate": 6.215369790411949e-06, "loss": 0.4218, "step": 5160 }, { "epoch": 0.62, "learning_rate": 6.227415080703445e-06, "loss": 0.4599, "step": 5170 }, { "epoch": 0.62, "learning_rate": 6.239460370994941e-06, "loss": 0.4596, "step": 5180 }, { "epoch": 0.63, "learning_rate": 6.251505661286437e-06, "loss": 0.5694, "step": 5190 }, { "epoch": 0.63, "learning_rate": 6.263550951577933e-06, "loss": 0.3899, "step": 5200 }, { "epoch": 0.63, "learning_rate": 6.275596241869429e-06, "loss": 0.3831, "step": 5210 }, { "epoch": 0.63, "learning_rate": 6.2876415321609244e-06, "loss": 0.379, "step": 5220 }, { "epoch": 0.63, "learning_rate": 6.299686822452422e-06, "loss": 0.4253, "step": 5230 }, { "epoch": 0.63, "learning_rate": 6.311732112743917e-06, "loss": 0.4373, "step": 5240 }, { "epoch": 0.63, "learning_rate": 6.323777403035413e-06, "loss": 0.4003, "step": 5250 }, { "epoch": 0.63, "learning_rate": 6.335822693326909e-06, "loss": 0.6563, "step": 5260 }, { "epoch": 0.63, "learning_rate": 6.3478679836184055e-06, "loss": 0.5389, "step": 5270 }, { "epoch": 0.64, "learning_rate": 6.359913273909901e-06, "loss": 0.3611, "step": 5280 }, { "epoch": 0.64, "learning_rate": 6.371958564201398e-06, "loss": 0.4825, "step": 5290 }, { "epoch": 0.64, "learning_rate": 6.384003854492894e-06, "loss": 0.3614, "step": 5300 }, { "epoch": 0.64, "learning_rate": 6.39604914478439e-06, "loss": 0.4205, "step": 5310 }, { "epoch": 0.64, "learning_rate": 6.408094435075886e-06, "loss": 0.5031, "step": 5320 }, { "epoch": 0.64, "learning_rate": 6.420139725367382e-06, "loss": 0.4996, "step": 5330 }, { "epoch": 0.64, "learning_rate": 6.432185015658878e-06, "loss": 0.5412, "step": 5340 }, { "epoch": 0.64, "learning_rate": 6.444230305950373e-06, "loss": 0.5283, "step": 5350 }, { "epoch": 0.65, "learning_rate": 6.456275596241871e-06, "loss": 0.5612, "step": 5360 }, { "epoch": 0.65, "learning_rate": 6.468320886533366e-06, "loss": 0.5204, "step": 5370 }, { "epoch": 0.65, "learning_rate": 6.480366176824862e-06, "loss": 0.6218, "step": 5380 }, { "epoch": 0.65, "learning_rate": 6.492411467116358e-06, "loss": 0.5053, "step": 5390 }, { "epoch": 0.65, "learning_rate": 6.504456757407854e-06, "loss": 0.5456, "step": 5400 }, { "epoch": 0.65, "learning_rate": 6.516502047699349e-06, "loss": 0.6857, "step": 5410 }, { "epoch": 0.65, "learning_rate": 6.528547337990847e-06, "loss": 0.6966, "step": 5420 }, { "epoch": 0.65, "learning_rate": 6.540592628282342e-06, "loss": 0.3692, "step": 5430 }, { "epoch": 0.66, "learning_rate": 6.552637918573838e-06, "loss": 0.5127, "step": 5440 }, { "epoch": 0.66, "learning_rate": 6.564683208865334e-06, "loss": 0.538, "step": 5450 }, { "epoch": 0.66, "learning_rate": 6.57672849915683e-06, "loss": 0.4563, "step": 5460 }, { "epoch": 0.66, "learning_rate": 6.588773789448325e-06, "loss": 0.3748, "step": 5470 }, { "epoch": 0.66, "learning_rate": 6.600819079739821e-06, "loss": 0.4877, "step": 5480 }, { "epoch": 0.66, "learning_rate": 6.612864370031319e-06, "loss": 0.6102, "step": 5490 }, { "epoch": 0.66, "learning_rate": 6.624909660322814e-06, "loss": 0.753, "step": 5500 }, { "epoch": 0.66, "learning_rate": 6.63695495061431e-06, "loss": 0.5277, "step": 5510 }, { "epoch": 0.66, "learning_rate": 6.649000240905806e-06, "loss": 0.5599, "step": 5520 }, { "epoch": 0.67, "learning_rate": 6.661045531197302e-06, "loss": 0.4621, "step": 5530 }, { "epoch": 0.67, "learning_rate": 6.6730908214887975e-06, "loss": 0.3551, "step": 5540 }, { "epoch": 0.67, "learning_rate": 6.685136111780295e-06, "loss": 0.467, "step": 5550 }, { "epoch": 0.67, "learning_rate": 6.69718140207179e-06, "loss": 0.3381, "step": 5560 }, { "epoch": 0.67, "learning_rate": 6.709226692363286e-06, "loss": 0.5466, "step": 5570 }, { "epoch": 0.67, "learning_rate": 6.7212719826547824e-06, "loss": 0.4579, "step": 5580 }, { "epoch": 0.67, "learning_rate": 6.7333172729462785e-06, "loss": 0.407, "step": 5590 }, { "epoch": 0.67, "learning_rate": 6.745362563237774e-06, "loss": 0.4037, "step": 5600 }, { "epoch": 0.68, "learning_rate": 6.75740785352927e-06, "loss": 0.5914, "step": 5610 }, { "epoch": 0.68, "learning_rate": 6.769453143820767e-06, "loss": 0.4636, "step": 5620 }, { "epoch": 0.68, "learning_rate": 6.781498434112263e-06, "loss": 0.4421, "step": 5630 }, { "epoch": 0.68, "learning_rate": 6.793543724403759e-06, "loss": 0.4942, "step": 5640 }, { "epoch": 0.68, "learning_rate": 6.805589014695255e-06, "loss": 0.6613, "step": 5650 }, { "epoch": 0.68, "learning_rate": 6.81763430498675e-06, "loss": 0.3269, "step": 5660 }, { "epoch": 0.68, "learning_rate": 6.829679595278246e-06, "loss": 0.4642, "step": 5670 }, { "epoch": 0.68, "learning_rate": 6.841724885569742e-06, "loss": 0.3119, "step": 5680 }, { "epoch": 0.69, "learning_rate": 6.853770175861239e-06, "loss": 0.4078, "step": 5690 }, { "epoch": 0.69, "learning_rate": 6.865815466152735e-06, "loss": 0.4629, "step": 5700 }, { "epoch": 0.69, "learning_rate": 6.877860756444231e-06, "loss": 0.4968, "step": 5710 }, { "epoch": 0.69, "learning_rate": 6.889906046735727e-06, "loss": 0.6665, "step": 5720 }, { "epoch": 0.69, "learning_rate": 6.901951337027222e-06, "loss": 0.3961, "step": 5730 }, { "epoch": 0.69, "learning_rate": 6.913996627318718e-06, "loss": 0.3748, "step": 5740 }, { "epoch": 0.69, "learning_rate": 6.926041917610215e-06, "loss": 0.4046, "step": 5750 }, { "epoch": 0.69, "learning_rate": 6.938087207901711e-06, "loss": 0.4587, "step": 5760 }, { "epoch": 0.69, "learning_rate": 6.950132498193207e-06, "loss": 0.4221, "step": 5770 }, { "epoch": 0.7, "learning_rate": 6.962177788484703e-06, "loss": 0.5593, "step": 5780 }, { "epoch": 0.7, "learning_rate": 6.974223078776198e-06, "loss": 0.4419, "step": 5790 }, { "epoch": 0.7, "learning_rate": 6.986268369067694e-06, "loss": 0.4444, "step": 5800 }, { "epoch": 0.7, "learning_rate": 6.99831365935919e-06, "loss": 0.4194, "step": 5810 }, { "epoch": 0.7, "learning_rate": 7.010358949650687e-06, "loss": 0.4815, "step": 5820 }, { "epoch": 0.7, "learning_rate": 7.022404239942183e-06, "loss": 0.4017, "step": 5830 }, { "epoch": 0.7, "learning_rate": 7.034449530233679e-06, "loss": 0.5477, "step": 5840 }, { "epoch": 0.7, "learning_rate": 7.046494820525175e-06, "loss": 0.4697, "step": 5850 }, { "epoch": 0.71, "learning_rate": 7.0585401108166705e-06, "loss": 0.3987, "step": 5860 }, { "epoch": 0.71, "learning_rate": 7.0705854011081665e-06, "loss": 0.519, "step": 5870 }, { "epoch": 0.71, "learning_rate": 7.082630691399663e-06, "loss": 0.4937, "step": 5880 }, { "epoch": 0.71, "learning_rate": 7.0946759816911594e-06, "loss": 0.4815, "step": 5890 }, { "epoch": 0.71, "learning_rate": 7.1067212719826555e-06, "loss": 0.2911, "step": 5900 }, { "epoch": 0.71, "learning_rate": 7.1187665622741515e-06, "loss": 0.4578, "step": 5910 }, { "epoch": 0.71, "learning_rate": 7.130811852565647e-06, "loss": 0.5965, "step": 5920 }, { "epoch": 0.71, "learning_rate": 7.142857142857143e-06, "loss": 0.4649, "step": 5930 }, { "epoch": 0.72, "learning_rate": 7.154902433148639e-06, "loss": 0.2398, "step": 5940 }, { "epoch": 0.72, "learning_rate": 7.166947723440136e-06, "loss": 0.3879, "step": 5950 }, { "epoch": 0.72, "learning_rate": 7.178993013731632e-06, "loss": 0.4985, "step": 5960 }, { "epoch": 0.72, "learning_rate": 7.191038304023128e-06, "loss": 0.5634, "step": 5970 }, { "epoch": 0.72, "learning_rate": 7.203083594314623e-06, "loss": 0.5043, "step": 5980 }, { "epoch": 0.72, "learning_rate": 7.215128884606119e-06, "loss": 0.4324, "step": 5990 }, { "epoch": 0.72, "learning_rate": 7.227174174897615e-06, "loss": 0.4749, "step": 6000 }, { "epoch": 0.72, "learning_rate": 7.239219465189112e-06, "loss": 0.5217, "step": 6010 }, { "epoch": 0.73, "learning_rate": 7.251264755480608e-06, "loss": 0.4153, "step": 6020 }, { "epoch": 0.73, "learning_rate": 7.263310045772104e-06, "loss": 0.4597, "step": 6030 }, { "epoch": 0.73, "learning_rate": 7.2753553360636e-06, "loss": 0.5156, "step": 6040 }, { "epoch": 0.73, "learning_rate": 7.287400626355095e-06, "loss": 0.3854, "step": 6050 }, { "epoch": 0.73, "learning_rate": 7.299445916646591e-06, "loss": 0.4904, "step": 6060 }, { "epoch": 0.73, "learning_rate": 7.311491206938087e-06, "loss": 0.2574, "step": 6070 }, { "epoch": 0.73, "learning_rate": 7.323536497229584e-06, "loss": 0.361, "step": 6080 }, { "epoch": 0.73, "learning_rate": 7.33558178752108e-06, "loss": 0.5998, "step": 6090 }, { "epoch": 0.73, "learning_rate": 7.347627077812576e-06, "loss": 0.444, "step": 6100 }, { "epoch": 0.74, "learning_rate": 7.359672368104071e-06, "loss": 0.4503, "step": 6110 }, { "epoch": 0.74, "learning_rate": 7.371717658395567e-06, "loss": 0.5508, "step": 6120 }, { "epoch": 0.74, "learning_rate": 7.383762948687063e-06, "loss": 0.4293, "step": 6130 }, { "epoch": 0.74, "learning_rate": 7.395808238978559e-06, "loss": 0.5193, "step": 6140 }, { "epoch": 0.74, "learning_rate": 7.407853529270056e-06, "loss": 0.6465, "step": 6150 }, { "epoch": 0.74, "learning_rate": 7.419898819561552e-06, "loss": 0.3654, "step": 6160 }, { "epoch": 0.74, "learning_rate": 7.4319441098530475e-06, "loss": 0.5274, "step": 6170 }, { "epoch": 0.74, "learning_rate": 7.4439894001445435e-06, "loss": 0.5087, "step": 6180 }, { "epoch": 0.75, "learning_rate": 7.4560346904360395e-06, "loss": 0.4028, "step": 6190 }, { "epoch": 0.75, "learning_rate": 7.468079980727536e-06, "loss": 0.5037, "step": 6200 }, { "epoch": 0.75, "learning_rate": 7.4801252710190325e-06, "loss": 0.5378, "step": 6210 }, { "epoch": 0.75, "learning_rate": 7.4921705613105285e-06, "loss": 0.4784, "step": 6220 }, { "epoch": 0.75, "learning_rate": 7.5042158516020245e-06, "loss": 0.5758, "step": 6230 }, { "epoch": 0.75, "learning_rate": 7.51626114189352e-06, "loss": 0.4719, "step": 6240 }, { "epoch": 0.75, "learning_rate": 7.528306432185016e-06, "loss": 0.4406, "step": 6250 }, { "epoch": 0.75, "learning_rate": 7.540351722476512e-06, "loss": 0.5498, "step": 6260 }, { "epoch": 0.76, "learning_rate": 7.552397012768008e-06, "loss": 0.3951, "step": 6270 }, { "epoch": 0.76, "learning_rate": 7.564442303059505e-06, "loss": 0.4104, "step": 6280 }, { "epoch": 0.76, "learning_rate": 7.576487593351001e-06, "loss": 0.41, "step": 6290 }, { "epoch": 0.76, "learning_rate": 7.588532883642496e-06, "loss": 0.4746, "step": 6300 }, { "epoch": 0.76, "learning_rate": 7.600578173933992e-06, "loss": 0.4351, "step": 6310 }, { "epoch": 0.76, "learning_rate": 7.612623464225488e-06, "loss": 0.4973, "step": 6320 }, { "epoch": 0.76, "learning_rate": 7.624668754516984e-06, "loss": 0.5322, "step": 6330 }, { "epoch": 0.76, "learning_rate": 7.636714044808481e-06, "loss": 0.5684, "step": 6340 }, { "epoch": 0.76, "learning_rate": 7.648759335099976e-06, "loss": 0.4497, "step": 6350 }, { "epoch": 0.77, "learning_rate": 7.660804625391473e-06, "loss": 0.5994, "step": 6360 }, { "epoch": 0.77, "learning_rate": 7.672849915682968e-06, "loss": 0.4334, "step": 6370 }, { "epoch": 0.77, "learning_rate": 7.684895205974465e-06, "loss": 0.5416, "step": 6380 }, { "epoch": 0.77, "learning_rate": 7.69694049626596e-06, "loss": 0.3739, "step": 6390 }, { "epoch": 0.77, "learning_rate": 7.708985786557455e-06, "loss": 0.4078, "step": 6400 }, { "epoch": 0.77, "learning_rate": 7.721031076848952e-06, "loss": 0.4728, "step": 6410 }, { "epoch": 0.77, "learning_rate": 7.733076367140449e-06, "loss": 0.4738, "step": 6420 }, { "epoch": 0.77, "learning_rate": 7.745121657431944e-06, "loss": 0.3702, "step": 6430 }, { "epoch": 0.78, "learning_rate": 7.757166947723441e-06, "loss": 0.3881, "step": 6440 }, { "epoch": 0.78, "learning_rate": 7.769212238014936e-06, "loss": 0.3192, "step": 6450 }, { "epoch": 0.78, "learning_rate": 7.781257528306432e-06, "loss": 0.3915, "step": 6460 }, { "epoch": 0.78, "learning_rate": 7.793302818597928e-06, "loss": 0.4486, "step": 6470 }, { "epoch": 0.78, "learning_rate": 7.805348108889425e-06, "loss": 0.4052, "step": 6480 }, { "epoch": 0.78, "learning_rate": 7.81739339918092e-06, "loss": 0.5315, "step": 6490 }, { "epoch": 0.78, "learning_rate": 7.829438689472417e-06, "loss": 0.4918, "step": 6500 }, { "epoch": 0.78, "learning_rate": 7.841483979763913e-06, "loss": 0.4985, "step": 6510 }, { "epoch": 0.79, "learning_rate": 7.853529270055408e-06, "loss": 0.412, "step": 6520 }, { "epoch": 0.79, "learning_rate": 7.865574560346905e-06, "loss": 0.4001, "step": 6530 }, { "epoch": 0.79, "learning_rate": 7.877619850638402e-06, "loss": 0.382, "step": 6540 }, { "epoch": 0.79, "learning_rate": 7.889665140929897e-06, "loss": 0.3336, "step": 6550 }, { "epoch": 0.79, "learning_rate": 7.901710431221394e-06, "loss": 0.4339, "step": 6560 }, { "epoch": 0.79, "learning_rate": 7.913755721512889e-06, "loss": 0.5157, "step": 6570 }, { "epoch": 0.79, "learning_rate": 7.925801011804384e-06, "loss": 0.5192, "step": 6580 }, { "epoch": 0.79, "learning_rate": 7.93784630209588e-06, "loss": 0.366, "step": 6590 }, { "epoch": 0.79, "learning_rate": 7.949891592387376e-06, "loss": 0.3981, "step": 6600 }, { "epoch": 0.8, "learning_rate": 7.961936882678873e-06, "loss": 0.4308, "step": 6610 }, { "epoch": 0.8, "learning_rate": 7.97398217297037e-06, "loss": 0.5161, "step": 6620 }, { "epoch": 0.8, "learning_rate": 7.986027463261865e-06, "loss": 0.2851, "step": 6630 }, { "epoch": 0.8, "learning_rate": 7.99807275355336e-06, "loss": 0.4614, "step": 6640 }, { "epoch": 0.8, "learning_rate": 8.010118043844857e-06, "loss": 0.5586, "step": 6650 }, { "epoch": 0.8, "learning_rate": 8.022163334136352e-06, "loss": 0.338, "step": 6660 }, { "epoch": 0.8, "learning_rate": 8.034208624427849e-06, "loss": 0.3199, "step": 6670 }, { "epoch": 0.8, "learning_rate": 8.046253914719346e-06, "loss": 0.2405, "step": 6680 }, { "epoch": 0.81, "learning_rate": 8.058299205010841e-06, "loss": 0.3322, "step": 6690 }, { "epoch": 0.81, "learning_rate": 8.070344495302336e-06, "loss": 0.4648, "step": 6700 }, { "epoch": 0.81, "learning_rate": 8.082389785593833e-06, "loss": 0.5113, "step": 6710 }, { "epoch": 0.81, "learning_rate": 8.094435075885328e-06, "loss": 0.6286, "step": 6720 }, { "epoch": 0.81, "learning_rate": 8.106480366176825e-06, "loss": 0.4101, "step": 6730 }, { "epoch": 0.81, "learning_rate": 8.118525656468322e-06, "loss": 0.5602, "step": 6740 }, { "epoch": 0.81, "learning_rate": 8.130570946759817e-06, "loss": 0.3852, "step": 6750 }, { "epoch": 0.81, "learning_rate": 8.142616237051314e-06, "loss": 0.4589, "step": 6760 }, { "epoch": 0.82, "learning_rate": 8.15466152734281e-06, "loss": 0.6245, "step": 6770 }, { "epoch": 0.82, "learning_rate": 8.166706817634305e-06, "loss": 0.5513, "step": 6780 }, { "epoch": 0.82, "learning_rate": 8.178752107925801e-06, "loss": 0.3232, "step": 6790 }, { "epoch": 0.82, "learning_rate": 8.190797398217298e-06, "loss": 0.4243, "step": 6800 }, { "epoch": 0.82, "learning_rate": 8.202842688508794e-06, "loss": 0.3588, "step": 6810 }, { "epoch": 0.82, "learning_rate": 8.21488797880029e-06, "loss": 0.4692, "step": 6820 }, { "epoch": 0.82, "learning_rate": 8.226933269091786e-06, "loss": 0.3904, "step": 6830 }, { "epoch": 0.82, "learning_rate": 8.23897855938328e-06, "loss": 0.2821, "step": 6840 }, { "epoch": 0.83, "learning_rate": 8.251023849674778e-06, "loss": 0.4965, "step": 6850 }, { "epoch": 0.83, "learning_rate": 8.263069139966273e-06, "loss": 0.3623, "step": 6860 }, { "epoch": 0.83, "learning_rate": 8.27511443025777e-06, "loss": 0.408, "step": 6870 }, { "epoch": 0.83, "learning_rate": 8.287159720549267e-06, "loss": 0.2602, "step": 6880 }, { "epoch": 0.83, "learning_rate": 8.299205010840762e-06, "loss": 0.412, "step": 6890 }, { "epoch": 0.83, "learning_rate": 8.311250301132257e-06, "loss": 0.5776, "step": 6900 }, { "epoch": 0.83, "learning_rate": 8.323295591423754e-06, "loss": 0.355, "step": 6910 }, { "epoch": 0.83, "learning_rate": 8.335340881715249e-06, "loss": 0.4796, "step": 6920 }, { "epoch": 0.83, "learning_rate": 8.347386172006746e-06, "loss": 0.4038, "step": 6930 }, { "epoch": 0.84, "learning_rate": 8.359431462298243e-06, "loss": 0.3637, "step": 6940 }, { "epoch": 0.84, "learning_rate": 8.371476752589738e-06, "loss": 0.5986, "step": 6950 }, { "epoch": 0.84, "learning_rate": 8.383522042881233e-06, "loss": 0.5413, "step": 6960 }, { "epoch": 0.84, "learning_rate": 8.39556733317273e-06, "loss": 0.5352, "step": 6970 }, { "epoch": 0.84, "learning_rate": 8.407612623464225e-06, "loss": 0.4325, "step": 6980 }, { "epoch": 0.84, "learning_rate": 8.419657913755722e-06, "loss": 0.2381, "step": 6990 }, { "epoch": 0.84, "learning_rate": 8.431703204047219e-06, "loss": 0.3547, "step": 7000 }, { "epoch": 0.84, "learning_rate": 8.443748494338714e-06, "loss": 0.4185, "step": 7010 }, { "epoch": 0.85, "learning_rate": 8.45579378463021e-06, "loss": 0.5981, "step": 7020 }, { "epoch": 0.85, "learning_rate": 8.467839074921706e-06, "loss": 0.3505, "step": 7030 }, { "epoch": 0.85, "learning_rate": 8.479884365213201e-06, "loss": 0.3978, "step": 7040 }, { "epoch": 0.85, "learning_rate": 8.491929655504698e-06, "loss": 0.6432, "step": 7050 }, { "epoch": 0.85, "learning_rate": 8.503974945796193e-06, "loss": 0.5061, "step": 7060 }, { "epoch": 0.85, "learning_rate": 8.51602023608769e-06, "loss": 0.4733, "step": 7070 }, { "epoch": 0.85, "learning_rate": 8.528065526379187e-06, "loss": 0.4144, "step": 7080 }, { "epoch": 0.85, "learning_rate": 8.540110816670682e-06, "loss": 0.3947, "step": 7090 }, { "epoch": 0.86, "learning_rate": 8.552156106962178e-06, "loss": 0.4079, "step": 7100 }, { "epoch": 0.86, "learning_rate": 8.564201397253674e-06, "loss": 0.3967, "step": 7110 }, { "epoch": 0.86, "learning_rate": 8.57624668754517e-06, "loss": 0.5846, "step": 7120 }, { "epoch": 0.86, "learning_rate": 8.588291977836667e-06, "loss": 0.4633, "step": 7130 }, { "epoch": 0.86, "learning_rate": 8.600337268128163e-06, "loss": 0.4765, "step": 7140 }, { "epoch": 0.86, "learning_rate": 8.612382558419659e-06, "loss": 0.4172, "step": 7150 }, { "epoch": 0.86, "learning_rate": 8.624427848711154e-06, "loss": 0.3337, "step": 7160 }, { "epoch": 0.86, "learning_rate": 8.63647313900265e-06, "loss": 0.356, "step": 7170 }, { "epoch": 0.86, "learning_rate": 8.648518429294146e-06, "loss": 0.3738, "step": 7180 }, { "epoch": 0.87, "learning_rate": 8.660563719585641e-06, "loss": 0.5098, "step": 7190 }, { "epoch": 0.87, "learning_rate": 8.67260900987714e-06, "loss": 0.3821, "step": 7200 }, { "epoch": 0.87, "learning_rate": 8.684654300168635e-06, "loss": 0.5555, "step": 7210 }, { "epoch": 0.87, "learning_rate": 8.69669959046013e-06, "loss": 0.6138, "step": 7220 }, { "epoch": 0.87, "learning_rate": 8.708744880751627e-06, "loss": 0.2638, "step": 7230 }, { "epoch": 0.87, "learning_rate": 8.720790171043122e-06, "loss": 0.3204, "step": 7240 }, { "epoch": 0.87, "learning_rate": 8.732835461334617e-06, "loss": 0.4521, "step": 7250 }, { "epoch": 0.87, "learning_rate": 8.744880751626116e-06, "loss": 0.3897, "step": 7260 }, { "epoch": 0.88, "learning_rate": 8.756926041917611e-06, "loss": 0.3247, "step": 7270 }, { "epoch": 0.88, "learning_rate": 8.768971332209106e-06, "loss": 0.3119, "step": 7280 }, { "epoch": 0.88, "learning_rate": 8.781016622500603e-06, "loss": 0.3051, "step": 7290 }, { "epoch": 0.88, "learning_rate": 8.793061912792098e-06, "loss": 0.4196, "step": 7300 }, { "epoch": 0.88, "learning_rate": 8.805107203083595e-06, "loss": 0.4143, "step": 7310 }, { "epoch": 0.88, "learning_rate": 8.81715249337509e-06, "loss": 0.4781, "step": 7320 }, { "epoch": 0.88, "learning_rate": 8.829197783666587e-06, "loss": 0.4575, "step": 7330 }, { "epoch": 0.88, "learning_rate": 8.841243073958082e-06, "loss": 0.4867, "step": 7340 }, { "epoch": 0.89, "learning_rate": 8.85328836424958e-06, "loss": 0.4684, "step": 7350 }, { "epoch": 0.89, "learning_rate": 8.865333654541074e-06, "loss": 0.3675, "step": 7360 }, { "epoch": 0.89, "learning_rate": 8.877378944832571e-06, "loss": 0.5028, "step": 7370 }, { "epoch": 0.89, "learning_rate": 8.889424235124067e-06, "loss": 0.398, "step": 7380 }, { "epoch": 0.89, "learning_rate": 8.901469525415563e-06, "loss": 0.5037, "step": 7390 }, { "epoch": 0.89, "learning_rate": 8.91351481570706e-06, "loss": 0.3101, "step": 7400 }, { "epoch": 0.89, "learning_rate": 8.925560105998555e-06, "loss": 0.4027, "step": 7410 }, { "epoch": 0.89, "learning_rate": 8.93760539629005e-06, "loss": 0.4091, "step": 7420 }, { "epoch": 0.89, "learning_rate": 8.949650686581548e-06, "loss": 0.3533, "step": 7430 }, { "epoch": 0.9, "learning_rate": 8.961695976873043e-06, "loss": 0.4538, "step": 7440 }, { "epoch": 0.9, "learning_rate": 8.973741267164538e-06, "loss": 0.5103, "step": 7450 }, { "epoch": 0.9, "learning_rate": 8.985786557456036e-06, "loss": 0.5349, "step": 7460 }, { "epoch": 0.9, "learning_rate": 8.997831847747532e-06, "loss": 0.3082, "step": 7470 }, { "epoch": 0.9, "learning_rate": 9.009877138039027e-06, "loss": 0.4585, "step": 7480 }, { "epoch": 0.9, "learning_rate": 9.021922428330524e-06, "loss": 0.5309, "step": 7490 }, { "epoch": 0.9, "learning_rate": 9.033967718622019e-06, "loss": 0.3636, "step": 7500 }, { "epoch": 0.9, "learning_rate": 9.046013008913514e-06, "loss": 0.4219, "step": 7510 }, { "epoch": 0.91, "learning_rate": 9.058058299205011e-06, "loss": 0.3823, "step": 7520 }, { "epoch": 0.91, "learning_rate": 9.070103589496508e-06, "loss": 0.3659, "step": 7530 }, { "epoch": 0.91, "learning_rate": 9.082148879788003e-06, "loss": 0.6386, "step": 7540 }, { "epoch": 0.91, "learning_rate": 9.0941941700795e-06, "loss": 0.3494, "step": 7550 }, { "epoch": 0.91, "learning_rate": 9.106239460370995e-06, "loss": 0.307, "step": 7560 }, { "epoch": 0.91, "learning_rate": 9.11828475066249e-06, "loss": 0.4388, "step": 7570 }, { "epoch": 0.91, "learning_rate": 9.130330040953987e-06, "loss": 0.3636, "step": 7580 }, { "epoch": 0.91, "learning_rate": 9.142375331245484e-06, "loss": 0.3525, "step": 7590 }, { "epoch": 0.92, "learning_rate": 9.15442062153698e-06, "loss": 0.5595, "step": 7600 }, { "epoch": 0.92, "learning_rate": 9.166465911828476e-06, "loss": 0.5447, "step": 7610 }, { "epoch": 0.92, "learning_rate": 9.178511202119971e-06, "loss": 0.4755, "step": 7620 }, { "epoch": 0.92, "learning_rate": 9.190556492411468e-06, "loss": 0.3651, "step": 7630 }, { "epoch": 0.92, "learning_rate": 9.202601782702963e-06, "loss": 0.3949, "step": 7640 }, { "epoch": 0.92, "learning_rate": 9.214647072994459e-06, "loss": 0.4126, "step": 7650 }, { "epoch": 0.92, "learning_rate": 9.226692363285955e-06, "loss": 0.5115, "step": 7660 }, { "epoch": 0.92, "learning_rate": 9.238737653577452e-06, "loss": 0.3829, "step": 7670 }, { "epoch": 0.93, "learning_rate": 9.250782943868947e-06, "loss": 0.4345, "step": 7680 }, { "epoch": 0.93, "learning_rate": 9.262828234160444e-06, "loss": 0.4506, "step": 7690 }, { "epoch": 0.93, "learning_rate": 9.27487352445194e-06, "loss": 0.5201, "step": 7700 }, { "epoch": 0.93, "learning_rate": 9.286918814743435e-06, "loss": 0.6001, "step": 7710 }, { "epoch": 0.93, "learning_rate": 9.298964105034932e-06, "loss": 0.3452, "step": 7720 }, { "epoch": 0.93, "learning_rate": 9.311009395326428e-06, "loss": 0.5694, "step": 7730 }, { "epoch": 0.93, "learning_rate": 9.323054685617924e-06, "loss": 0.4353, "step": 7740 }, { "epoch": 0.93, "learning_rate": 9.33509997590942e-06, "loss": 0.4726, "step": 7750 }, { "epoch": 0.93, "learning_rate": 9.347145266200916e-06, "loss": 0.4451, "step": 7760 }, { "epoch": 0.94, "learning_rate": 9.359190556492411e-06, "loss": 0.3735, "step": 7770 }, { "epoch": 0.94, "learning_rate": 9.371235846783908e-06, "loss": 0.51, "step": 7780 }, { "epoch": 0.94, "learning_rate": 9.383281137075405e-06, "loss": 0.5572, "step": 7790 }, { "epoch": 0.94, "learning_rate": 9.3953264273669e-06, "loss": 0.5198, "step": 7800 }, { "epoch": 0.94, "learning_rate": 9.407371717658397e-06, "loss": 0.4625, "step": 7810 }, { "epoch": 0.94, "learning_rate": 9.419417007949892e-06, "loss": 0.4026, "step": 7820 }, { "epoch": 0.94, "learning_rate": 9.431462298241387e-06, "loss": 0.4422, "step": 7830 }, { "epoch": 0.94, "learning_rate": 9.443507588532884e-06, "loss": 0.3775, "step": 7840 }, { "epoch": 0.95, "learning_rate": 9.455552878824381e-06, "loss": 0.395, "step": 7850 }, { "epoch": 0.95, "learning_rate": 9.467598169115876e-06, "loss": 0.3672, "step": 7860 }, { "epoch": 0.95, "learning_rate": 9.479643459407373e-06, "loss": 0.4422, "step": 7870 }, { "epoch": 0.95, "learning_rate": 9.491688749698868e-06, "loss": 0.2415, "step": 7880 }, { "epoch": 0.95, "learning_rate": 9.503734039990363e-06, "loss": 0.3443, "step": 7890 }, { "epoch": 0.95, "learning_rate": 9.51577933028186e-06, "loss": 0.2497, "step": 7900 }, { "epoch": 0.95, "learning_rate": 9.527824620573355e-06, "loss": 0.4515, "step": 7910 }, { "epoch": 0.95, "learning_rate": 9.539869910864852e-06, "loss": 0.3656, "step": 7920 }, { "epoch": 0.96, "learning_rate": 9.551915201156349e-06, "loss": 0.3656, "step": 7930 }, { "epoch": 0.96, "learning_rate": 9.563960491447844e-06, "loss": 0.3278, "step": 7940 }, { "epoch": 0.96, "learning_rate": 9.57600578173934e-06, "loss": 0.3987, "step": 7950 }, { "epoch": 0.96, "learning_rate": 9.588051072030836e-06, "loss": 0.4519, "step": 7960 }, { "epoch": 0.96, "learning_rate": 9.600096362322332e-06, "loss": 0.3105, "step": 7970 }, { "epoch": 0.96, "learning_rate": 9.612141652613828e-06, "loss": 0.395, "step": 7980 }, { "epoch": 0.96, "learning_rate": 9.624186942905325e-06, "loss": 0.3364, "step": 7990 }, { "epoch": 0.96, "learning_rate": 9.63623223319682e-06, "loss": 0.3145, "step": 8000 }, { "epoch": 0.96, "learning_rate": 9.648277523488317e-06, "loss": 0.4203, "step": 8010 }, { "epoch": 0.97, "learning_rate": 9.660322813779813e-06, "loss": 0.3641, "step": 8020 }, { "epoch": 0.97, "learning_rate": 9.672368104071308e-06, "loss": 0.2783, "step": 8030 }, { "epoch": 0.97, "learning_rate": 9.684413394362805e-06, "loss": 0.3757, "step": 8040 }, { "epoch": 0.97, "learning_rate": 9.696458684654301e-06, "loss": 0.2093, "step": 8050 }, { "epoch": 0.97, "learning_rate": 9.708503974945797e-06, "loss": 0.3385, "step": 8060 }, { "epoch": 0.97, "learning_rate": 9.720549265237294e-06, "loss": 0.3739, "step": 8070 }, { "epoch": 0.97, "learning_rate": 9.732594555528789e-06, "loss": 0.2254, "step": 8080 }, { "epoch": 0.97, "learning_rate": 9.744639845820284e-06, "loss": 0.5837, "step": 8090 }, { "epoch": 0.98, "learning_rate": 9.75668513611178e-06, "loss": 0.4332, "step": 8100 }, { "epoch": 0.98, "learning_rate": 9.768730426403276e-06, "loss": 0.4345, "step": 8110 }, { "epoch": 0.98, "learning_rate": 9.780775716694773e-06, "loss": 0.3919, "step": 8120 }, { "epoch": 0.98, "learning_rate": 9.79282100698627e-06, "loss": 0.3164, "step": 8130 }, { "epoch": 0.98, "learning_rate": 9.804866297277765e-06, "loss": 0.6981, "step": 8140 }, { "epoch": 0.98, "learning_rate": 9.81691158756926e-06, "loss": 0.3197, "step": 8150 }, { "epoch": 0.98, "learning_rate": 9.828956877860757e-06, "loss": 0.4367, "step": 8160 }, { "epoch": 0.98, "learning_rate": 9.841002168152252e-06, "loss": 0.2865, "step": 8170 }, { "epoch": 0.99, "learning_rate": 9.853047458443749e-06, "loss": 0.5013, "step": 8180 }, { "epoch": 0.99, "learning_rate": 9.865092748735246e-06, "loss": 0.3213, "step": 8190 }, { "epoch": 0.99, "learning_rate": 9.877138039026741e-06, "loss": 0.4613, "step": 8200 }, { "epoch": 0.99, "learning_rate": 9.889183329318236e-06, "loss": 0.4298, "step": 8210 }, { "epoch": 0.99, "learning_rate": 9.901228619609733e-06, "loss": 0.3051, "step": 8220 }, { "epoch": 0.99, "learning_rate": 9.913273909901228e-06, "loss": 0.3338, "step": 8230 }, { "epoch": 0.99, "learning_rate": 9.925319200192725e-06, "loss": 0.4111, "step": 8240 }, { "epoch": 0.99, "learning_rate": 9.937364490484222e-06, "loss": 0.4244, "step": 8250 }, { "epoch": 0.99, "learning_rate": 9.949409780775717e-06, "loss": 0.313, "step": 8260 }, { "epoch": 1.0, "learning_rate": 9.961455071067212e-06, "loss": 0.1936, "step": 8270 }, { "epoch": 1.0, "learning_rate": 9.97350036135871e-06, "loss": 0.4937, "step": 8280 }, { "epoch": 1.0, "learning_rate": 9.985545651650205e-06, "loss": 0.3544, "step": 8290 }, { "epoch": 1.0, "learning_rate": 9.997590941941701e-06, "loss": 0.37, "step": 8300 }, { "epoch": 1.0, "eval_accuracy": { "accuracy": 0.8933453778982234 }, "eval_f1": { "f1": 0.8942100052466936 }, "eval_loss": 0.34615740180015564, "eval_precision": { "precision": 0.8984250247518094 }, "eval_recall": { "recall": 0.8931370564158605 }, "eval_runtime": 655.829, "eval_samples_per_second": 25.319, "eval_steps_per_second": 12.66, "step": 8302 }, { "epoch": 1.0, "learning_rate": 1.0009636232233198e-05, "loss": 0.3992, "step": 8310 }, { "epoch": 1.0, "learning_rate": 1.0021681522524694e-05, "loss": 0.4586, "step": 8320 }, { "epoch": 1.0, "learning_rate": 1.003372681281619e-05, "loss": 0.3856, "step": 8330 }, { "epoch": 1.0, "learning_rate": 1.0045772103107686e-05, "loss": 0.6489, "step": 8340 }, { "epoch": 1.01, "learning_rate": 1.005781739339918e-05, "loss": 0.3425, "step": 8350 }, { "epoch": 1.01, "learning_rate": 1.0069862683690678e-05, "loss": 0.2965, "step": 8360 }, { "epoch": 1.01, "learning_rate": 1.0081907973982173e-05, "loss": 0.4241, "step": 8370 }, { "epoch": 1.01, "learning_rate": 1.009395326427367e-05, "loss": 0.6012, "step": 8380 }, { "epoch": 1.01, "learning_rate": 1.0105998554565167e-05, "loss": 0.2937, "step": 8390 }, { "epoch": 1.01, "learning_rate": 1.0118043844856662e-05, "loss": 0.4138, "step": 8400 }, { "epoch": 1.01, "learning_rate": 1.0130089135148157e-05, "loss": 0.449, "step": 8410 }, { "epoch": 1.01, "learning_rate": 1.0142134425439654e-05, "loss": 0.5532, "step": 8420 }, { "epoch": 1.02, "learning_rate": 1.0154179715731149e-05, "loss": 0.3685, "step": 8430 }, { "epoch": 1.02, "learning_rate": 1.0166225006022646e-05, "loss": 0.3075, "step": 8440 }, { "epoch": 1.02, "learning_rate": 1.0178270296314143e-05, "loss": 0.4835, "step": 8450 }, { "epoch": 1.02, "learning_rate": 1.0190315586605638e-05, "loss": 0.3404, "step": 8460 }, { "epoch": 1.02, "learning_rate": 1.0202360876897133e-05, "loss": 0.385, "step": 8470 }, { "epoch": 1.02, "learning_rate": 1.021440616718863e-05, "loss": 0.4614, "step": 8480 }, { "epoch": 1.02, "learning_rate": 1.0226451457480125e-05, "loss": 0.6181, "step": 8490 }, { "epoch": 1.02, "learning_rate": 1.023849674777162e-05, "loss": 0.3392, "step": 8500 }, { "epoch": 1.02, "learning_rate": 1.0250542038063119e-05, "loss": 0.3725, "step": 8510 }, { "epoch": 1.03, "learning_rate": 1.0262587328354614e-05, "loss": 0.4357, "step": 8520 }, { "epoch": 1.03, "learning_rate": 1.027463261864611e-05, "loss": 0.3347, "step": 8530 }, { "epoch": 1.03, "learning_rate": 1.0286677908937606e-05, "loss": 0.3535, "step": 8540 }, { "epoch": 1.03, "learning_rate": 1.0298723199229101e-05, "loss": 0.3758, "step": 8550 }, { "epoch": 1.03, "learning_rate": 1.0310768489520598e-05, "loss": 0.3641, "step": 8560 }, { "epoch": 1.03, "learning_rate": 1.0322813779812093e-05, "loss": 0.3299, "step": 8570 }, { "epoch": 1.03, "learning_rate": 1.033485907010359e-05, "loss": 0.4344, "step": 8580 }, { "epoch": 1.03, "learning_rate": 1.0346904360395086e-05, "loss": 0.2793, "step": 8590 }, { "epoch": 1.04, "learning_rate": 1.0358949650686582e-05, "loss": 0.4237, "step": 8600 }, { "epoch": 1.04, "learning_rate": 1.0370994940978078e-05, "loss": 0.2914, "step": 8610 }, { "epoch": 1.04, "learning_rate": 1.0383040231269574e-05, "loss": 0.3509, "step": 8620 }, { "epoch": 1.04, "learning_rate": 1.039508552156107e-05, "loss": 0.3521, "step": 8630 }, { "epoch": 1.04, "learning_rate": 1.0407130811852567e-05, "loss": 0.4525, "step": 8640 }, { "epoch": 1.04, "learning_rate": 1.0419176102144063e-05, "loss": 0.5443, "step": 8650 }, { "epoch": 1.04, "learning_rate": 1.0431221392435559e-05, "loss": 0.4707, "step": 8660 }, { "epoch": 1.04, "learning_rate": 1.0443266682727054e-05, "loss": 0.3739, "step": 8670 }, { "epoch": 1.05, "learning_rate": 1.045531197301855e-05, "loss": 0.2813, "step": 8680 }, { "epoch": 1.05, "learning_rate": 1.0467357263310046e-05, "loss": 0.4741, "step": 8690 }, { "epoch": 1.05, "learning_rate": 1.0479402553601541e-05, "loss": 0.2454, "step": 8700 }, { "epoch": 1.05, "learning_rate": 1.049144784389304e-05, "loss": 0.2791, "step": 8710 }, { "epoch": 1.05, "learning_rate": 1.0503493134184535e-05, "loss": 0.2423, "step": 8720 }, { "epoch": 1.05, "learning_rate": 1.051553842447603e-05, "loss": 0.4175, "step": 8730 }, { "epoch": 1.05, "learning_rate": 1.0527583714767527e-05, "loss": 0.4691, "step": 8740 }, { "epoch": 1.05, "learning_rate": 1.0539629005059022e-05, "loss": 0.2991, "step": 8750 }, { "epoch": 1.06, "learning_rate": 1.0551674295350517e-05, "loss": 0.4591, "step": 8760 }, { "epoch": 1.06, "learning_rate": 1.0563719585642016e-05, "loss": 0.458, "step": 8770 }, { "epoch": 1.06, "learning_rate": 1.0575764875933511e-05, "loss": 0.3077, "step": 8780 }, { "epoch": 1.06, "learning_rate": 1.0587810166225006e-05, "loss": 0.3674, "step": 8790 }, { "epoch": 1.06, "learning_rate": 1.0599855456516503e-05, "loss": 0.2467, "step": 8800 }, { "epoch": 1.06, "learning_rate": 1.0611900746807998e-05, "loss": 0.3926, "step": 8810 }, { "epoch": 1.06, "learning_rate": 1.0623946037099493e-05, "loss": 0.5028, "step": 8820 }, { "epoch": 1.06, "learning_rate": 1.063599132739099e-05, "loss": 0.3821, "step": 8830 }, { "epoch": 1.06, "learning_rate": 1.0648036617682487e-05, "loss": 0.3092, "step": 8840 }, { "epoch": 1.07, "learning_rate": 1.0660081907973982e-05, "loss": 0.5485, "step": 8850 }, { "epoch": 1.07, "learning_rate": 1.067212719826548e-05, "loss": 0.4605, "step": 8860 }, { "epoch": 1.07, "learning_rate": 1.0684172488556974e-05, "loss": 0.3811, "step": 8870 }, { "epoch": 1.07, "learning_rate": 1.0696217778848471e-05, "loss": 0.4402, "step": 8880 }, { "epoch": 1.07, "learning_rate": 1.0708263069139966e-05, "loss": 0.2989, "step": 8890 }, { "epoch": 1.07, "learning_rate": 1.0720308359431463e-05, "loss": 0.2415, "step": 8900 }, { "epoch": 1.07, "learning_rate": 1.0732353649722959e-05, "loss": 0.4552, "step": 8910 }, { "epoch": 1.07, "learning_rate": 1.0744398940014455e-05, "loss": 0.2925, "step": 8920 }, { "epoch": 1.08, "learning_rate": 1.075644423030595e-05, "loss": 0.4753, "step": 8930 }, { "epoch": 1.08, "learning_rate": 1.0768489520597447e-05, "loss": 0.4537, "step": 8940 }, { "epoch": 1.08, "learning_rate": 1.0780534810888943e-05, "loss": 0.3199, "step": 8950 }, { "epoch": 1.08, "learning_rate": 1.0792580101180438e-05, "loss": 0.3307, "step": 8960 }, { "epoch": 1.08, "learning_rate": 1.0804625391471936e-05, "loss": 0.5302, "step": 8970 }, { "epoch": 1.08, "learning_rate": 1.0816670681763432e-05, "loss": 0.4864, "step": 8980 }, { "epoch": 1.08, "learning_rate": 1.0828715972054927e-05, "loss": 0.3208, "step": 8990 }, { "epoch": 1.08, "learning_rate": 1.0840761262346424e-05, "loss": 0.52, "step": 9000 }, { "epoch": 1.09, "learning_rate": 1.0852806552637919e-05, "loss": 0.3761, "step": 9010 }, { "epoch": 1.09, "learning_rate": 1.0864851842929414e-05, "loss": 0.3495, "step": 9020 }, { "epoch": 1.09, "learning_rate": 1.0876897133220911e-05, "loss": 0.2601, "step": 9030 }, { "epoch": 1.09, "learning_rate": 1.0888942423512408e-05, "loss": 0.4162, "step": 9040 }, { "epoch": 1.09, "learning_rate": 1.0900987713803903e-05, "loss": 0.614, "step": 9050 }, { "epoch": 1.09, "learning_rate": 1.09130330040954e-05, "loss": 0.4303, "step": 9060 }, { "epoch": 1.09, "learning_rate": 1.0925078294386895e-05, "loss": 0.4174, "step": 9070 }, { "epoch": 1.09, "learning_rate": 1.093712358467839e-05, "loss": 0.2943, "step": 9080 }, { "epoch": 1.09, "learning_rate": 1.0949168874969887e-05, "loss": 0.3932, "step": 9090 }, { "epoch": 1.1, "learning_rate": 1.0961214165261384e-05, "loss": 0.3248, "step": 9100 }, { "epoch": 1.1, "learning_rate": 1.097325945555288e-05, "loss": 0.2313, "step": 9110 }, { "epoch": 1.1, "learning_rate": 1.0985304745844376e-05, "loss": 0.3972, "step": 9120 }, { "epoch": 1.1, "learning_rate": 1.0997350036135871e-05, "loss": 0.5109, "step": 9130 }, { "epoch": 1.1, "learning_rate": 1.1009395326427366e-05, "loss": 0.2809, "step": 9140 }, { "epoch": 1.1, "learning_rate": 1.1021440616718863e-05, "loss": 0.4447, "step": 9150 }, { "epoch": 1.1, "learning_rate": 1.1033485907010358e-05, "loss": 0.4777, "step": 9160 }, { "epoch": 1.1, "learning_rate": 1.1045531197301855e-05, "loss": 0.3876, "step": 9170 }, { "epoch": 1.11, "learning_rate": 1.1057576487593352e-05, "loss": 0.2433, "step": 9180 }, { "epoch": 1.11, "learning_rate": 1.1069621777884847e-05, "loss": 0.3221, "step": 9190 }, { "epoch": 1.11, "learning_rate": 1.1081667068176344e-05, "loss": 0.5058, "step": 9200 }, { "epoch": 1.11, "learning_rate": 1.109371235846784e-05, "loss": 0.3444, "step": 9210 }, { "epoch": 1.11, "learning_rate": 1.1105757648759335e-05, "loss": 0.3818, "step": 9220 }, { "epoch": 1.11, "learning_rate": 1.1117802939050832e-05, "loss": 0.3573, "step": 9230 }, { "epoch": 1.11, "learning_rate": 1.1129848229342328e-05, "loss": 0.3912, "step": 9240 }, { "epoch": 1.11, "learning_rate": 1.1141893519633824e-05, "loss": 0.4674, "step": 9250 }, { "epoch": 1.12, "learning_rate": 1.115393880992532e-05, "loss": 0.2574, "step": 9260 }, { "epoch": 1.12, "learning_rate": 1.1165984100216816e-05, "loss": 0.5234, "step": 9270 }, { "epoch": 1.12, "learning_rate": 1.1178029390508311e-05, "loss": 0.4797, "step": 9280 }, { "epoch": 1.12, "learning_rate": 1.1190074680799808e-05, "loss": 0.3532, "step": 9290 }, { "epoch": 1.12, "learning_rate": 1.1202119971091305e-05, "loss": 0.2743, "step": 9300 }, { "epoch": 1.12, "learning_rate": 1.12141652613828e-05, "loss": 0.2132, "step": 9310 }, { "epoch": 1.12, "learning_rate": 1.1226210551674297e-05, "loss": 0.356, "step": 9320 }, { "epoch": 1.12, "learning_rate": 1.1238255841965792e-05, "loss": 0.3616, "step": 9330 }, { "epoch": 1.12, "learning_rate": 1.1250301132257287e-05, "loss": 0.3397, "step": 9340 }, { "epoch": 1.13, "learning_rate": 1.1262346422548784e-05, "loss": 0.3918, "step": 9350 }, { "epoch": 1.13, "learning_rate": 1.127439171284028e-05, "loss": 0.3928, "step": 9360 }, { "epoch": 1.13, "learning_rate": 1.1286437003131776e-05, "loss": 0.3664, "step": 9370 }, { "epoch": 1.13, "learning_rate": 1.1298482293423273e-05, "loss": 0.517, "step": 9380 }, { "epoch": 1.13, "learning_rate": 1.1310527583714768e-05, "loss": 0.4504, "step": 9390 }, { "epoch": 1.13, "learning_rate": 1.1322572874006263e-05, "loss": 0.5106, "step": 9400 }, { "epoch": 1.13, "learning_rate": 1.133461816429776e-05, "loss": 0.4331, "step": 9410 }, { "epoch": 1.13, "learning_rate": 1.1346663454589255e-05, "loss": 0.3698, "step": 9420 }, { "epoch": 1.14, "learning_rate": 1.1358708744880752e-05, "loss": 0.4401, "step": 9430 }, { "epoch": 1.14, "learning_rate": 1.1370754035172249e-05, "loss": 0.5724, "step": 9440 }, { "epoch": 1.14, "learning_rate": 1.1382799325463744e-05, "loss": 0.3299, "step": 9450 }, { "epoch": 1.14, "learning_rate": 1.139484461575524e-05, "loss": 0.4635, "step": 9460 }, { "epoch": 1.14, "learning_rate": 1.1406889906046736e-05, "loss": 0.4139, "step": 9470 }, { "epoch": 1.14, "learning_rate": 1.1418935196338232e-05, "loss": 0.3881, "step": 9480 }, { "epoch": 1.14, "learning_rate": 1.1430980486629728e-05, "loss": 0.4271, "step": 9490 }, { "epoch": 1.14, "learning_rate": 1.1443025776921225e-05, "loss": 0.2873, "step": 9500 }, { "epoch": 1.15, "learning_rate": 1.145507106721272e-05, "loss": 0.3298, "step": 9510 }, { "epoch": 1.15, "learning_rate": 1.1467116357504216e-05, "loss": 0.2938, "step": 9520 }, { "epoch": 1.15, "learning_rate": 1.1479161647795713e-05, "loss": 0.6126, "step": 9530 }, { "epoch": 1.15, "learning_rate": 1.1491206938087208e-05, "loss": 0.3208, "step": 9540 }, { "epoch": 1.15, "learning_rate": 1.1503252228378705e-05, "loss": 0.3831, "step": 9550 }, { "epoch": 1.15, "learning_rate": 1.1515297518670201e-05, "loss": 0.2818, "step": 9560 }, { "epoch": 1.15, "learning_rate": 1.1527342808961697e-05, "loss": 0.5447, "step": 9570 }, { "epoch": 1.15, "learning_rate": 1.1539388099253194e-05, "loss": 0.2883, "step": 9580 }, { "epoch": 1.16, "learning_rate": 1.1551433389544689e-05, "loss": 0.5855, "step": 9590 }, { "epoch": 1.16, "learning_rate": 1.1563478679836184e-05, "loss": 0.5799, "step": 9600 }, { "epoch": 1.16, "learning_rate": 1.157552397012768e-05, "loss": 0.5013, "step": 9610 }, { "epoch": 1.16, "learning_rate": 1.1587569260419176e-05, "loss": 0.1882, "step": 9620 }, { "epoch": 1.16, "learning_rate": 1.1599614550710673e-05, "loss": 0.2029, "step": 9630 }, { "epoch": 1.16, "learning_rate": 1.161165984100217e-05, "loss": 0.4352, "step": 9640 }, { "epoch": 1.16, "learning_rate": 1.1623705131293665e-05, "loss": 0.4715, "step": 9650 }, { "epoch": 1.16, "learning_rate": 1.163575042158516e-05, "loss": 0.4523, "step": 9660 }, { "epoch": 1.16, "learning_rate": 1.1647795711876657e-05, "loss": 0.3172, "step": 9670 }, { "epoch": 1.17, "learning_rate": 1.1659841002168152e-05, "loss": 0.3978, "step": 9680 }, { "epoch": 1.17, "learning_rate": 1.1671886292459649e-05, "loss": 0.4146, "step": 9690 }, { "epoch": 1.17, "learning_rate": 1.1683931582751146e-05, "loss": 0.3802, "step": 9700 }, { "epoch": 1.17, "learning_rate": 1.1695976873042641e-05, "loss": 0.3293, "step": 9710 }, { "epoch": 1.17, "learning_rate": 1.1708022163334136e-05, "loss": 0.3704, "step": 9720 }, { "epoch": 1.17, "learning_rate": 1.1720067453625633e-05, "loss": 0.2578, "step": 9730 }, { "epoch": 1.17, "learning_rate": 1.1732112743917128e-05, "loss": 0.6322, "step": 9740 }, { "epoch": 1.17, "learning_rate": 1.1744158034208624e-05, "loss": 0.2488, "step": 9750 }, { "epoch": 1.18, "learning_rate": 1.1756203324500122e-05, "loss": 0.3944, "step": 9760 }, { "epoch": 1.18, "learning_rate": 1.1768248614791617e-05, "loss": 0.5959, "step": 9770 }, { "epoch": 1.18, "learning_rate": 1.1780293905083112e-05, "loss": 0.3774, "step": 9780 }, { "epoch": 1.18, "learning_rate": 1.179233919537461e-05, "loss": 0.3298, "step": 9790 }, { "epoch": 1.18, "learning_rate": 1.1804384485666105e-05, "loss": 0.4073, "step": 9800 }, { "epoch": 1.18, "learning_rate": 1.1816429775957601e-05, "loss": 0.3669, "step": 9810 }, { "epoch": 1.18, "learning_rate": 1.1828475066249098e-05, "loss": 0.327, "step": 9820 }, { "epoch": 1.18, "learning_rate": 1.1840520356540593e-05, "loss": 0.3494, "step": 9830 }, { "epoch": 1.19, "learning_rate": 1.1852565646832089e-05, "loss": 0.5345, "step": 9840 }, { "epoch": 1.19, "learning_rate": 1.1864610937123586e-05, "loss": 0.4357, "step": 9850 }, { "epoch": 1.19, "learning_rate": 1.187665622741508e-05, "loss": 0.2851, "step": 9860 }, { "epoch": 1.19, "learning_rate": 1.1888701517706578e-05, "loss": 0.2626, "step": 9870 }, { "epoch": 1.19, "learning_rate": 1.1900746807998073e-05, "loss": 0.4714, "step": 9880 }, { "epoch": 1.19, "learning_rate": 1.191279209828957e-05, "loss": 0.3791, "step": 9890 }, { "epoch": 1.19, "learning_rate": 1.1924837388581067e-05, "loss": 0.4164, "step": 9900 }, { "epoch": 1.19, "learning_rate": 1.1936882678872562e-05, "loss": 0.2851, "step": 9910 }, { "epoch": 1.19, "learning_rate": 1.1948927969164057e-05, "loss": 0.3515, "step": 9920 }, { "epoch": 1.2, "learning_rate": 1.1960973259455554e-05, "loss": 0.3537, "step": 9930 }, { "epoch": 1.2, "learning_rate": 1.1973018549747049e-05, "loss": 0.2608, "step": 9940 }, { "epoch": 1.2, "learning_rate": 1.1985063840038544e-05, "loss": 0.2467, "step": 9950 }, { "epoch": 1.2, "learning_rate": 1.1997109130330043e-05, "loss": 0.4219, "step": 9960 }, { "epoch": 1.2, "learning_rate": 1.2009154420621538e-05, "loss": 0.2925, "step": 9970 }, { "epoch": 1.2, "learning_rate": 1.2021199710913033e-05, "loss": 0.2669, "step": 9980 }, { "epoch": 1.2, "learning_rate": 1.203324500120453e-05, "loss": 0.5412, "step": 9990 }, { "epoch": 1.2, "learning_rate": 1.2045290291496025e-05, "loss": 0.4498, "step": 10000 }, { "epoch": 1.21, "learning_rate": 1.205733558178752e-05, "loss": 0.2609, "step": 10010 }, { "epoch": 1.21, "learning_rate": 1.2069380872079019e-05, "loss": 0.2133, "step": 10020 }, { "epoch": 1.21, "learning_rate": 1.2081426162370514e-05, "loss": 0.4027, "step": 10030 }, { "epoch": 1.21, "learning_rate": 1.209347145266201e-05, "loss": 0.256, "step": 10040 }, { "epoch": 1.21, "learning_rate": 1.2105516742953506e-05, "loss": 0.4117, "step": 10050 }, { "epoch": 1.21, "learning_rate": 1.2117562033245001e-05, "loss": 0.5944, "step": 10060 }, { "epoch": 1.21, "learning_rate": 1.2129607323536497e-05, "loss": 0.4816, "step": 10070 }, { "epoch": 1.21, "learning_rate": 1.2141652613827993e-05, "loss": 0.301, "step": 10080 }, { "epoch": 1.22, "learning_rate": 1.215369790411949e-05, "loss": 0.3282, "step": 10090 }, { "epoch": 1.22, "learning_rate": 1.2165743194410986e-05, "loss": 0.4364, "step": 10100 }, { "epoch": 1.22, "learning_rate": 1.2177788484702482e-05, "loss": 0.2965, "step": 10110 }, { "epoch": 1.22, "learning_rate": 1.2189833774993978e-05, "loss": 0.548, "step": 10120 }, { "epoch": 1.22, "learning_rate": 1.2201879065285474e-05, "loss": 0.1975, "step": 10130 }, { "epoch": 1.22, "learning_rate": 1.221392435557697e-05, "loss": 0.4656, "step": 10140 }, { "epoch": 1.22, "learning_rate": 1.2225969645868467e-05, "loss": 0.3649, "step": 10150 }, { "epoch": 1.22, "learning_rate": 1.2238014936159962e-05, "loss": 0.4439, "step": 10160 }, { "epoch": 1.22, "learning_rate": 1.2250060226451459e-05, "loss": 0.4052, "step": 10170 }, { "epoch": 1.23, "learning_rate": 1.2262105516742954e-05, "loss": 0.313, "step": 10180 }, { "epoch": 1.23, "learning_rate": 1.227415080703445e-05, "loss": 0.3402, "step": 10190 }, { "epoch": 1.23, "learning_rate": 1.2286196097325946e-05, "loss": 0.6, "step": 10200 }, { "epoch": 1.23, "learning_rate": 1.2298241387617441e-05, "loss": 0.3297, "step": 10210 }, { "epoch": 1.23, "learning_rate": 1.231028667790894e-05, "loss": 0.2274, "step": 10220 }, { "epoch": 1.23, "learning_rate": 1.2322331968200435e-05, "loss": 0.3509, "step": 10230 }, { "epoch": 1.23, "learning_rate": 1.233437725849193e-05, "loss": 0.2626, "step": 10240 }, { "epoch": 1.23, "learning_rate": 1.2346422548783427e-05, "loss": 0.4718, "step": 10250 }, { "epoch": 1.24, "learning_rate": 1.2358467839074922e-05, "loss": 0.4903, "step": 10260 }, { "epoch": 1.24, "learning_rate": 1.2370513129366417e-05, "loss": 0.1601, "step": 10270 }, { "epoch": 1.24, "learning_rate": 1.2382558419657916e-05, "loss": 0.6041, "step": 10280 }, { "epoch": 1.24, "learning_rate": 1.2394603709949411e-05, "loss": 0.3548, "step": 10290 }, { "epoch": 1.24, "learning_rate": 1.2406649000240906e-05, "loss": 0.3877, "step": 10300 }, { "epoch": 1.24, "learning_rate": 1.2418694290532403e-05, "loss": 0.3786, "step": 10310 }, { "epoch": 1.24, "learning_rate": 1.2430739580823898e-05, "loss": 0.2247, "step": 10320 }, { "epoch": 1.24, "learning_rate": 1.2442784871115393e-05, "loss": 0.3873, "step": 10330 }, { "epoch": 1.25, "learning_rate": 1.245483016140689e-05, "loss": 0.3348, "step": 10340 }, { "epoch": 1.25, "learning_rate": 1.2466875451698387e-05, "loss": 0.3212, "step": 10350 }, { "epoch": 1.25, "learning_rate": 1.2478920741989882e-05, "loss": 0.4174, "step": 10360 }, { "epoch": 1.25, "learning_rate": 1.249096603228138e-05, "loss": 0.2686, "step": 10370 }, { "epoch": 1.25, "learning_rate": 1.2503011322572874e-05, "loss": 0.2603, "step": 10380 }, { "epoch": 1.25, "learning_rate": 1.251505661286437e-05, "loss": 0.3362, "step": 10390 }, { "epoch": 1.25, "learning_rate": 1.2527101903155866e-05, "loss": 0.2682, "step": 10400 }, { "epoch": 1.25, "learning_rate": 1.2539147193447362e-05, "loss": 0.2858, "step": 10410 }, { "epoch": 1.26, "learning_rate": 1.2551192483738859e-05, "loss": 0.3353, "step": 10420 }, { "epoch": 1.26, "learning_rate": 1.2563237774030354e-05, "loss": 0.4343, "step": 10430 }, { "epoch": 1.26, "learning_rate": 1.2575283064321849e-05, "loss": 0.8225, "step": 10440 }, { "epoch": 1.26, "learning_rate": 1.2587328354613347e-05, "loss": 0.5498, "step": 10450 }, { "epoch": 1.26, "learning_rate": 1.2599373644904844e-05, "loss": 0.4412, "step": 10460 }, { "epoch": 1.26, "learning_rate": 1.261141893519634e-05, "loss": 0.4852, "step": 10470 }, { "epoch": 1.26, "learning_rate": 1.2623464225487835e-05, "loss": 0.2934, "step": 10480 }, { "epoch": 1.26, "learning_rate": 1.2635509515779332e-05, "loss": 0.2178, "step": 10490 }, { "epoch": 1.26, "learning_rate": 1.2647554806070827e-05, "loss": 0.3313, "step": 10500 }, { "epoch": 1.27, "learning_rate": 1.2659600096362324e-05, "loss": 0.4309, "step": 10510 }, { "epoch": 1.27, "learning_rate": 1.2671645386653819e-05, "loss": 0.4869, "step": 10520 }, { "epoch": 1.27, "learning_rate": 1.2683690676945314e-05, "loss": 0.3481, "step": 10530 }, { "epoch": 1.27, "learning_rate": 1.2695735967236811e-05, "loss": 0.5286, "step": 10540 }, { "epoch": 1.27, "learning_rate": 1.2707781257528306e-05, "loss": 0.4182, "step": 10550 }, { "epoch": 1.27, "learning_rate": 1.2719826547819801e-05, "loss": 0.3799, "step": 10560 }, { "epoch": 1.27, "learning_rate": 1.2731871838111298e-05, "loss": 0.3756, "step": 10570 }, { "epoch": 1.27, "learning_rate": 1.2743917128402797e-05, "loss": 0.3464, "step": 10580 }, { "epoch": 1.28, "learning_rate": 1.2755962418694292e-05, "loss": 0.2329, "step": 10590 }, { "epoch": 1.28, "learning_rate": 1.2768007708985789e-05, "loss": 0.4941, "step": 10600 }, { "epoch": 1.28, "learning_rate": 1.2780052999277284e-05, "loss": 0.3596, "step": 10610 }, { "epoch": 1.28, "learning_rate": 1.279209828956878e-05, "loss": 0.2119, "step": 10620 }, { "epoch": 1.28, "learning_rate": 1.2804143579860276e-05, "loss": 0.4331, "step": 10630 }, { "epoch": 1.28, "learning_rate": 1.2816188870151771e-05, "loss": 0.3859, "step": 10640 }, { "epoch": 1.28, "learning_rate": 1.2828234160443266e-05, "loss": 0.3466, "step": 10650 }, { "epoch": 1.28, "learning_rate": 1.2840279450734763e-05, "loss": 0.2155, "step": 10660 }, { "epoch": 1.29, "learning_rate": 1.2852324741026258e-05, "loss": 0.3596, "step": 10670 }, { "epoch": 1.29, "learning_rate": 1.2864370031317755e-05, "loss": 0.5381, "step": 10680 }, { "epoch": 1.29, "learning_rate": 1.287641532160925e-05, "loss": 0.4123, "step": 10690 }, { "epoch": 1.29, "learning_rate": 1.2888460611900746e-05, "loss": 0.3519, "step": 10700 }, { "epoch": 1.29, "learning_rate": 1.2900505902192244e-05, "loss": 0.2735, "step": 10710 }, { "epoch": 1.29, "learning_rate": 1.2912551192483741e-05, "loss": 0.3256, "step": 10720 }, { "epoch": 1.29, "learning_rate": 1.2924596482775236e-05, "loss": 0.3526, "step": 10730 }, { "epoch": 1.29, "learning_rate": 1.2936641773066732e-05, "loss": 0.3734, "step": 10740 }, { "epoch": 1.29, "learning_rate": 1.2948687063358228e-05, "loss": 0.3678, "step": 10750 }, { "epoch": 1.3, "learning_rate": 1.2960732353649724e-05, "loss": 0.4595, "step": 10760 }, { "epoch": 1.3, "learning_rate": 1.2972777643941219e-05, "loss": 0.2911, "step": 10770 }, { "epoch": 1.3, "learning_rate": 1.2984822934232716e-05, "loss": 0.3885, "step": 10780 }, { "epoch": 1.3, "learning_rate": 1.2996868224524211e-05, "loss": 0.239, "step": 10790 }, { "epoch": 1.3, "learning_rate": 1.3008913514815708e-05, "loss": 0.3013, "step": 10800 }, { "epoch": 1.3, "learning_rate": 1.3020958805107203e-05, "loss": 0.3748, "step": 10810 }, { "epoch": 1.3, "learning_rate": 1.3033004095398698e-05, "loss": 0.2455, "step": 10820 }, { "epoch": 1.3, "learning_rate": 1.3045049385690195e-05, "loss": 0.4706, "step": 10830 }, { "epoch": 1.31, "learning_rate": 1.3057094675981694e-05, "loss": 0.4589, "step": 10840 }, { "epoch": 1.31, "learning_rate": 1.3069139966273189e-05, "loss": 0.3049, "step": 10850 }, { "epoch": 1.31, "learning_rate": 1.3081185256564684e-05, "loss": 0.3244, "step": 10860 }, { "epoch": 1.31, "learning_rate": 1.309323054685618e-05, "loss": 0.4397, "step": 10870 }, { "epoch": 1.31, "learning_rate": 1.3105275837147676e-05, "loss": 0.2534, "step": 10880 }, { "epoch": 1.31, "learning_rate": 1.3117321127439173e-05, "loss": 0.1862, "step": 10890 }, { "epoch": 1.31, "learning_rate": 1.3129366417730668e-05, "loss": 0.6529, "step": 10900 }, { "epoch": 1.31, "learning_rate": 1.3141411708022163e-05, "loss": 0.361, "step": 10910 }, { "epoch": 1.32, "learning_rate": 1.315345699831366e-05, "loss": 0.4102, "step": 10920 }, { "epoch": 1.32, "learning_rate": 1.3165502288605155e-05, "loss": 0.4684, "step": 10930 }, { "epoch": 1.32, "learning_rate": 1.317754757889665e-05, "loss": 0.3702, "step": 10940 }, { "epoch": 1.32, "learning_rate": 1.3189592869188147e-05, "loss": 0.3239, "step": 10950 }, { "epoch": 1.32, "learning_rate": 1.3201638159479643e-05, "loss": 0.245, "step": 10960 }, { "epoch": 1.32, "learning_rate": 1.3213683449771141e-05, "loss": 0.284, "step": 10970 }, { "epoch": 1.32, "learning_rate": 1.3225728740062638e-05, "loss": 0.3126, "step": 10980 }, { "epoch": 1.32, "learning_rate": 1.3237774030354133e-05, "loss": 0.4432, "step": 10990 }, { "epoch": 1.32, "learning_rate": 1.3249819320645628e-05, "loss": 0.2476, "step": 11000 }, { "epoch": 1.33, "learning_rate": 1.3261864610937125e-05, "loss": 0.3528, "step": 11010 }, { "epoch": 1.33, "learning_rate": 1.327390990122862e-05, "loss": 0.2475, "step": 11020 }, { "epoch": 1.33, "learning_rate": 1.3285955191520116e-05, "loss": 0.3123, "step": 11030 }, { "epoch": 1.33, "learning_rate": 1.3298000481811613e-05, "loss": 0.2975, "step": 11040 }, { "epoch": 1.33, "learning_rate": 1.3310045772103108e-05, "loss": 0.2531, "step": 11050 }, { "epoch": 1.33, "learning_rate": 1.3322091062394605e-05, "loss": 0.5253, "step": 11060 }, { "epoch": 1.33, "learning_rate": 1.33341363526861e-05, "loss": 0.2272, "step": 11070 }, { "epoch": 1.33, "learning_rate": 1.3346181642977595e-05, "loss": 0.5105, "step": 11080 }, { "epoch": 1.34, "learning_rate": 1.3358226933269092e-05, "loss": 0.4141, "step": 11090 }, { "epoch": 1.34, "learning_rate": 1.337027222356059e-05, "loss": 0.5504, "step": 11100 }, { "epoch": 1.34, "learning_rate": 1.3382317513852086e-05, "loss": 0.3032, "step": 11110 }, { "epoch": 1.34, "learning_rate": 1.339436280414358e-05, "loss": 0.4668, "step": 11120 }, { "epoch": 1.34, "learning_rate": 1.3406408094435078e-05, "loss": 0.306, "step": 11130 }, { "epoch": 1.34, "learning_rate": 1.3418453384726573e-05, "loss": 0.4476, "step": 11140 }, { "epoch": 1.34, "learning_rate": 1.343049867501807e-05, "loss": 0.3189, "step": 11150 }, { "epoch": 1.34, "learning_rate": 1.3442543965309565e-05, "loss": 0.2039, "step": 11160 }, { "epoch": 1.35, "learning_rate": 1.345458925560106e-05, "loss": 0.4322, "step": 11170 }, { "epoch": 1.35, "learning_rate": 1.3466634545892557e-05, "loss": 0.3732, "step": 11180 }, { "epoch": 1.35, "learning_rate": 1.3478679836184052e-05, "loss": 0.3319, "step": 11190 }, { "epoch": 1.35, "learning_rate": 1.3490725126475547e-05, "loss": 0.47, "step": 11200 }, { "epoch": 1.35, "learning_rate": 1.3502770416767044e-05, "loss": 0.3992, "step": 11210 }, { "epoch": 1.35, "learning_rate": 1.351481570705854e-05, "loss": 0.5242, "step": 11220 }, { "epoch": 1.35, "learning_rate": 1.3526860997350036e-05, "loss": 0.2599, "step": 11230 }, { "epoch": 1.35, "learning_rate": 1.3538906287641535e-05, "loss": 0.4286, "step": 11240 }, { "epoch": 1.36, "learning_rate": 1.355095157793303e-05, "loss": 0.2825, "step": 11250 }, { "epoch": 1.36, "learning_rate": 1.3562996868224525e-05, "loss": 0.5086, "step": 11260 }, { "epoch": 1.36, "learning_rate": 1.3575042158516022e-05, "loss": 0.2748, "step": 11270 }, { "epoch": 1.36, "learning_rate": 1.3587087448807517e-05, "loss": 0.2844, "step": 11280 }, { "epoch": 1.36, "learning_rate": 1.3599132739099012e-05, "loss": 0.3002, "step": 11290 }, { "epoch": 1.36, "learning_rate": 1.361117802939051e-05, "loss": 0.5451, "step": 11300 }, { "epoch": 1.36, "learning_rate": 1.3623223319682005e-05, "loss": 0.3293, "step": 11310 }, { "epoch": 1.36, "learning_rate": 1.36352686099735e-05, "loss": 0.5627, "step": 11320 }, { "epoch": 1.36, "learning_rate": 1.3647313900264997e-05, "loss": 0.242, "step": 11330 }, { "epoch": 1.37, "learning_rate": 1.3659359190556492e-05, "loss": 0.5285, "step": 11340 }, { "epoch": 1.37, "learning_rate": 1.3671404480847989e-05, "loss": 0.5053, "step": 11350 }, { "epoch": 1.37, "learning_rate": 1.3683449771139484e-05, "loss": 0.4217, "step": 11360 }, { "epoch": 1.37, "learning_rate": 1.3695495061430982e-05, "loss": 0.4739, "step": 11370 }, { "epoch": 1.37, "learning_rate": 1.3707540351722478e-05, "loss": 0.3974, "step": 11380 }, { "epoch": 1.37, "learning_rate": 1.3719585642013974e-05, "loss": 0.4268, "step": 11390 }, { "epoch": 1.37, "learning_rate": 1.373163093230547e-05, "loss": 0.4897, "step": 11400 }, { "epoch": 1.37, "learning_rate": 1.3743676222596965e-05, "loss": 0.2236, "step": 11410 }, { "epoch": 1.38, "learning_rate": 1.3755721512888462e-05, "loss": 0.2109, "step": 11420 }, { "epoch": 1.38, "learning_rate": 1.3767766803179957e-05, "loss": 0.2655, "step": 11430 }, { "epoch": 1.38, "learning_rate": 1.3779812093471454e-05, "loss": 0.2143, "step": 11440 }, { "epoch": 1.38, "learning_rate": 1.3791857383762949e-05, "loss": 0.3833, "step": 11450 }, { "epoch": 1.38, "learning_rate": 1.3803902674054444e-05, "loss": 0.3263, "step": 11460 }, { "epoch": 1.38, "learning_rate": 1.3815947964345941e-05, "loss": 0.327, "step": 11470 }, { "epoch": 1.38, "learning_rate": 1.3827993254637436e-05, "loss": 0.4004, "step": 11480 }, { "epoch": 1.38, "learning_rate": 1.3840038544928931e-05, "loss": 0.4076, "step": 11490 }, { "epoch": 1.39, "learning_rate": 1.385208383522043e-05, "loss": 0.2048, "step": 11500 }, { "epoch": 1.39, "learning_rate": 1.3864129125511927e-05, "loss": 0.269, "step": 11510 }, { "epoch": 1.39, "learning_rate": 1.3876174415803422e-05, "loss": 0.4911, "step": 11520 }, { "epoch": 1.39, "learning_rate": 1.3888219706094919e-05, "loss": 0.5099, "step": 11530 }, { "epoch": 1.39, "learning_rate": 1.3900264996386414e-05, "loss": 0.4482, "step": 11540 }, { "epoch": 1.39, "learning_rate": 1.391231028667791e-05, "loss": 0.3959, "step": 11550 }, { "epoch": 1.39, "learning_rate": 1.3924355576969406e-05, "loss": 0.279, "step": 11560 }, { "epoch": 1.39, "learning_rate": 1.3936400867260901e-05, "loss": 0.3413, "step": 11570 }, { "epoch": 1.39, "learning_rate": 1.3948446157552397e-05, "loss": 0.4297, "step": 11580 }, { "epoch": 1.4, "learning_rate": 1.3960491447843893e-05, "loss": 0.2419, "step": 11590 }, { "epoch": 1.4, "learning_rate": 1.3972536738135389e-05, "loss": 0.1612, "step": 11600 }, { "epoch": 1.4, "learning_rate": 1.3984582028426885e-05, "loss": 0.2313, "step": 11610 }, { "epoch": 1.4, "learning_rate": 1.399662731871838e-05, "loss": 0.3355, "step": 11620 }, { "epoch": 1.4, "learning_rate": 1.400867260900988e-05, "loss": 0.4352, "step": 11630 }, { "epoch": 1.4, "learning_rate": 1.4020717899301374e-05, "loss": 0.1389, "step": 11640 }, { "epoch": 1.4, "learning_rate": 1.4032763189592871e-05, "loss": 0.4996, "step": 11650 }, { "epoch": 1.4, "learning_rate": 1.4044808479884366e-05, "loss": 0.5073, "step": 11660 }, { "epoch": 1.41, "learning_rate": 1.4056853770175862e-05, "loss": 0.3096, "step": 11670 }, { "epoch": 1.41, "learning_rate": 1.4068899060467359e-05, "loss": 0.3518, "step": 11680 }, { "epoch": 1.41, "learning_rate": 1.4080944350758854e-05, "loss": 0.3444, "step": 11690 }, { "epoch": 1.41, "learning_rate": 1.409298964105035e-05, "loss": 0.2343, "step": 11700 }, { "epoch": 1.41, "learning_rate": 1.4105034931341846e-05, "loss": 0.2693, "step": 11710 }, { "epoch": 1.41, "learning_rate": 1.4117080221633341e-05, "loss": 0.5802, "step": 11720 }, { "epoch": 1.41, "learning_rate": 1.4129125511924838e-05, "loss": 0.1507, "step": 11730 }, { "epoch": 1.41, "learning_rate": 1.4141170802216333e-05, "loss": 0.3518, "step": 11740 }, { "epoch": 1.42, "learning_rate": 1.4153216092507828e-05, "loss": 0.4109, "step": 11750 }, { "epoch": 1.42, "learning_rate": 1.4165261382799327e-05, "loss": 0.4011, "step": 11760 }, { "epoch": 1.42, "learning_rate": 1.4177306673090824e-05, "loss": 0.2582, "step": 11770 }, { "epoch": 1.42, "learning_rate": 1.4189351963382319e-05, "loss": 0.2418, "step": 11780 }, { "epoch": 1.42, "learning_rate": 1.4201397253673816e-05, "loss": 0.3764, "step": 11790 }, { "epoch": 1.42, "learning_rate": 1.4213442543965311e-05, "loss": 0.4159, "step": 11800 }, { "epoch": 1.42, "learning_rate": 1.4225487834256806e-05, "loss": 0.2722, "step": 11810 }, { "epoch": 1.42, "learning_rate": 1.4237533124548303e-05, "loss": 0.4073, "step": 11820 }, { "epoch": 1.42, "learning_rate": 1.4249578414839798e-05, "loss": 0.3489, "step": 11830 }, { "epoch": 1.43, "learning_rate": 1.4261623705131293e-05, "loss": 0.3851, "step": 11840 }, { "epoch": 1.43, "learning_rate": 1.427366899542279e-05, "loss": 0.3512, "step": 11850 }, { "epoch": 1.43, "learning_rate": 1.4285714285714285e-05, "loss": 0.3986, "step": 11860 }, { "epoch": 1.43, "learning_rate": 1.429775957600578e-05, "loss": 0.3274, "step": 11870 }, { "epoch": 1.43, "learning_rate": 1.4309804866297277e-05, "loss": 0.3681, "step": 11880 }, { "epoch": 1.43, "learning_rate": 1.4321850156588776e-05, "loss": 0.5337, "step": 11890 }, { "epoch": 1.43, "learning_rate": 1.4333895446880271e-05, "loss": 0.2649, "step": 11900 }, { "epoch": 1.43, "learning_rate": 1.4345940737171768e-05, "loss": 0.3293, "step": 11910 }, { "epoch": 1.44, "learning_rate": 1.4357986027463263e-05, "loss": 0.1857, "step": 11920 }, { "epoch": 1.44, "learning_rate": 1.4370031317754759e-05, "loss": 0.2115, "step": 11930 }, { "epoch": 1.44, "learning_rate": 1.4382076608046255e-05, "loss": 0.4037, "step": 11940 }, { "epoch": 1.44, "learning_rate": 1.439412189833775e-05, "loss": 0.4671, "step": 11950 }, { "epoch": 1.44, "learning_rate": 1.4406167188629246e-05, "loss": 0.3055, "step": 11960 }, { "epoch": 1.44, "learning_rate": 1.4418212478920743e-05, "loss": 0.1612, "step": 11970 }, { "epoch": 1.44, "learning_rate": 1.4430257769212238e-05, "loss": 0.4918, "step": 11980 }, { "epoch": 1.44, "learning_rate": 1.4442303059503735e-05, "loss": 0.3537, "step": 11990 }, { "epoch": 1.45, "learning_rate": 1.445434834979523e-05, "loss": 0.5459, "step": 12000 }, { "epoch": 1.45, "learning_rate": 1.4466393640086725e-05, "loss": 0.309, "step": 12010 }, { "epoch": 1.45, "learning_rate": 1.4478438930378224e-05, "loss": 0.4333, "step": 12020 }, { "epoch": 1.45, "learning_rate": 1.449048422066972e-05, "loss": 0.3093, "step": 12030 }, { "epoch": 1.45, "learning_rate": 1.4502529510961216e-05, "loss": 0.4552, "step": 12040 }, { "epoch": 1.45, "learning_rate": 1.4514574801252711e-05, "loss": 0.4958, "step": 12050 }, { "epoch": 1.45, "learning_rate": 1.4526620091544208e-05, "loss": 0.4371, "step": 12060 }, { "epoch": 1.45, "learning_rate": 1.4538665381835703e-05, "loss": 0.2332, "step": 12070 }, { "epoch": 1.45, "learning_rate": 1.45507106721272e-05, "loss": 0.3764, "step": 12080 }, { "epoch": 1.46, "learning_rate": 1.4562755962418695e-05, "loss": 0.6118, "step": 12090 }, { "epoch": 1.46, "learning_rate": 1.457480125271019e-05, "loss": 0.2334, "step": 12100 }, { "epoch": 1.46, "learning_rate": 1.4586846543001687e-05, "loss": 0.3318, "step": 12110 }, { "epoch": 1.46, "learning_rate": 1.4598891833293182e-05, "loss": 0.3161, "step": 12120 }, { "epoch": 1.46, "learning_rate": 1.4610937123584677e-05, "loss": 0.2787, "step": 12130 }, { "epoch": 1.46, "learning_rate": 1.4622982413876174e-05, "loss": 0.5164, "step": 12140 }, { "epoch": 1.46, "learning_rate": 1.463502770416767e-05, "loss": 0.5754, "step": 12150 }, { "epoch": 1.46, "learning_rate": 1.4647072994459168e-05, "loss": 0.3001, "step": 12160 }, { "epoch": 1.47, "learning_rate": 1.4659118284750665e-05, "loss": 0.4091, "step": 12170 }, { "epoch": 1.47, "learning_rate": 1.467116357504216e-05, "loss": 0.3288, "step": 12180 }, { "epoch": 1.47, "learning_rate": 1.4683208865333655e-05, "loss": 0.2347, "step": 12190 }, { "epoch": 1.47, "learning_rate": 1.4695254155625152e-05, "loss": 0.5855, "step": 12200 }, { "epoch": 1.47, "learning_rate": 1.4707299445916647e-05, "loss": 0.3206, "step": 12210 }, { "epoch": 1.47, "learning_rate": 1.4719344736208143e-05, "loss": 0.1844, "step": 12220 }, { "epoch": 1.47, "learning_rate": 1.473139002649964e-05, "loss": 0.3442, "step": 12230 }, { "epoch": 1.47, "learning_rate": 1.4743435316791135e-05, "loss": 0.2072, "step": 12240 }, { "epoch": 1.48, "learning_rate": 1.4755480607082632e-05, "loss": 0.2881, "step": 12250 }, { "epoch": 1.48, "learning_rate": 1.4767525897374127e-05, "loss": 0.5588, "step": 12260 }, { "epoch": 1.48, "learning_rate": 1.4779571187665622e-05, "loss": 0.2865, "step": 12270 }, { "epoch": 1.48, "learning_rate": 1.4791616477957119e-05, "loss": 0.3952, "step": 12280 }, { "epoch": 1.48, "learning_rate": 1.4803661768248617e-05, "loss": 0.2356, "step": 12290 }, { "epoch": 1.48, "learning_rate": 1.4815707058540113e-05, "loss": 0.4747, "step": 12300 }, { "epoch": 1.48, "learning_rate": 1.4827752348831608e-05, "loss": 0.5004, "step": 12310 }, { "epoch": 1.48, "learning_rate": 1.4839797639123105e-05, "loss": 0.235, "step": 12320 }, { "epoch": 1.49, "learning_rate": 1.48518429294146e-05, "loss": 0.3126, "step": 12330 }, { "epoch": 1.49, "learning_rate": 1.4863888219706095e-05, "loss": 0.381, "step": 12340 }, { "epoch": 1.49, "learning_rate": 1.4875933509997592e-05, "loss": 0.4341, "step": 12350 }, { "epoch": 1.49, "learning_rate": 1.4887978800289087e-05, "loss": 0.4029, "step": 12360 }, { "epoch": 1.49, "learning_rate": 1.4900024090580584e-05, "loss": 0.293, "step": 12370 }, { "epoch": 1.49, "learning_rate": 1.4912069380872079e-05, "loss": 0.3431, "step": 12380 }, { "epoch": 1.49, "learning_rate": 1.4924114671163574e-05, "loss": 0.3762, "step": 12390 }, { "epoch": 1.49, "learning_rate": 1.4936159961455071e-05, "loss": 0.3159, "step": 12400 }, { "epoch": 1.49, "learning_rate": 1.4948205251746566e-05, "loss": 0.4036, "step": 12410 }, { "epoch": 1.5, "learning_rate": 1.4960250542038065e-05, "loss": 0.3693, "step": 12420 }, { "epoch": 1.5, "learning_rate": 1.497229583232956e-05, "loss": 0.3619, "step": 12430 }, { "epoch": 1.5, "learning_rate": 1.4984341122621057e-05, "loss": 0.4055, "step": 12440 }, { "epoch": 1.5, "learning_rate": 1.4996386412912552e-05, "loss": 0.2692, "step": 12450 }, { "epoch": 1.5, "learning_rate": 1.5008431703204049e-05, "loss": 0.3584, "step": 12460 }, { "epoch": 1.5, "learning_rate": 1.5020476993495544e-05, "loss": 0.2851, "step": 12470 }, { "epoch": 1.5, "learning_rate": 1.503252228378704e-05, "loss": 0.3085, "step": 12480 }, { "epoch": 1.5, "learning_rate": 1.5044567574078536e-05, "loss": 0.4008, "step": 12490 }, { "epoch": 1.51, "learning_rate": 1.5056612864370031e-05, "loss": 0.4373, "step": 12500 }, { "epoch": 1.51, "learning_rate": 1.5068658154661527e-05, "loss": 0.1429, "step": 12510 }, { "epoch": 1.51, "learning_rate": 1.5080703444953024e-05, "loss": 0.2767, "step": 12520 }, { "epoch": 1.51, "learning_rate": 1.5092748735244519e-05, "loss": 0.3339, "step": 12530 }, { "epoch": 1.51, "learning_rate": 1.5104794025536016e-05, "loss": 0.4184, "step": 12540 }, { "epoch": 1.51, "learning_rate": 1.5116839315827514e-05, "loss": 0.3385, "step": 12550 }, { "epoch": 1.51, "learning_rate": 1.512888460611901e-05, "loss": 0.2424, "step": 12560 }, { "epoch": 1.51, "learning_rate": 1.5140929896410505e-05, "loss": 0.195, "step": 12570 }, { "epoch": 1.52, "learning_rate": 1.5152975186702001e-05, "loss": 0.3646, "step": 12580 }, { "epoch": 1.52, "learning_rate": 1.5165020476993497e-05, "loss": 0.2277, "step": 12590 }, { "epoch": 1.52, "learning_rate": 1.5177065767284992e-05, "loss": 0.1904, "step": 12600 }, { "epoch": 1.52, "learning_rate": 1.5189111057576489e-05, "loss": 0.4617, "step": 12610 }, { "epoch": 1.52, "learning_rate": 1.5201156347867984e-05, "loss": 0.2848, "step": 12620 }, { "epoch": 1.52, "learning_rate": 1.521320163815948e-05, "loss": 0.3798, "step": 12630 }, { "epoch": 1.52, "learning_rate": 1.5225246928450976e-05, "loss": 0.2946, "step": 12640 }, { "epoch": 1.52, "learning_rate": 1.5237292218742471e-05, "loss": 0.3064, "step": 12650 }, { "epoch": 1.52, "learning_rate": 1.5249337509033968e-05, "loss": 0.4657, "step": 12660 }, { "epoch": 1.53, "learning_rate": 1.5261382799325463e-05, "loss": 0.348, "step": 12670 }, { "epoch": 1.53, "learning_rate": 1.5273428089616962e-05, "loss": 0.3528, "step": 12680 }, { "epoch": 1.53, "learning_rate": 1.5285473379908457e-05, "loss": 0.2054, "step": 12690 }, { "epoch": 1.53, "learning_rate": 1.5297518670199952e-05, "loss": 0.5211, "step": 12700 }, { "epoch": 1.53, "learning_rate": 1.530956396049145e-05, "loss": 0.3843, "step": 12710 }, { "epoch": 1.53, "learning_rate": 1.5321609250782946e-05, "loss": 0.5084, "step": 12720 }, { "epoch": 1.53, "learning_rate": 1.533365454107444e-05, "loss": 0.2415, "step": 12730 }, { "epoch": 1.53, "learning_rate": 1.5345699831365936e-05, "loss": 0.4332, "step": 12740 }, { "epoch": 1.54, "learning_rate": 1.535774512165743e-05, "loss": 0.2622, "step": 12750 }, { "epoch": 1.54, "learning_rate": 1.536979041194893e-05, "loss": 0.2234, "step": 12760 }, { "epoch": 1.54, "learning_rate": 1.5381835702240425e-05, "loss": 0.4075, "step": 12770 }, { "epoch": 1.54, "learning_rate": 1.539388099253192e-05, "loss": 0.3508, "step": 12780 }, { "epoch": 1.54, "learning_rate": 1.5405926282823416e-05, "loss": 0.5162, "step": 12790 }, { "epoch": 1.54, "learning_rate": 1.541797157311491e-05, "loss": 0.4172, "step": 12800 }, { "epoch": 1.54, "learning_rate": 1.543001686340641e-05, "loss": 0.2841, "step": 12810 }, { "epoch": 1.54, "learning_rate": 1.5442062153697904e-05, "loss": 0.4602, "step": 12820 }, { "epoch": 1.55, "learning_rate": 1.5454107443989403e-05, "loss": 0.3063, "step": 12830 }, { "epoch": 1.55, "learning_rate": 1.5466152734280898e-05, "loss": 0.3252, "step": 12840 }, { "epoch": 1.55, "learning_rate": 1.5478198024572393e-05, "loss": 0.3649, "step": 12850 }, { "epoch": 1.55, "learning_rate": 1.549024331486389e-05, "loss": 0.4884, "step": 12860 }, { "epoch": 1.55, "learning_rate": 1.5502288605155384e-05, "loss": 0.3019, "step": 12870 }, { "epoch": 1.55, "learning_rate": 1.5514333895446882e-05, "loss": 0.3976, "step": 12880 }, { "epoch": 1.55, "learning_rate": 1.5526379185738378e-05, "loss": 0.2001, "step": 12890 }, { "epoch": 1.55, "learning_rate": 1.5538424476029873e-05, "loss": 0.291, "step": 12900 }, { "epoch": 1.55, "learning_rate": 1.5550469766321368e-05, "loss": 0.1655, "step": 12910 }, { "epoch": 1.56, "learning_rate": 1.5562515056612863e-05, "loss": 0.1467, "step": 12920 }, { "epoch": 1.56, "learning_rate": 1.557456034690436e-05, "loss": 0.4834, "step": 12930 }, { "epoch": 1.56, "learning_rate": 1.5586605637195857e-05, "loss": 0.3435, "step": 12940 }, { "epoch": 1.56, "learning_rate": 1.5598650927487355e-05, "loss": 0.5977, "step": 12950 }, { "epoch": 1.56, "learning_rate": 1.561069621777885e-05, "loss": 0.2923, "step": 12960 }, { "epoch": 1.56, "learning_rate": 1.5622741508070346e-05, "loss": 0.3485, "step": 12970 }, { "epoch": 1.56, "learning_rate": 1.563478679836184e-05, "loss": 0.2219, "step": 12980 }, { "epoch": 1.56, "learning_rate": 1.5646832088653336e-05, "loss": 0.2824, "step": 12990 }, { "epoch": 1.57, "learning_rate": 1.5658877378944835e-05, "loss": 0.2508, "step": 13000 }, { "epoch": 1.57, "learning_rate": 1.567092266923633e-05, "loss": 0.2949, "step": 13010 }, { "epoch": 1.57, "learning_rate": 1.5682967959527825e-05, "loss": 0.2749, "step": 13020 }, { "epoch": 1.57, "learning_rate": 1.569501324981932e-05, "loss": 0.2485, "step": 13030 }, { "epoch": 1.57, "learning_rate": 1.5707058540110816e-05, "loss": 0.3601, "step": 13040 }, { "epoch": 1.57, "learning_rate": 1.5719103830402314e-05, "loss": 0.2874, "step": 13050 }, { "epoch": 1.57, "learning_rate": 1.573114912069381e-05, "loss": 0.4797, "step": 13060 }, { "epoch": 1.57, "learning_rate": 1.5743194410985304e-05, "loss": 0.2689, "step": 13070 }, { "epoch": 1.58, "learning_rate": 1.5755239701276803e-05, "loss": 0.4906, "step": 13080 }, { "epoch": 1.58, "learning_rate": 1.5767284991568298e-05, "loss": 0.4924, "step": 13090 }, { "epoch": 1.58, "learning_rate": 1.5779330281859793e-05, "loss": 0.2001, "step": 13100 }, { "epoch": 1.58, "learning_rate": 1.579137557215129e-05, "loss": 0.4605, "step": 13110 }, { "epoch": 1.58, "learning_rate": 1.5803420862442787e-05, "loss": 0.3408, "step": 13120 }, { "epoch": 1.58, "learning_rate": 1.5815466152734282e-05, "loss": 0.2144, "step": 13130 }, { "epoch": 1.58, "learning_rate": 1.5827511443025778e-05, "loss": 0.6161, "step": 13140 }, { "epoch": 1.58, "learning_rate": 1.5839556733317273e-05, "loss": 0.3949, "step": 13150 }, { "epoch": 1.59, "learning_rate": 1.5851602023608768e-05, "loss": 0.3768, "step": 13160 }, { "epoch": 1.59, "learning_rate": 1.5863647313900266e-05, "loss": 0.2804, "step": 13170 }, { "epoch": 1.59, "learning_rate": 1.587569260419176e-05, "loss": 0.4068, "step": 13180 }, { "epoch": 1.59, "learning_rate": 1.5887737894483257e-05, "loss": 0.4433, "step": 13190 }, { "epoch": 1.59, "learning_rate": 1.5899783184774752e-05, "loss": 0.294, "step": 13200 }, { "epoch": 1.59, "learning_rate": 1.591182847506625e-05, "loss": 0.4707, "step": 13210 }, { "epoch": 1.59, "learning_rate": 1.5923873765357746e-05, "loss": 0.2421, "step": 13220 }, { "epoch": 1.59, "learning_rate": 1.5935919055649244e-05, "loss": 0.2884, "step": 13230 }, { "epoch": 1.59, "learning_rate": 1.594796434594074e-05, "loss": 0.4972, "step": 13240 }, { "epoch": 1.6, "learning_rate": 1.5960009636232235e-05, "loss": 0.3178, "step": 13250 }, { "epoch": 1.6, "learning_rate": 1.597205492652373e-05, "loss": 0.2397, "step": 13260 }, { "epoch": 1.6, "learning_rate": 1.5984100216815225e-05, "loss": 0.3195, "step": 13270 }, { "epoch": 1.6, "learning_rate": 1.599614550710672e-05, "loss": 0.3819, "step": 13280 }, { "epoch": 1.6, "learning_rate": 1.600819079739822e-05, "loss": 0.3761, "step": 13290 }, { "epoch": 1.6, "learning_rate": 1.6020236087689714e-05, "loss": 0.2743, "step": 13300 }, { "epoch": 1.6, "learning_rate": 1.603228137798121e-05, "loss": 0.3008, "step": 13310 }, { "epoch": 1.6, "learning_rate": 1.6044326668272704e-05, "loss": 0.4692, "step": 13320 }, { "epoch": 1.61, "learning_rate": 1.60563719585642e-05, "loss": 0.4879, "step": 13330 }, { "epoch": 1.61, "learning_rate": 1.6068417248855698e-05, "loss": 0.4925, "step": 13340 }, { "epoch": 1.61, "learning_rate": 1.6080462539147197e-05, "loss": 0.2146, "step": 13350 }, { "epoch": 1.61, "learning_rate": 1.6092507829438692e-05, "loss": 0.2931, "step": 13360 }, { "epoch": 1.61, "learning_rate": 1.6104553119730187e-05, "loss": 0.3135, "step": 13370 }, { "epoch": 1.61, "learning_rate": 1.6116598410021682e-05, "loss": 0.4451, "step": 13380 }, { "epoch": 1.61, "learning_rate": 1.6128643700313177e-05, "loss": 0.5261, "step": 13390 }, { "epoch": 1.61, "learning_rate": 1.6140688990604673e-05, "loss": 0.3075, "step": 13400 }, { "epoch": 1.62, "learning_rate": 1.615273428089617e-05, "loss": 0.4711, "step": 13410 }, { "epoch": 1.62, "learning_rate": 1.6164779571187666e-05, "loss": 0.4821, "step": 13420 }, { "epoch": 1.62, "learning_rate": 1.617682486147916e-05, "loss": 0.1759, "step": 13430 }, { "epoch": 1.62, "learning_rate": 1.6188870151770657e-05, "loss": 0.2648, "step": 13440 }, { "epoch": 1.62, "learning_rate": 1.6200915442062152e-05, "loss": 0.2824, "step": 13450 }, { "epoch": 1.62, "learning_rate": 1.621296073235365e-05, "loss": 0.4145, "step": 13460 }, { "epoch": 1.62, "learning_rate": 1.622500602264515e-05, "loss": 0.4369, "step": 13470 }, { "epoch": 1.62, "learning_rate": 1.6237051312936644e-05, "loss": 0.3467, "step": 13480 }, { "epoch": 1.62, "learning_rate": 1.624909660322814e-05, "loss": 0.4327, "step": 13490 }, { "epoch": 1.63, "learning_rate": 1.6261141893519635e-05, "loss": 0.3503, "step": 13500 }, { "epoch": 1.63, "learning_rate": 1.627318718381113e-05, "loss": 0.3238, "step": 13510 }, { "epoch": 1.63, "learning_rate": 1.628523247410263e-05, "loss": 0.2933, "step": 13520 }, { "epoch": 1.63, "learning_rate": 1.6297277764394124e-05, "loss": 0.1341, "step": 13530 }, { "epoch": 1.63, "learning_rate": 1.630932305468562e-05, "loss": 0.3176, "step": 13540 }, { "epoch": 1.63, "learning_rate": 1.6321368344977114e-05, "loss": 0.3101, "step": 13550 }, { "epoch": 1.63, "learning_rate": 1.633341363526861e-05, "loss": 0.3854, "step": 13560 }, { "epoch": 1.63, "learning_rate": 1.6345458925560104e-05, "loss": 0.5578, "step": 13570 }, { "epoch": 1.64, "learning_rate": 1.6357504215851603e-05, "loss": 0.2908, "step": 13580 }, { "epoch": 1.64, "learning_rate": 1.6369549506143098e-05, "loss": 0.2367, "step": 13590 }, { "epoch": 1.64, "learning_rate": 1.6381594796434597e-05, "loss": 0.2731, "step": 13600 }, { "epoch": 1.64, "learning_rate": 1.6393640086726092e-05, "loss": 0.4618, "step": 13610 }, { "epoch": 1.64, "learning_rate": 1.6405685377017587e-05, "loss": 0.4616, "step": 13620 }, { "epoch": 1.64, "learning_rate": 1.6417730667309082e-05, "loss": 0.2046, "step": 13630 }, { "epoch": 1.64, "learning_rate": 1.642977595760058e-05, "loss": 0.4087, "step": 13640 }, { "epoch": 1.64, "learning_rate": 1.6441821247892076e-05, "loss": 0.1865, "step": 13650 }, { "epoch": 1.65, "learning_rate": 1.645386653818357e-05, "loss": 0.2905, "step": 13660 }, { "epoch": 1.65, "learning_rate": 1.6465911828475066e-05, "loss": 0.1579, "step": 13670 }, { "epoch": 1.65, "learning_rate": 1.647795711876656e-05, "loss": 0.3411, "step": 13680 }, { "epoch": 1.65, "learning_rate": 1.649000240905806e-05, "loss": 0.6369, "step": 13690 }, { "epoch": 1.65, "learning_rate": 1.6502047699349555e-05, "loss": 0.3428, "step": 13700 }, { "epoch": 1.65, "learning_rate": 1.651409298964105e-05, "loss": 0.4011, "step": 13710 }, { "epoch": 1.65, "learning_rate": 1.6526138279932546e-05, "loss": 0.3921, "step": 13720 }, { "epoch": 1.65, "learning_rate": 1.6538183570224044e-05, "loss": 0.3676, "step": 13730 }, { "epoch": 1.65, "learning_rate": 1.655022886051554e-05, "loss": 0.307, "step": 13740 }, { "epoch": 1.66, "learning_rate": 1.6562274150807035e-05, "loss": 0.2852, "step": 13750 }, { "epoch": 1.66, "learning_rate": 1.6574319441098533e-05, "loss": 0.41, "step": 13760 }, { "epoch": 1.66, "learning_rate": 1.658636473139003e-05, "loss": 0.076, "step": 13770 }, { "epoch": 1.66, "learning_rate": 1.6598410021681524e-05, "loss": 0.258, "step": 13780 }, { "epoch": 1.66, "learning_rate": 1.661045531197302e-05, "loss": 0.0401, "step": 13790 }, { "epoch": 1.66, "learning_rate": 1.6622500602264514e-05, "loss": 0.2154, "step": 13800 }, { "epoch": 1.66, "learning_rate": 1.6634545892556013e-05, "loss": 0.3682, "step": 13810 }, { "epoch": 1.66, "learning_rate": 1.6646591182847508e-05, "loss": 0.5444, "step": 13820 }, { "epoch": 1.67, "learning_rate": 1.6658636473139003e-05, "loss": 0.4248, "step": 13830 }, { "epoch": 1.67, "learning_rate": 1.6670681763430498e-05, "loss": 0.4025, "step": 13840 }, { "epoch": 1.67, "learning_rate": 1.6682727053721993e-05, "loss": 0.1882, "step": 13850 }, { "epoch": 1.67, "learning_rate": 1.6694772344013492e-05, "loss": 0.2606, "step": 13860 }, { "epoch": 1.67, "learning_rate": 1.6706817634304987e-05, "loss": 0.2605, "step": 13870 }, { "epoch": 1.67, "learning_rate": 1.6718862924596486e-05, "loss": 0.3162, "step": 13880 }, { "epoch": 1.67, "learning_rate": 1.673090821488798e-05, "loss": 0.3562, "step": 13890 }, { "epoch": 1.67, "learning_rate": 1.6742953505179476e-05, "loss": 0.2218, "step": 13900 }, { "epoch": 1.68, "learning_rate": 1.675499879547097e-05, "loss": 0.4068, "step": 13910 }, { "epoch": 1.68, "learning_rate": 1.6767044085762466e-05, "loss": 0.465, "step": 13920 }, { "epoch": 1.68, "learning_rate": 1.6779089376053965e-05, "loss": 0.2806, "step": 13930 }, { "epoch": 1.68, "learning_rate": 1.679113466634546e-05, "loss": 0.4725, "step": 13940 }, { "epoch": 1.68, "learning_rate": 1.6803179956636955e-05, "loss": 0.216, "step": 13950 }, { "epoch": 1.68, "learning_rate": 1.681522524692845e-05, "loss": 0.2318, "step": 13960 }, { "epoch": 1.68, "learning_rate": 1.6827270537219946e-05, "loss": 0.1856, "step": 13970 }, { "epoch": 1.68, "learning_rate": 1.6839315827511444e-05, "loss": 0.3509, "step": 13980 }, { "epoch": 1.69, "learning_rate": 1.6851361117802943e-05, "loss": 0.4952, "step": 13990 }, { "epoch": 1.69, "learning_rate": 1.6863406408094438e-05, "loss": 0.124, "step": 14000 }, { "epoch": 1.69, "learning_rate": 1.6875451698385933e-05, "loss": 0.2282, "step": 14010 }, { "epoch": 1.69, "learning_rate": 1.688749698867743e-05, "loss": 0.3701, "step": 14020 }, { "epoch": 1.69, "learning_rate": 1.6899542278968924e-05, "loss": 0.4223, "step": 14030 }, { "epoch": 1.69, "learning_rate": 1.691158756926042e-05, "loss": 0.202, "step": 14040 }, { "epoch": 1.69, "learning_rate": 1.6923632859551917e-05, "loss": 0.3866, "step": 14050 }, { "epoch": 1.69, "learning_rate": 1.6935678149843412e-05, "loss": 0.3391, "step": 14060 }, { "epoch": 1.69, "learning_rate": 1.6947723440134908e-05, "loss": 0.3552, "step": 14070 }, { "epoch": 1.7, "learning_rate": 1.6959768730426403e-05, "loss": 0.2743, "step": 14080 }, { "epoch": 1.7, "learning_rate": 1.6971814020717898e-05, "loss": 0.1825, "step": 14090 }, { "epoch": 1.7, "learning_rate": 1.6983859311009397e-05, "loss": 0.454, "step": 14100 }, { "epoch": 1.7, "learning_rate": 1.6995904601300892e-05, "loss": 0.2734, "step": 14110 }, { "epoch": 1.7, "learning_rate": 1.7007949891592387e-05, "loss": 0.5079, "step": 14120 }, { "epoch": 1.7, "learning_rate": 1.7019995181883886e-05, "loss": 0.2319, "step": 14130 }, { "epoch": 1.7, "learning_rate": 1.703204047217538e-05, "loss": 0.56, "step": 14140 }, { "epoch": 1.7, "learning_rate": 1.7044085762466876e-05, "loss": 0.3642, "step": 14150 }, { "epoch": 1.71, "learning_rate": 1.7056131052758374e-05, "loss": 0.4052, "step": 14160 }, { "epoch": 1.71, "learning_rate": 1.706817634304987e-05, "loss": 0.3666, "step": 14170 }, { "epoch": 1.71, "learning_rate": 1.7080221633341365e-05, "loss": 0.4243, "step": 14180 }, { "epoch": 1.71, "learning_rate": 1.709226692363286e-05, "loss": 0.47, "step": 14190 }, { "epoch": 1.71, "learning_rate": 1.7104312213924355e-05, "loss": 0.318, "step": 14200 }, { "epoch": 1.71, "learning_rate": 1.711635750421585e-05, "loss": 0.3056, "step": 14210 }, { "epoch": 1.71, "learning_rate": 1.712840279450735e-05, "loss": 0.2701, "step": 14220 }, { "epoch": 1.71, "learning_rate": 1.7140448084798844e-05, "loss": 0.4098, "step": 14230 }, { "epoch": 1.72, "learning_rate": 1.715249337509034e-05, "loss": 0.3387, "step": 14240 }, { "epoch": 1.72, "learning_rate": 1.7164538665381835e-05, "loss": 0.3452, "step": 14250 }, { "epoch": 1.72, "learning_rate": 1.7176583955673333e-05, "loss": 0.3347, "step": 14260 }, { "epoch": 1.72, "learning_rate": 1.7188629245964828e-05, "loss": 0.1865, "step": 14270 }, { "epoch": 1.72, "learning_rate": 1.7200674536256327e-05, "loss": 0.2718, "step": 14280 }, { "epoch": 1.72, "learning_rate": 1.7212719826547822e-05, "loss": 0.4819, "step": 14290 }, { "epoch": 1.72, "learning_rate": 1.7224765116839317e-05, "loss": 0.3818, "step": 14300 }, { "epoch": 1.72, "learning_rate": 1.7236810407130812e-05, "loss": 0.2123, "step": 14310 }, { "epoch": 1.72, "learning_rate": 1.7248855697422308e-05, "loss": 0.2762, "step": 14320 }, { "epoch": 1.73, "learning_rate": 1.7260900987713806e-05, "loss": 0.3271, "step": 14330 }, { "epoch": 1.73, "learning_rate": 1.72729462780053e-05, "loss": 0.2539, "step": 14340 }, { "epoch": 1.73, "learning_rate": 1.7284991568296797e-05, "loss": 0.2468, "step": 14350 }, { "epoch": 1.73, "learning_rate": 1.7297036858588292e-05, "loss": 0.3247, "step": 14360 }, { "epoch": 1.73, "learning_rate": 1.7309082148879787e-05, "loss": 0.3987, "step": 14370 }, { "epoch": 1.73, "learning_rate": 1.7321127439171282e-05, "loss": 0.2014, "step": 14380 }, { "epoch": 1.73, "learning_rate": 1.733317272946278e-05, "loss": 0.3405, "step": 14390 }, { "epoch": 1.73, "learning_rate": 1.734521801975428e-05, "loss": 0.3838, "step": 14400 }, { "epoch": 1.74, "learning_rate": 1.7357263310045774e-05, "loss": 0.3863, "step": 14410 }, { "epoch": 1.74, "learning_rate": 1.736930860033727e-05, "loss": 0.5188, "step": 14420 }, { "epoch": 1.74, "learning_rate": 1.7381353890628765e-05, "loss": 0.1735, "step": 14430 }, { "epoch": 1.74, "learning_rate": 1.739339918092026e-05, "loss": 0.2148, "step": 14440 }, { "epoch": 1.74, "learning_rate": 1.740544447121176e-05, "loss": 0.4637, "step": 14450 }, { "epoch": 1.74, "learning_rate": 1.7417489761503254e-05, "loss": 0.2373, "step": 14460 }, { "epoch": 1.74, "learning_rate": 1.742953505179475e-05, "loss": 0.2514, "step": 14470 }, { "epoch": 1.74, "learning_rate": 1.7441580342086244e-05, "loss": 0.3044, "step": 14480 }, { "epoch": 1.75, "learning_rate": 1.745362563237774e-05, "loss": 0.3063, "step": 14490 }, { "epoch": 1.75, "learning_rate": 1.7465670922669234e-05, "loss": 0.2131, "step": 14500 }, { "epoch": 1.75, "learning_rate": 1.7477716212960733e-05, "loss": 0.3743, "step": 14510 }, { "epoch": 1.75, "learning_rate": 1.748976150325223e-05, "loss": 0.2936, "step": 14520 }, { "epoch": 1.75, "learning_rate": 1.7501806793543727e-05, "loss": 0.2125, "step": 14530 }, { "epoch": 1.75, "learning_rate": 1.7513852083835222e-05, "loss": 0.4889, "step": 14540 }, { "epoch": 1.75, "learning_rate": 1.7525897374126717e-05, "loss": 0.4442, "step": 14550 }, { "epoch": 1.75, "learning_rate": 1.7537942664418212e-05, "loss": 0.2729, "step": 14560 }, { "epoch": 1.75, "learning_rate": 1.754998795470971e-05, "loss": 0.3475, "step": 14570 }, { "epoch": 1.76, "learning_rate": 1.7562033245001206e-05, "loss": 0.3476, "step": 14580 }, { "epoch": 1.76, "learning_rate": 1.75740785352927e-05, "loss": 0.3827, "step": 14590 }, { "epoch": 1.76, "learning_rate": 1.7586123825584196e-05, "loss": 0.309, "step": 14600 }, { "epoch": 1.76, "learning_rate": 1.759816911587569e-05, "loss": 0.2272, "step": 14610 }, { "epoch": 1.76, "learning_rate": 1.761021440616719e-05, "loss": 0.2344, "step": 14620 }, { "epoch": 1.76, "learning_rate": 1.7622259696458685e-05, "loss": 0.1675, "step": 14630 }, { "epoch": 1.76, "learning_rate": 1.763430498675018e-05, "loss": 0.3641, "step": 14640 }, { "epoch": 1.76, "learning_rate": 1.764635027704168e-05, "loss": 0.2936, "step": 14650 }, { "epoch": 1.77, "learning_rate": 1.7658395567333174e-05, "loss": 0.411, "step": 14660 }, { "epoch": 1.77, "learning_rate": 1.767044085762467e-05, "loss": 0.4164, "step": 14670 }, { "epoch": 1.77, "learning_rate": 1.7682486147916165e-05, "loss": 0.2581, "step": 14680 }, { "epoch": 1.77, "learning_rate": 1.7694531438207663e-05, "loss": 0.1938, "step": 14690 }, { "epoch": 1.77, "learning_rate": 1.770657672849916e-05, "loss": 0.2734, "step": 14700 }, { "epoch": 1.77, "learning_rate": 1.7718622018790654e-05, "loss": 0.4574, "step": 14710 }, { "epoch": 1.77, "learning_rate": 1.773066730908215e-05, "loss": 0.3181, "step": 14720 }, { "epoch": 1.77, "learning_rate": 1.7742712599373644e-05, "loss": 0.353, "step": 14730 }, { "epoch": 1.78, "learning_rate": 1.7754757889665143e-05, "loss": 0.3924, "step": 14740 }, { "epoch": 1.78, "learning_rate": 1.7766803179956638e-05, "loss": 0.2645, "step": 14750 }, { "epoch": 1.78, "learning_rate": 1.7778848470248133e-05, "loss": 0.2565, "step": 14760 }, { "epoch": 1.78, "learning_rate": 1.7790893760539628e-05, "loss": 0.3047, "step": 14770 }, { "epoch": 1.78, "learning_rate": 1.7802939050831127e-05, "loss": 0.2603, "step": 14780 }, { "epoch": 1.78, "learning_rate": 1.7814984341122622e-05, "loss": 0.2049, "step": 14790 }, { "epoch": 1.78, "learning_rate": 1.782702963141412e-05, "loss": 0.2337, "step": 14800 }, { "epoch": 1.78, "learning_rate": 1.7839074921705616e-05, "loss": 0.2218, "step": 14810 }, { "epoch": 1.79, "learning_rate": 1.785112021199711e-05, "loss": 0.3291, "step": 14820 }, { "epoch": 1.79, "learning_rate": 1.7863165502288606e-05, "loss": 0.2858, "step": 14830 }, { "epoch": 1.79, "learning_rate": 1.78752107925801e-05, "loss": 0.1728, "step": 14840 }, { "epoch": 1.79, "learning_rate": 1.7887256082871596e-05, "loss": 0.2959, "step": 14850 }, { "epoch": 1.79, "learning_rate": 1.7899301373163095e-05, "loss": 0.3913, "step": 14860 }, { "epoch": 1.79, "learning_rate": 1.791134666345459e-05, "loss": 0.2921, "step": 14870 }, { "epoch": 1.79, "learning_rate": 1.7923391953746085e-05, "loss": 0.296, "step": 14880 }, { "epoch": 1.79, "learning_rate": 1.793543724403758e-05, "loss": 0.2029, "step": 14890 }, { "epoch": 1.79, "learning_rate": 1.7947482534329076e-05, "loss": 0.5612, "step": 14900 }, { "epoch": 1.8, "learning_rate": 1.7959527824620574e-05, "loss": 0.3015, "step": 14910 }, { "epoch": 1.8, "learning_rate": 1.7971573114912073e-05, "loss": 0.2218, "step": 14920 }, { "epoch": 1.8, "learning_rate": 1.7983618405203568e-05, "loss": 0.1559, "step": 14930 }, { "epoch": 1.8, "learning_rate": 1.7995663695495063e-05, "loss": 0.2545, "step": 14940 }, { "epoch": 1.8, "learning_rate": 1.800770898578656e-05, "loss": 0.3471, "step": 14950 }, { "epoch": 1.8, "learning_rate": 1.8019754276078054e-05, "loss": 0.2408, "step": 14960 }, { "epoch": 1.8, "learning_rate": 1.803179956636955e-05, "loss": 0.3018, "step": 14970 }, { "epoch": 1.8, "learning_rate": 1.8043844856661047e-05, "loss": 0.2983, "step": 14980 }, { "epoch": 1.81, "learning_rate": 1.8055890146952543e-05, "loss": 0.2574, "step": 14990 }, { "epoch": 1.81, "learning_rate": 1.8067935437244038e-05, "loss": 0.2971, "step": 15000 }, { "epoch": 1.81, "learning_rate": 1.8079980727535533e-05, "loss": 0.3266, "step": 15010 }, { "epoch": 1.81, "learning_rate": 1.8092026017827028e-05, "loss": 0.4204, "step": 15020 }, { "epoch": 1.81, "learning_rate": 1.8104071308118527e-05, "loss": 0.2603, "step": 15030 }, { "epoch": 1.81, "learning_rate": 1.8116116598410022e-05, "loss": 0.3498, "step": 15040 }, { "epoch": 1.81, "learning_rate": 1.812816188870152e-05, "loss": 0.2594, "step": 15050 }, { "epoch": 1.81, "learning_rate": 1.8140207178993016e-05, "loss": 0.3597, "step": 15060 }, { "epoch": 1.82, "learning_rate": 1.815225246928451e-05, "loss": 0.4581, "step": 15070 }, { "epoch": 1.82, "learning_rate": 1.8164297759576006e-05, "loss": 0.2563, "step": 15080 }, { "epoch": 1.82, "learning_rate": 1.8176343049867505e-05, "loss": 0.2272, "step": 15090 }, { "epoch": 1.82, "learning_rate": 1.8188388340159e-05, "loss": 0.3986, "step": 15100 }, { "epoch": 1.82, "learning_rate": 1.8200433630450495e-05, "loss": 0.3184, "step": 15110 }, { "epoch": 1.82, "learning_rate": 1.821247892074199e-05, "loss": 0.5686, "step": 15120 }, { "epoch": 1.82, "learning_rate": 1.8224524211033485e-05, "loss": 0.5083, "step": 15130 }, { "epoch": 1.82, "learning_rate": 1.823656950132498e-05, "loss": 0.1569, "step": 15140 }, { "epoch": 1.82, "learning_rate": 1.824861479161648e-05, "loss": 0.3912, "step": 15150 }, { "epoch": 1.83, "learning_rate": 1.8260660081907974e-05, "loss": 0.4585, "step": 15160 }, { "epoch": 1.83, "learning_rate": 1.827270537219947e-05, "loss": 0.219, "step": 15170 }, { "epoch": 1.83, "learning_rate": 1.8284750662490968e-05, "loss": 0.2725, "step": 15180 }, { "epoch": 1.83, "learning_rate": 1.8296795952782463e-05, "loss": 0.1995, "step": 15190 }, { "epoch": 1.83, "learning_rate": 1.830884124307396e-05, "loss": 0.5374, "step": 15200 }, { "epoch": 1.83, "learning_rate": 1.8320886533365457e-05, "loss": 0.2917, "step": 15210 }, { "epoch": 1.83, "learning_rate": 1.8332931823656952e-05, "loss": 0.2898, "step": 15220 }, { "epoch": 1.83, "learning_rate": 1.8344977113948447e-05, "loss": 0.2882, "step": 15230 }, { "epoch": 1.84, "learning_rate": 1.8357022404239943e-05, "loss": 0.4494, "step": 15240 }, { "epoch": 1.84, "learning_rate": 1.8369067694531438e-05, "loss": 0.4427, "step": 15250 }, { "epoch": 1.84, "learning_rate": 1.8381112984822936e-05, "loss": 0.6007, "step": 15260 }, { "epoch": 1.84, "learning_rate": 1.839315827511443e-05, "loss": 0.18, "step": 15270 }, { "epoch": 1.84, "learning_rate": 1.8405203565405927e-05, "loss": 0.4799, "step": 15280 }, { "epoch": 1.84, "learning_rate": 1.8417248855697422e-05, "loss": 0.3997, "step": 15290 }, { "epoch": 1.84, "learning_rate": 1.8429294145988917e-05, "loss": 0.2451, "step": 15300 }, { "epoch": 1.84, "learning_rate": 1.8441339436280416e-05, "loss": 0.1734, "step": 15310 }, { "epoch": 1.85, "learning_rate": 1.845338472657191e-05, "loss": 0.4535, "step": 15320 }, { "epoch": 1.85, "learning_rate": 1.846543001686341e-05, "loss": 0.508, "step": 15330 }, { "epoch": 1.85, "learning_rate": 1.8477475307154905e-05, "loss": 0.2511, "step": 15340 }, { "epoch": 1.85, "learning_rate": 1.84895205974464e-05, "loss": 0.285, "step": 15350 }, { "epoch": 1.85, "learning_rate": 1.8501565887737895e-05, "loss": 0.4584, "step": 15360 }, { "epoch": 1.85, "learning_rate": 1.851361117802939e-05, "loss": 0.2, "step": 15370 }, { "epoch": 1.85, "learning_rate": 1.852565646832089e-05, "loss": 0.3063, "step": 15380 }, { "epoch": 1.85, "learning_rate": 1.8537701758612384e-05, "loss": 0.244, "step": 15390 }, { "epoch": 1.85, "learning_rate": 1.854974704890388e-05, "loss": 0.406, "step": 15400 }, { "epoch": 1.86, "learning_rate": 1.8561792339195374e-05, "loss": 0.1243, "step": 15410 }, { "epoch": 1.86, "learning_rate": 1.857383762948687e-05, "loss": 0.2677, "step": 15420 }, { "epoch": 1.86, "learning_rate": 1.8585882919778365e-05, "loss": 0.271, "step": 15430 }, { "epoch": 1.86, "learning_rate": 1.8597928210069863e-05, "loss": 0.1796, "step": 15440 }, { "epoch": 1.86, "learning_rate": 1.8609973500361362e-05, "loss": 0.4569, "step": 15450 }, { "epoch": 1.86, "learning_rate": 1.8622018790652857e-05, "loss": 0.3373, "step": 15460 }, { "epoch": 1.86, "learning_rate": 1.8634064080944352e-05, "loss": 0.3345, "step": 15470 }, { "epoch": 1.86, "learning_rate": 1.8646109371235847e-05, "loss": 0.2536, "step": 15480 }, { "epoch": 1.87, "learning_rate": 1.8658154661527342e-05, "loss": 0.3955, "step": 15490 }, { "epoch": 1.87, "learning_rate": 1.867019995181884e-05, "loss": 0.3549, "step": 15500 }, { "epoch": 1.87, "learning_rate": 1.8682245242110336e-05, "loss": 0.4603, "step": 15510 }, { "epoch": 1.87, "learning_rate": 1.869429053240183e-05, "loss": 0.3886, "step": 15520 }, { "epoch": 1.87, "learning_rate": 1.8706335822693327e-05, "loss": 0.5313, "step": 15530 }, { "epoch": 1.87, "learning_rate": 1.8718381112984822e-05, "loss": 0.2165, "step": 15540 }, { "epoch": 1.87, "learning_rate": 1.873042640327632e-05, "loss": 0.2749, "step": 15550 }, { "epoch": 1.87, "learning_rate": 1.8742471693567816e-05, "loss": 0.3809, "step": 15560 }, { "epoch": 1.88, "learning_rate": 1.8754516983859314e-05, "loss": 0.3826, "step": 15570 }, { "epoch": 1.88, "learning_rate": 1.876656227415081e-05, "loss": 0.2446, "step": 15580 }, { "epoch": 1.88, "learning_rate": 1.8778607564442305e-05, "loss": 0.396, "step": 15590 }, { "epoch": 1.88, "learning_rate": 1.87906528547338e-05, "loss": 0.2011, "step": 15600 }, { "epoch": 1.88, "learning_rate": 1.8802698145025295e-05, "loss": 0.1042, "step": 15610 }, { "epoch": 1.88, "learning_rate": 1.8814743435316793e-05, "loss": 0.3822, "step": 15620 }, { "epoch": 1.88, "learning_rate": 1.882678872560829e-05, "loss": 0.3027, "step": 15630 }, { "epoch": 1.88, "learning_rate": 1.8838834015899784e-05, "loss": 0.3612, "step": 15640 }, { "epoch": 1.88, "learning_rate": 1.885087930619128e-05, "loss": 0.3436, "step": 15650 }, { "epoch": 1.89, "learning_rate": 1.8862924596482774e-05, "loss": 0.2867, "step": 15660 }, { "epoch": 1.89, "learning_rate": 1.8874969886774273e-05, "loss": 0.1464, "step": 15670 }, { "epoch": 1.89, "learning_rate": 1.8887015177065768e-05, "loss": 0.3208, "step": 15680 }, { "epoch": 1.89, "learning_rate": 1.8899060467357263e-05, "loss": 0.4024, "step": 15690 }, { "epoch": 1.89, "learning_rate": 1.8911105757648762e-05, "loss": 0.3799, "step": 15700 }, { "epoch": 1.89, "learning_rate": 1.8923151047940257e-05, "loss": 0.4071, "step": 15710 }, { "epoch": 1.89, "learning_rate": 1.8935196338231752e-05, "loss": 0.201, "step": 15720 }, { "epoch": 1.89, "learning_rate": 1.894724162852325e-05, "loss": 0.2279, "step": 15730 }, { "epoch": 1.9, "learning_rate": 1.8959286918814746e-05, "loss": 0.3912, "step": 15740 }, { "epoch": 1.9, "learning_rate": 1.897133220910624e-05, "loss": 0.3171, "step": 15750 }, { "epoch": 1.9, "learning_rate": 1.8983377499397736e-05, "loss": 0.3496, "step": 15760 }, { "epoch": 1.9, "learning_rate": 1.899542278968923e-05, "loss": 0.4952, "step": 15770 }, { "epoch": 1.9, "learning_rate": 1.9007468079980727e-05, "loss": 0.4063, "step": 15780 }, { "epoch": 1.9, "learning_rate": 1.9019513370272225e-05, "loss": 0.4042, "step": 15790 }, { "epoch": 1.9, "learning_rate": 1.903155866056372e-05, "loss": 0.2415, "step": 15800 }, { "epoch": 1.9, "learning_rate": 1.9043603950855216e-05, "loss": 0.3367, "step": 15810 }, { "epoch": 1.91, "learning_rate": 1.905564924114671e-05, "loss": 0.2631, "step": 15820 }, { "epoch": 1.91, "learning_rate": 1.906769453143821e-05, "loss": 0.4188, "step": 15830 }, { "epoch": 1.91, "learning_rate": 1.9079739821729704e-05, "loss": 0.3314, "step": 15840 }, { "epoch": 1.91, "learning_rate": 1.9091785112021203e-05, "loss": 0.3671, "step": 15850 }, { "epoch": 1.91, "learning_rate": 1.9103830402312698e-05, "loss": 0.5761, "step": 15860 }, { "epoch": 1.91, "learning_rate": 1.9115875692604193e-05, "loss": 0.3058, "step": 15870 }, { "epoch": 1.91, "learning_rate": 1.912792098289569e-05, "loss": 0.2773, "step": 15880 }, { "epoch": 1.91, "learning_rate": 1.9139966273187184e-05, "loss": 0.2895, "step": 15890 }, { "epoch": 1.92, "learning_rate": 1.915201156347868e-05, "loss": 0.4341, "step": 15900 }, { "epoch": 1.92, "learning_rate": 1.9164056853770178e-05, "loss": 0.3176, "step": 15910 }, { "epoch": 1.92, "learning_rate": 1.9176102144061673e-05, "loss": 0.4531, "step": 15920 }, { "epoch": 1.92, "learning_rate": 1.9188147434353168e-05, "loss": 0.3601, "step": 15930 }, { "epoch": 1.92, "learning_rate": 1.9200192724644663e-05, "loss": 0.3109, "step": 15940 }, { "epoch": 1.92, "learning_rate": 1.9212238014936158e-05, "loss": 0.3579, "step": 15950 }, { "epoch": 1.92, "learning_rate": 1.9224283305227657e-05, "loss": 0.3883, "step": 15960 }, { "epoch": 1.92, "learning_rate": 1.9236328595519155e-05, "loss": 0.2297, "step": 15970 }, { "epoch": 1.92, "learning_rate": 1.924837388581065e-05, "loss": 0.2892, "step": 15980 }, { "epoch": 1.93, "learning_rate": 1.9260419176102146e-05, "loss": 0.1598, "step": 15990 }, { "epoch": 1.93, "learning_rate": 1.927246446639364e-05, "loss": 0.4829, "step": 16000 }, { "epoch": 1.93, "learning_rate": 1.9284509756685136e-05, "loss": 0.4168, "step": 16010 }, { "epoch": 1.93, "learning_rate": 1.9296555046976635e-05, "loss": 0.2772, "step": 16020 }, { "epoch": 1.93, "learning_rate": 1.930860033726813e-05, "loss": 0.2266, "step": 16030 }, { "epoch": 1.93, "learning_rate": 1.9320645627559625e-05, "loss": 0.3934, "step": 16040 }, { "epoch": 1.93, "learning_rate": 1.933269091785112e-05, "loss": 0.3363, "step": 16050 }, { "epoch": 1.93, "learning_rate": 1.9344736208142615e-05, "loss": 0.4, "step": 16060 }, { "epoch": 1.94, "learning_rate": 1.935678149843411e-05, "loss": 0.3144, "step": 16070 }, { "epoch": 1.94, "learning_rate": 1.936882678872561e-05, "loss": 0.3816, "step": 16080 }, { "epoch": 1.94, "learning_rate": 1.9380872079017104e-05, "loss": 0.275, "step": 16090 }, { "epoch": 1.94, "learning_rate": 1.9392917369308603e-05, "loss": 0.3182, "step": 16100 }, { "epoch": 1.94, "learning_rate": 1.9404962659600098e-05, "loss": 0.2334, "step": 16110 }, { "epoch": 1.94, "learning_rate": 1.9417007949891593e-05, "loss": 0.4573, "step": 16120 }, { "epoch": 1.94, "learning_rate": 1.942905324018309e-05, "loss": 0.4261, "step": 16130 }, { "epoch": 1.94, "learning_rate": 1.9441098530474587e-05, "loss": 0.2527, "step": 16140 }, { "epoch": 1.95, "learning_rate": 1.9453143820766082e-05, "loss": 0.2289, "step": 16150 }, { "epoch": 1.95, "learning_rate": 1.9465189111057577e-05, "loss": 0.4979, "step": 16160 }, { "epoch": 1.95, "learning_rate": 1.9477234401349073e-05, "loss": 0.3014, "step": 16170 }, { "epoch": 1.95, "learning_rate": 1.9489279691640568e-05, "loss": 0.2372, "step": 16180 }, { "epoch": 1.95, "learning_rate": 1.9501324981932066e-05, "loss": 0.2293, "step": 16190 }, { "epoch": 1.95, "learning_rate": 1.951337027222356e-05, "loss": 0.514, "step": 16200 }, { "epoch": 1.95, "learning_rate": 1.9525415562515057e-05, "loss": 0.2959, "step": 16210 }, { "epoch": 1.95, "learning_rate": 1.9537460852806552e-05, "loss": 0.4496, "step": 16220 }, { "epoch": 1.95, "learning_rate": 1.954950614309805e-05, "loss": 0.3025, "step": 16230 }, { "epoch": 1.96, "learning_rate": 1.9561551433389546e-05, "loss": 0.3887, "step": 16240 }, { "epoch": 1.96, "learning_rate": 1.957359672368104e-05, "loss": 0.267, "step": 16250 }, { "epoch": 1.96, "learning_rate": 1.958564201397254e-05, "loss": 0.4207, "step": 16260 }, { "epoch": 1.96, "learning_rate": 1.9597687304264035e-05, "loss": 0.3424, "step": 16270 }, { "epoch": 1.96, "learning_rate": 1.960973259455553e-05, "loss": 0.2413, "step": 16280 }, { "epoch": 1.96, "learning_rate": 1.9621777884847025e-05, "loss": 0.1827, "step": 16290 }, { "epoch": 1.96, "learning_rate": 1.963382317513852e-05, "loss": 0.2709, "step": 16300 }, { "epoch": 1.96, "learning_rate": 1.964586846543002e-05, "loss": 0.2257, "step": 16310 }, { "epoch": 1.97, "learning_rate": 1.9657913755721514e-05, "loss": 0.324, "step": 16320 }, { "epoch": 1.97, "learning_rate": 1.966995904601301e-05, "loss": 0.1879, "step": 16330 }, { "epoch": 1.97, "learning_rate": 1.9682004336304504e-05, "loss": 0.2552, "step": 16340 }, { "epoch": 1.97, "learning_rate": 1.9694049626596e-05, "loss": 0.2307, "step": 16350 }, { "epoch": 1.97, "learning_rate": 1.9706094916887498e-05, "loss": 0.3287, "step": 16360 }, { "epoch": 1.97, "learning_rate": 1.9718140207178997e-05, "loss": 0.4426, "step": 16370 }, { "epoch": 1.97, "learning_rate": 1.9730185497470492e-05, "loss": 0.3599, "step": 16380 }, { "epoch": 1.97, "learning_rate": 1.9742230787761987e-05, "loss": 0.457, "step": 16390 }, { "epoch": 1.98, "learning_rate": 1.9754276078053482e-05, "loss": 0.2401, "step": 16400 }, { "epoch": 1.98, "learning_rate": 1.9766321368344977e-05, "loss": 0.3895, "step": 16410 }, { "epoch": 1.98, "learning_rate": 1.9778366658636473e-05, "loss": 0.2891, "step": 16420 }, { "epoch": 1.98, "learning_rate": 1.979041194892797e-05, "loss": 0.3111, "step": 16430 }, { "epoch": 1.98, "learning_rate": 1.9802457239219466e-05, "loss": 0.2576, "step": 16440 }, { "epoch": 1.98, "learning_rate": 1.981450252951096e-05, "loss": 0.2424, "step": 16450 }, { "epoch": 1.98, "learning_rate": 1.9826547819802457e-05, "loss": 0.5225, "step": 16460 }, { "epoch": 1.98, "learning_rate": 1.9838593110093952e-05, "loss": 0.3513, "step": 16470 }, { "epoch": 1.98, "learning_rate": 1.985063840038545e-05, "loss": 0.1253, "step": 16480 }, { "epoch": 1.99, "learning_rate": 1.986268369067695e-05, "loss": 0.3113, "step": 16490 }, { "epoch": 1.99, "learning_rate": 1.9874728980968444e-05, "loss": 0.2435, "step": 16500 }, { "epoch": 1.99, "learning_rate": 1.988677427125994e-05, "loss": 0.2349, "step": 16510 }, { "epoch": 1.99, "learning_rate": 1.9898819561551435e-05, "loss": 0.3336, "step": 16520 }, { "epoch": 1.99, "learning_rate": 1.991086485184293e-05, "loss": 0.2742, "step": 16530 }, { "epoch": 1.99, "learning_rate": 1.9922910142134425e-05, "loss": 0.3472, "step": 16540 }, { "epoch": 1.99, "learning_rate": 1.9934955432425924e-05, "loss": 0.31, "step": 16550 }, { "epoch": 1.99, "learning_rate": 1.994700072271742e-05, "loss": 0.3302, "step": 16560 }, { "epoch": 2.0, "learning_rate": 1.9959046013008914e-05, "loss": 0.2803, "step": 16570 }, { "epoch": 2.0, "learning_rate": 1.997109130330041e-05, "loss": 0.1179, "step": 16580 }, { "epoch": 2.0, "learning_rate": 1.9983136593591904e-05, "loss": 0.4523, "step": 16590 }, { "epoch": 2.0, "learning_rate": 1.9995181883883403e-05, "loss": 0.2375, "step": 16600 }, { "epoch": 2.0, "eval_accuracy": { "accuracy": 0.9053297199638664 }, "eval_f1": { "f1": 0.9062005892826234 }, "eval_loss": 0.33534759283065796, "eval_precision": { "precision": 0.912717242831991 }, "eval_recall": { "recall": 0.9052684275828231 }, "eval_runtime": 649.6653, "eval_samples_per_second": 25.559, "eval_steps_per_second": 12.78, "step": 16605 } ], "logging_steps": 10, "max_steps": 415100, "num_input_tokens_seen": 0, "num_train_epochs": 50, "save_steps": 500, "total_flos": 1.029415769881919e+19, "train_batch_size": 2, "trial_name": null, "trial_params": null }