{ "best_metric": 0.25953179597854614, "best_model_checkpoint": "vit-base-patch16-224-in21k-bridgedefectVIT\\checkpoint-41512", "epoch": 4.999939777175549, "eval_steps": 500, "global_step": 41512, "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 }, { "epoch": 2.0, "learning_rate": 2.0007227174174898e-05, "loss": 0.2914, "step": 16610 }, { "epoch": 2.0, "learning_rate": 2.0019272464466397e-05, "loss": 0.305, "step": 16620 }, { "epoch": 2.0, "learning_rate": 2.0031317754757892e-05, "loss": 0.2866, "step": 16630 }, { "epoch": 2.0, "learning_rate": 2.0043363045049387e-05, "loss": 0.3344, "step": 16640 }, { "epoch": 2.01, "learning_rate": 2.0055408335340882e-05, "loss": 0.1779, "step": 16650 }, { "epoch": 2.01, "learning_rate": 2.006745362563238e-05, "loss": 0.3102, "step": 16660 }, { "epoch": 2.01, "learning_rate": 2.0079498915923876e-05, "loss": 0.357, "step": 16670 }, { "epoch": 2.01, "learning_rate": 2.009154420621537e-05, "loss": 0.1414, "step": 16680 }, { "epoch": 2.01, "learning_rate": 2.0103589496506866e-05, "loss": 0.3453, "step": 16690 }, { "epoch": 2.01, "learning_rate": 2.011563478679836e-05, "loss": 0.1864, "step": 16700 }, { "epoch": 2.01, "learning_rate": 2.0127680077089857e-05, "loss": 0.2298, "step": 16710 }, { "epoch": 2.01, "learning_rate": 2.0139725367381355e-05, "loss": 0.2703, "step": 16720 }, { "epoch": 2.02, "learning_rate": 2.015177065767285e-05, "loss": 0.3941, "step": 16730 }, { "epoch": 2.02, "learning_rate": 2.0163815947964346e-05, "loss": 0.0865, "step": 16740 }, { "epoch": 2.02, "learning_rate": 2.0175861238255844e-05, "loss": 0.1091, "step": 16750 }, { "epoch": 2.02, "learning_rate": 2.018790652854734e-05, "loss": 0.2061, "step": 16760 }, { "epoch": 2.02, "learning_rate": 2.0199951818838835e-05, "loss": 0.4656, "step": 16770 }, { "epoch": 2.02, "learning_rate": 2.0211997109130333e-05, "loss": 0.4075, "step": 16780 }, { "epoch": 2.02, "learning_rate": 2.022404239942183e-05, "loss": 0.3266, "step": 16790 }, { "epoch": 2.02, "learning_rate": 2.0236087689713324e-05, "loss": 0.3694, "step": 16800 }, { "epoch": 2.02, "learning_rate": 2.024813298000482e-05, "loss": 0.3631, "step": 16810 }, { "epoch": 2.03, "learning_rate": 2.0260178270296314e-05, "loss": 0.2557, "step": 16820 }, { "epoch": 2.03, "learning_rate": 2.0272223560587812e-05, "loss": 0.438, "step": 16830 }, { "epoch": 2.03, "learning_rate": 2.0284268850879308e-05, "loss": 0.2812, "step": 16840 }, { "epoch": 2.03, "learning_rate": 2.0296314141170803e-05, "loss": 0.2438, "step": 16850 }, { "epoch": 2.03, "learning_rate": 2.0308359431462298e-05, "loss": 0.1862, "step": 16860 }, { "epoch": 2.03, "learning_rate": 2.0320404721753793e-05, "loss": 0.3811, "step": 16870 }, { "epoch": 2.03, "learning_rate": 2.0332450012045292e-05, "loss": 0.2505, "step": 16880 }, { "epoch": 2.03, "learning_rate": 2.0344495302336787e-05, "loss": 0.4308, "step": 16890 }, { "epoch": 2.04, "learning_rate": 2.0356540592628286e-05, "loss": 0.2882, "step": 16900 }, { "epoch": 2.04, "learning_rate": 2.036858588291978e-05, "loss": 0.3546, "step": 16910 }, { "epoch": 2.04, "learning_rate": 2.0380631173211276e-05, "loss": 0.2764, "step": 16920 }, { "epoch": 2.04, "learning_rate": 2.039267646350277e-05, "loss": 0.2269, "step": 16930 }, { "epoch": 2.04, "learning_rate": 2.0404721753794266e-05, "loss": 0.2786, "step": 16940 }, { "epoch": 2.04, "learning_rate": 2.0416767044085765e-05, "loss": 0.1561, "step": 16950 }, { "epoch": 2.04, "learning_rate": 2.042881233437726e-05, "loss": 0.215, "step": 16960 }, { "epoch": 2.04, "learning_rate": 2.0440857624668755e-05, "loss": 0.292, "step": 16970 }, { "epoch": 2.05, "learning_rate": 2.045290291496025e-05, "loss": 0.3788, "step": 16980 }, { "epoch": 2.05, "learning_rate": 2.0464948205251746e-05, "loss": 0.1888, "step": 16990 }, { "epoch": 2.05, "learning_rate": 2.047699349554324e-05, "loss": 0.2586, "step": 17000 }, { "epoch": 2.05, "learning_rate": 2.048903878583474e-05, "loss": 0.2408, "step": 17010 }, { "epoch": 2.05, "learning_rate": 2.0501084076126238e-05, "loss": 0.532, "step": 17020 }, { "epoch": 2.05, "learning_rate": 2.0513129366417733e-05, "loss": 0.2302, "step": 17030 }, { "epoch": 2.05, "learning_rate": 2.0525174656709228e-05, "loss": 0.375, "step": 17040 }, { "epoch": 2.05, "learning_rate": 2.0537219947000723e-05, "loss": 0.1673, "step": 17050 }, { "epoch": 2.05, "learning_rate": 2.054926523729222e-05, "loss": 0.3951, "step": 17060 }, { "epoch": 2.06, "learning_rate": 2.0561310527583717e-05, "loss": 0.254, "step": 17070 }, { "epoch": 2.06, "learning_rate": 2.0573355817875212e-05, "loss": 0.3323, "step": 17080 }, { "epoch": 2.06, "learning_rate": 2.0585401108166708e-05, "loss": 0.201, "step": 17090 }, { "epoch": 2.06, "learning_rate": 2.0597446398458203e-05, "loss": 0.25, "step": 17100 }, { "epoch": 2.06, "learning_rate": 2.0609491688749698e-05, "loss": 0.3145, "step": 17110 }, { "epoch": 2.06, "learning_rate": 2.0621536979041197e-05, "loss": 0.2449, "step": 17120 }, { "epoch": 2.06, "learning_rate": 2.0633582269332692e-05, "loss": 0.1466, "step": 17130 }, { "epoch": 2.06, "learning_rate": 2.0645627559624187e-05, "loss": 0.2614, "step": 17140 }, { "epoch": 2.07, "learning_rate": 2.0657672849915685e-05, "loss": 0.2366, "step": 17150 }, { "epoch": 2.07, "learning_rate": 2.066971814020718e-05, "loss": 0.333, "step": 17160 }, { "epoch": 2.07, "learning_rate": 2.0681763430498676e-05, "loss": 0.6166, "step": 17170 }, { "epoch": 2.07, "learning_rate": 2.069380872079017e-05, "loss": 0.2378, "step": 17180 }, { "epoch": 2.07, "learning_rate": 2.070585401108167e-05, "loss": 0.3643, "step": 17190 }, { "epoch": 2.07, "learning_rate": 2.0717899301373165e-05, "loss": 0.3749, "step": 17200 }, { "epoch": 2.07, "learning_rate": 2.072994459166466e-05, "loss": 0.3215, "step": 17210 }, { "epoch": 2.07, "learning_rate": 2.0741989881956155e-05, "loss": 0.249, "step": 17220 }, { "epoch": 2.08, "learning_rate": 2.075403517224765e-05, "loss": 0.2997, "step": 17230 }, { "epoch": 2.08, "learning_rate": 2.076608046253915e-05, "loss": 0.405, "step": 17240 }, { "epoch": 2.08, "learning_rate": 2.0778125752830644e-05, "loss": 0.3253, "step": 17250 }, { "epoch": 2.08, "learning_rate": 2.079017104312214e-05, "loss": 0.4503, "step": 17260 }, { "epoch": 2.08, "learning_rate": 2.0802216333413634e-05, "loss": 0.5839, "step": 17270 }, { "epoch": 2.08, "learning_rate": 2.0814261623705133e-05, "loss": 0.2655, "step": 17280 }, { "epoch": 2.08, "learning_rate": 2.0826306913996628e-05, "loss": 0.4254, "step": 17290 }, { "epoch": 2.08, "learning_rate": 2.0838352204288127e-05, "loss": 0.23, "step": 17300 }, { "epoch": 2.08, "learning_rate": 2.0850397494579622e-05, "loss": 0.1886, "step": 17310 }, { "epoch": 2.09, "learning_rate": 2.0862442784871117e-05, "loss": 0.242, "step": 17320 }, { "epoch": 2.09, "learning_rate": 2.0874488075162612e-05, "loss": 0.2815, "step": 17330 }, { "epoch": 2.09, "learning_rate": 2.0886533365454108e-05, "loss": 0.4134, "step": 17340 }, { "epoch": 2.09, "learning_rate": 2.0898578655745603e-05, "loss": 0.2259, "step": 17350 }, { "epoch": 2.09, "learning_rate": 2.09106239460371e-05, "loss": 0.3114, "step": 17360 }, { "epoch": 2.09, "learning_rate": 2.0922669236328596e-05, "loss": 0.2106, "step": 17370 }, { "epoch": 2.09, "learning_rate": 2.093471452662009e-05, "loss": 0.4336, "step": 17380 }, { "epoch": 2.09, "learning_rate": 2.0946759816911587e-05, "loss": 0.1755, "step": 17390 }, { "epoch": 2.1, "learning_rate": 2.0958805107203082e-05, "loss": 0.5467, "step": 17400 }, { "epoch": 2.1, "learning_rate": 2.097085039749458e-05, "loss": 0.3853, "step": 17410 }, { "epoch": 2.1, "learning_rate": 2.098289568778608e-05, "loss": 0.2602, "step": 17420 }, { "epoch": 2.1, "learning_rate": 2.0994940978077574e-05, "loss": 0.3967, "step": 17430 }, { "epoch": 2.1, "learning_rate": 2.100698626836907e-05, "loss": 0.2118, "step": 17440 }, { "epoch": 2.1, "learning_rate": 2.1019031558660565e-05, "loss": 0.4101, "step": 17450 }, { "epoch": 2.1, "learning_rate": 2.103107684895206e-05, "loss": 0.2633, "step": 17460 }, { "epoch": 2.1, "learning_rate": 2.1043122139243555e-05, "loss": 0.2408, "step": 17470 }, { "epoch": 2.11, "learning_rate": 2.1055167429535054e-05, "loss": 0.448, "step": 17480 }, { "epoch": 2.11, "learning_rate": 2.106721271982655e-05, "loss": 0.3208, "step": 17490 }, { "epoch": 2.11, "learning_rate": 2.1079258010118044e-05, "loss": 0.1705, "step": 17500 }, { "epoch": 2.11, "learning_rate": 2.109130330040954e-05, "loss": 0.1935, "step": 17510 }, { "epoch": 2.11, "learning_rate": 2.1103348590701034e-05, "loss": 0.4944, "step": 17520 }, { "epoch": 2.11, "learning_rate": 2.1115393880992533e-05, "loss": 0.1404, "step": 17530 }, { "epoch": 2.11, "learning_rate": 2.112743917128403e-05, "loss": 0.3484, "step": 17540 }, { "epoch": 2.11, "learning_rate": 2.1139484461575527e-05, "loss": 0.2847, "step": 17550 }, { "epoch": 2.12, "learning_rate": 2.1151529751867022e-05, "loss": 0.3686, "step": 17560 }, { "epoch": 2.12, "learning_rate": 2.1163575042158517e-05, "loss": 0.2192, "step": 17570 }, { "epoch": 2.12, "learning_rate": 2.1175620332450012e-05, "loss": 0.3375, "step": 17580 }, { "epoch": 2.12, "learning_rate": 2.118766562274151e-05, "loss": 0.2324, "step": 17590 }, { "epoch": 2.12, "learning_rate": 2.1199710913033006e-05, "loss": 0.0811, "step": 17600 }, { "epoch": 2.12, "learning_rate": 2.12117562033245e-05, "loss": 0.2242, "step": 17610 }, { "epoch": 2.12, "learning_rate": 2.1223801493615996e-05, "loss": 0.1555, "step": 17620 }, { "epoch": 2.12, "learning_rate": 2.123584678390749e-05, "loss": 0.3489, "step": 17630 }, { "epoch": 2.12, "learning_rate": 2.1247892074198987e-05, "loss": 0.3034, "step": 17640 }, { "epoch": 2.13, "learning_rate": 2.1259937364490485e-05, "loss": 0.0997, "step": 17650 }, { "epoch": 2.13, "learning_rate": 2.127198265478198e-05, "loss": 0.2187, "step": 17660 }, { "epoch": 2.13, "learning_rate": 2.128402794507348e-05, "loss": 0.1508, "step": 17670 }, { "epoch": 2.13, "learning_rate": 2.1296073235364974e-05, "loss": 0.3971, "step": 17680 }, { "epoch": 2.13, "learning_rate": 2.130811852565647e-05, "loss": 0.3516, "step": 17690 }, { "epoch": 2.13, "learning_rate": 2.1320163815947965e-05, "loss": 0.2255, "step": 17700 }, { "epoch": 2.13, "learning_rate": 2.1332209106239463e-05, "loss": 0.1815, "step": 17710 }, { "epoch": 2.13, "learning_rate": 2.134425439653096e-05, "loss": 0.1485, "step": 17720 }, { "epoch": 2.14, "learning_rate": 2.1356299686822454e-05, "loss": 0.2289, "step": 17730 }, { "epoch": 2.14, "learning_rate": 2.136834497711395e-05, "loss": 0.2994, "step": 17740 }, { "epoch": 2.14, "learning_rate": 2.1380390267405444e-05, "loss": 0.4039, "step": 17750 }, { "epoch": 2.14, "learning_rate": 2.1392435557696943e-05, "loss": 0.0527, "step": 17760 }, { "epoch": 2.14, "learning_rate": 2.1404480847988438e-05, "loss": 0.1493, "step": 17770 }, { "epoch": 2.14, "learning_rate": 2.1416526138279933e-05, "loss": 0.2762, "step": 17780 }, { "epoch": 2.14, "learning_rate": 2.1428571428571428e-05, "loss": 0.45, "step": 17790 }, { "epoch": 2.14, "learning_rate": 2.1440616718862927e-05, "loss": 0.3264, "step": 17800 }, { "epoch": 2.15, "learning_rate": 2.1452662009154422e-05, "loss": 0.2049, "step": 17810 }, { "epoch": 2.15, "learning_rate": 2.1464707299445917e-05, "loss": 0.2005, "step": 17820 }, { "epoch": 2.15, "learning_rate": 2.1476752589737416e-05, "loss": 0.3575, "step": 17830 }, { "epoch": 2.15, "learning_rate": 2.148879788002891e-05, "loss": 0.3329, "step": 17840 }, { "epoch": 2.15, "learning_rate": 2.1500843170320406e-05, "loss": 0.3524, "step": 17850 }, { "epoch": 2.15, "learning_rate": 2.15128884606119e-05, "loss": 0.4009, "step": 17860 }, { "epoch": 2.15, "learning_rate": 2.1524933750903396e-05, "loss": 0.3011, "step": 17870 }, { "epoch": 2.15, "learning_rate": 2.1536979041194895e-05, "loss": 0.32, "step": 17880 }, { "epoch": 2.15, "learning_rate": 2.154902433148639e-05, "loss": 0.2219, "step": 17890 }, { "epoch": 2.16, "learning_rate": 2.1561069621777885e-05, "loss": 0.3954, "step": 17900 }, { "epoch": 2.16, "learning_rate": 2.157311491206938e-05, "loss": 0.2671, "step": 17910 }, { "epoch": 2.16, "learning_rate": 2.1585160202360876e-05, "loss": 0.2958, "step": 17920 }, { "epoch": 2.16, "learning_rate": 2.159720549265237e-05, "loss": 0.2239, "step": 17930 }, { "epoch": 2.16, "learning_rate": 2.1609250782943873e-05, "loss": 0.3014, "step": 17940 }, { "epoch": 2.16, "learning_rate": 2.1621296073235368e-05, "loss": 0.2026, "step": 17950 }, { "epoch": 2.16, "learning_rate": 2.1633341363526863e-05, "loss": 0.2436, "step": 17960 }, { "epoch": 2.16, "learning_rate": 2.164538665381836e-05, "loss": 0.3339, "step": 17970 }, { "epoch": 2.17, "learning_rate": 2.1657431944109854e-05, "loss": 0.3628, "step": 17980 }, { "epoch": 2.17, "learning_rate": 2.166947723440135e-05, "loss": 0.2508, "step": 17990 }, { "epoch": 2.17, "learning_rate": 2.1681522524692847e-05, "loss": 0.459, "step": 18000 }, { "epoch": 2.17, "learning_rate": 2.1693567814984343e-05, "loss": 0.3504, "step": 18010 }, { "epoch": 2.17, "learning_rate": 2.1705613105275838e-05, "loss": 0.2809, "step": 18020 }, { "epoch": 2.17, "learning_rate": 2.1717658395567333e-05, "loss": 0.2621, "step": 18030 }, { "epoch": 2.17, "learning_rate": 2.1729703685858828e-05, "loss": 0.28, "step": 18040 }, { "epoch": 2.17, "learning_rate": 2.1741748976150327e-05, "loss": 0.404, "step": 18050 }, { "epoch": 2.18, "learning_rate": 2.1753794266441822e-05, "loss": 0.3653, "step": 18060 }, { "epoch": 2.18, "learning_rate": 2.176583955673332e-05, "loss": 0.2114, "step": 18070 }, { "epoch": 2.18, "learning_rate": 2.1777884847024816e-05, "loss": 0.6794, "step": 18080 }, { "epoch": 2.18, "learning_rate": 2.178993013731631e-05, "loss": 0.2391, "step": 18090 }, { "epoch": 2.18, "learning_rate": 2.1801975427607806e-05, "loss": 0.3012, "step": 18100 }, { "epoch": 2.18, "learning_rate": 2.18140207178993e-05, "loss": 0.3213, "step": 18110 }, { "epoch": 2.18, "learning_rate": 2.18260660081908e-05, "loss": 0.2734, "step": 18120 }, { "epoch": 2.18, "learning_rate": 2.1838111298482295e-05, "loss": 0.0894, "step": 18130 }, { "epoch": 2.18, "learning_rate": 2.185015658877379e-05, "loss": 0.3062, "step": 18140 }, { "epoch": 2.19, "learning_rate": 2.1862201879065285e-05, "loss": 0.2842, "step": 18150 }, { "epoch": 2.19, "learning_rate": 2.187424716935678e-05, "loss": 0.086, "step": 18160 }, { "epoch": 2.19, "learning_rate": 2.188629245964828e-05, "loss": 0.353, "step": 18170 }, { "epoch": 2.19, "learning_rate": 2.1898337749939774e-05, "loss": 0.4306, "step": 18180 }, { "epoch": 2.19, "learning_rate": 2.191038304023127e-05, "loss": 0.1276, "step": 18190 }, { "epoch": 2.19, "learning_rate": 2.1922428330522768e-05, "loss": 0.2943, "step": 18200 }, { "epoch": 2.19, "learning_rate": 2.1934473620814263e-05, "loss": 0.2604, "step": 18210 }, { "epoch": 2.19, "learning_rate": 2.194651891110576e-05, "loss": 0.3189, "step": 18220 }, { "epoch": 2.2, "learning_rate": 2.1958564201397257e-05, "loss": 0.3042, "step": 18230 }, { "epoch": 2.2, "learning_rate": 2.1970609491688752e-05, "loss": 0.2671, "step": 18240 }, { "epoch": 2.2, "learning_rate": 2.1982654781980247e-05, "loss": 0.6304, "step": 18250 }, { "epoch": 2.2, "learning_rate": 2.1994700072271742e-05, "loss": 0.2382, "step": 18260 }, { "epoch": 2.2, "learning_rate": 2.2006745362563238e-05, "loss": 0.3108, "step": 18270 }, { "epoch": 2.2, "learning_rate": 2.2018790652854733e-05, "loss": 0.2383, "step": 18280 }, { "epoch": 2.2, "learning_rate": 2.203083594314623e-05, "loss": 0.2426, "step": 18290 }, { "epoch": 2.2, "learning_rate": 2.2042881233437727e-05, "loss": 0.4343, "step": 18300 }, { "epoch": 2.21, "learning_rate": 2.2054926523729222e-05, "loss": 0.3187, "step": 18310 }, { "epoch": 2.21, "learning_rate": 2.2066971814020717e-05, "loss": 0.3462, "step": 18320 }, { "epoch": 2.21, "learning_rate": 2.2079017104312216e-05, "loss": 0.2321, "step": 18330 }, { "epoch": 2.21, "learning_rate": 2.209106239460371e-05, "loss": 0.229, "step": 18340 }, { "epoch": 2.21, "learning_rate": 2.210310768489521e-05, "loss": 0.3456, "step": 18350 }, { "epoch": 2.21, "learning_rate": 2.2115152975186705e-05, "loss": 0.2307, "step": 18360 }, { "epoch": 2.21, "learning_rate": 2.21271982654782e-05, "loss": 0.3269, "step": 18370 }, { "epoch": 2.21, "learning_rate": 2.2139243555769695e-05, "loss": 0.2373, "step": 18380 }, { "epoch": 2.21, "learning_rate": 2.215128884606119e-05, "loss": 0.4131, "step": 18390 }, { "epoch": 2.22, "learning_rate": 2.216333413635269e-05, "loss": 0.3287, "step": 18400 }, { "epoch": 2.22, "learning_rate": 2.2175379426644184e-05, "loss": 0.3601, "step": 18410 }, { "epoch": 2.22, "learning_rate": 2.218742471693568e-05, "loss": 0.3011, "step": 18420 }, { "epoch": 2.22, "learning_rate": 2.2199470007227174e-05, "loss": 0.3371, "step": 18430 }, { "epoch": 2.22, "learning_rate": 2.221151529751867e-05, "loss": 0.432, "step": 18440 }, { "epoch": 2.22, "learning_rate": 2.2223560587810165e-05, "loss": 0.1302, "step": 18450 }, { "epoch": 2.22, "learning_rate": 2.2235605878101663e-05, "loss": 0.3449, "step": 18460 }, { "epoch": 2.22, "learning_rate": 2.2247651168393162e-05, "loss": 0.3238, "step": 18470 }, { "epoch": 2.23, "learning_rate": 2.2259696458684657e-05, "loss": 0.3781, "step": 18480 }, { "epoch": 2.23, "learning_rate": 2.2271741748976152e-05, "loss": 0.2215, "step": 18490 }, { "epoch": 2.23, "learning_rate": 2.2283787039267647e-05, "loss": 0.2023, "step": 18500 }, { "epoch": 2.23, "learning_rate": 2.2295832329559142e-05, "loss": 0.3182, "step": 18510 }, { "epoch": 2.23, "learning_rate": 2.230787761985064e-05, "loss": 0.2086, "step": 18520 }, { "epoch": 2.23, "learning_rate": 2.2319922910142136e-05, "loss": 0.333, "step": 18530 }, { "epoch": 2.23, "learning_rate": 2.233196820043363e-05, "loss": 0.4096, "step": 18540 }, { "epoch": 2.23, "learning_rate": 2.2344013490725127e-05, "loss": 0.2103, "step": 18550 }, { "epoch": 2.24, "learning_rate": 2.2356058781016622e-05, "loss": 0.3476, "step": 18560 }, { "epoch": 2.24, "learning_rate": 2.2368104071308117e-05, "loss": 0.5966, "step": 18570 }, { "epoch": 2.24, "learning_rate": 2.2380149361599616e-05, "loss": 0.244, "step": 18580 }, { "epoch": 2.24, "learning_rate": 2.2392194651891114e-05, "loss": 0.2959, "step": 18590 }, { "epoch": 2.24, "learning_rate": 2.240423994218261e-05, "loss": 0.3444, "step": 18600 }, { "epoch": 2.24, "learning_rate": 2.2416285232474104e-05, "loss": 0.5162, "step": 18610 }, { "epoch": 2.24, "learning_rate": 2.24283305227656e-05, "loss": 0.1157, "step": 18620 }, { "epoch": 2.24, "learning_rate": 2.2440375813057095e-05, "loss": 0.3528, "step": 18630 }, { "epoch": 2.25, "learning_rate": 2.2452421103348593e-05, "loss": 0.1366, "step": 18640 }, { "epoch": 2.25, "learning_rate": 2.246446639364009e-05, "loss": 0.2158, "step": 18650 }, { "epoch": 2.25, "learning_rate": 2.2476511683931584e-05, "loss": 0.221, "step": 18660 }, { "epoch": 2.25, "learning_rate": 2.248855697422308e-05, "loss": 0.2644, "step": 18670 }, { "epoch": 2.25, "learning_rate": 2.2500602264514574e-05, "loss": 0.2404, "step": 18680 }, { "epoch": 2.25, "learning_rate": 2.2512647554806073e-05, "loss": 0.3361, "step": 18690 }, { "epoch": 2.25, "learning_rate": 2.2524692845097568e-05, "loss": 0.2452, "step": 18700 }, { "epoch": 2.25, "learning_rate": 2.2536738135389063e-05, "loss": 0.3258, "step": 18710 }, { "epoch": 2.25, "learning_rate": 2.254878342568056e-05, "loss": 0.1848, "step": 18720 }, { "epoch": 2.26, "learning_rate": 2.2560828715972057e-05, "loss": 0.3118, "step": 18730 }, { "epoch": 2.26, "learning_rate": 2.2572874006263552e-05, "loss": 0.3791, "step": 18740 }, { "epoch": 2.26, "learning_rate": 2.2584919296555047e-05, "loss": 0.6041, "step": 18750 }, { "epoch": 2.26, "learning_rate": 2.2596964586846546e-05, "loss": 0.324, "step": 18760 }, { "epoch": 2.26, "learning_rate": 2.260900987713804e-05, "loss": 0.4027, "step": 18770 }, { "epoch": 2.26, "learning_rate": 2.2621055167429536e-05, "loss": 0.1868, "step": 18780 }, { "epoch": 2.26, "learning_rate": 2.263310045772103e-05, "loss": 0.195, "step": 18790 }, { "epoch": 2.26, "learning_rate": 2.2645145748012527e-05, "loss": 0.3552, "step": 18800 }, { "epoch": 2.27, "learning_rate": 2.2657191038304025e-05, "loss": 0.3043, "step": 18810 }, { "epoch": 2.27, "learning_rate": 2.266923632859552e-05, "loss": 0.4583, "step": 18820 }, { "epoch": 2.27, "learning_rate": 2.2681281618887015e-05, "loss": 0.2351, "step": 18830 }, { "epoch": 2.27, "learning_rate": 2.269332690917851e-05, "loss": 0.129, "step": 18840 }, { "epoch": 2.27, "learning_rate": 2.2705372199470006e-05, "loss": 0.456, "step": 18850 }, { "epoch": 2.27, "learning_rate": 2.2717417489761504e-05, "loss": 0.4243, "step": 18860 }, { "epoch": 2.27, "learning_rate": 2.2729462780053003e-05, "loss": 0.6836, "step": 18870 }, { "epoch": 2.27, "learning_rate": 2.2741508070344498e-05, "loss": 0.4416, "step": 18880 }, { "epoch": 2.28, "learning_rate": 2.2753553360635993e-05, "loss": 0.1777, "step": 18890 }, { "epoch": 2.28, "learning_rate": 2.276559865092749e-05, "loss": 0.1317, "step": 18900 }, { "epoch": 2.28, "learning_rate": 2.2777643941218984e-05, "loss": 0.4661, "step": 18910 }, { "epoch": 2.28, "learning_rate": 2.278968923151048e-05, "loss": 0.3253, "step": 18920 }, { "epoch": 2.28, "learning_rate": 2.2801734521801977e-05, "loss": 0.2412, "step": 18930 }, { "epoch": 2.28, "learning_rate": 2.2813779812093473e-05, "loss": 0.4389, "step": 18940 }, { "epoch": 2.28, "learning_rate": 2.2825825102384968e-05, "loss": 0.1474, "step": 18950 }, { "epoch": 2.28, "learning_rate": 2.2837870392676463e-05, "loss": 0.2352, "step": 18960 }, { "epoch": 2.28, "learning_rate": 2.2849915682967958e-05, "loss": 0.3675, "step": 18970 }, { "epoch": 2.29, "learning_rate": 2.2861960973259457e-05, "loss": 0.1691, "step": 18980 }, { "epoch": 2.29, "learning_rate": 2.2874006263550955e-05, "loss": 0.2047, "step": 18990 }, { "epoch": 2.29, "learning_rate": 2.288605155384245e-05, "loss": 0.2093, "step": 19000 }, { "epoch": 2.29, "learning_rate": 2.2898096844133946e-05, "loss": 0.1675, "step": 19010 }, { "epoch": 2.29, "learning_rate": 2.291014213442544e-05, "loss": 0.1145, "step": 19020 }, { "epoch": 2.29, "learning_rate": 2.2922187424716936e-05, "loss": 0.2146, "step": 19030 }, { "epoch": 2.29, "learning_rate": 2.293423271500843e-05, "loss": 0.2248, "step": 19040 }, { "epoch": 2.29, "learning_rate": 2.294627800529993e-05, "loss": 0.4549, "step": 19050 }, { "epoch": 2.3, "learning_rate": 2.2958323295591425e-05, "loss": 0.3018, "step": 19060 }, { "epoch": 2.3, "learning_rate": 2.297036858588292e-05, "loss": 0.2102, "step": 19070 }, { "epoch": 2.3, "learning_rate": 2.2982413876174415e-05, "loss": 0.3029, "step": 19080 }, { "epoch": 2.3, "learning_rate": 2.299445916646591e-05, "loss": 0.1932, "step": 19090 }, { "epoch": 2.3, "learning_rate": 2.300650445675741e-05, "loss": 0.3838, "step": 19100 }, { "epoch": 2.3, "learning_rate": 2.3018549747048904e-05, "loss": 0.2002, "step": 19110 }, { "epoch": 2.3, "learning_rate": 2.3030595037340403e-05, "loss": 0.4463, "step": 19120 }, { "epoch": 2.3, "learning_rate": 2.3042640327631898e-05, "loss": 0.2369, "step": 19130 }, { "epoch": 2.31, "learning_rate": 2.3054685617923393e-05, "loss": 0.2805, "step": 19140 }, { "epoch": 2.31, "learning_rate": 2.306673090821489e-05, "loss": 0.2233, "step": 19150 }, { "epoch": 2.31, "learning_rate": 2.3078776198506387e-05, "loss": 0.3876, "step": 19160 }, { "epoch": 2.31, "learning_rate": 2.3090821488797882e-05, "loss": 0.2707, "step": 19170 }, { "epoch": 2.31, "learning_rate": 2.3102866779089377e-05, "loss": 0.4088, "step": 19180 }, { "epoch": 2.31, "learning_rate": 2.3114912069380873e-05, "loss": 0.2962, "step": 19190 }, { "epoch": 2.31, "learning_rate": 2.3126957359672368e-05, "loss": 0.2919, "step": 19200 }, { "epoch": 2.31, "learning_rate": 2.3139002649963863e-05, "loss": 0.2467, "step": 19210 }, { "epoch": 2.31, "learning_rate": 2.315104794025536e-05, "loss": 0.2536, "step": 19220 }, { "epoch": 2.32, "learning_rate": 2.3163093230546857e-05, "loss": 0.1904, "step": 19230 }, { "epoch": 2.32, "learning_rate": 2.3175138520838352e-05, "loss": 0.2516, "step": 19240 }, { "epoch": 2.32, "learning_rate": 2.318718381112985e-05, "loss": 0.3618, "step": 19250 }, { "epoch": 2.32, "learning_rate": 2.3199229101421346e-05, "loss": 0.1558, "step": 19260 }, { "epoch": 2.32, "learning_rate": 2.321127439171284e-05, "loss": 0.2178, "step": 19270 }, { "epoch": 2.32, "learning_rate": 2.322331968200434e-05, "loss": 0.1738, "step": 19280 }, { "epoch": 2.32, "learning_rate": 2.3235364972295835e-05, "loss": 0.3009, "step": 19290 }, { "epoch": 2.32, "learning_rate": 2.324741026258733e-05, "loss": 0.2911, "step": 19300 }, { "epoch": 2.33, "learning_rate": 2.3259455552878825e-05, "loss": 0.2195, "step": 19310 }, { "epoch": 2.33, "learning_rate": 2.327150084317032e-05, "loss": 0.309, "step": 19320 }, { "epoch": 2.33, "learning_rate": 2.328354613346182e-05, "loss": 0.296, "step": 19330 }, { "epoch": 2.33, "learning_rate": 2.3295591423753314e-05, "loss": 0.4469, "step": 19340 }, { "epoch": 2.33, "learning_rate": 2.330763671404481e-05, "loss": 0.3804, "step": 19350 }, { "epoch": 2.33, "learning_rate": 2.3319682004336304e-05, "loss": 0.3234, "step": 19360 }, { "epoch": 2.33, "learning_rate": 2.33317272946278e-05, "loss": 0.1655, "step": 19370 }, { "epoch": 2.33, "learning_rate": 2.3343772584919298e-05, "loss": 0.201, "step": 19380 }, { "epoch": 2.34, "learning_rate": 2.3355817875210793e-05, "loss": 0.3988, "step": 19390 }, { "epoch": 2.34, "learning_rate": 2.3367863165502292e-05, "loss": 0.2351, "step": 19400 }, { "epoch": 2.34, "learning_rate": 2.3379908455793787e-05, "loss": 0.177, "step": 19410 }, { "epoch": 2.34, "learning_rate": 2.3391953746085282e-05, "loss": 0.3211, "step": 19420 }, { "epoch": 2.34, "learning_rate": 2.3403999036376777e-05, "loss": 0.4941, "step": 19430 }, { "epoch": 2.34, "learning_rate": 2.3416044326668273e-05, "loss": 0.2463, "step": 19440 }, { "epoch": 2.34, "learning_rate": 2.342808961695977e-05, "loss": 0.3675, "step": 19450 }, { "epoch": 2.34, "learning_rate": 2.3440134907251266e-05, "loss": 0.1161, "step": 19460 }, { "epoch": 2.35, "learning_rate": 2.345218019754276e-05, "loss": 0.2396, "step": 19470 }, { "epoch": 2.35, "learning_rate": 2.3464225487834257e-05, "loss": 0.2403, "step": 19480 }, { "epoch": 2.35, "learning_rate": 2.3476270778125752e-05, "loss": 0.3182, "step": 19490 }, { "epoch": 2.35, "learning_rate": 2.3488316068417247e-05, "loss": 0.4083, "step": 19500 }, { "epoch": 2.35, "learning_rate": 2.3500361358708746e-05, "loss": 0.2613, "step": 19510 }, { "epoch": 2.35, "learning_rate": 2.3512406649000244e-05, "loss": 0.2057, "step": 19520 }, { "epoch": 2.35, "learning_rate": 2.352445193929174e-05, "loss": 0.3623, "step": 19530 }, { "epoch": 2.35, "learning_rate": 2.3536497229583235e-05, "loss": 0.3913, "step": 19540 }, { "epoch": 2.35, "learning_rate": 2.354854251987473e-05, "loss": 0.2372, "step": 19550 }, { "epoch": 2.36, "learning_rate": 2.3560587810166225e-05, "loss": 0.427, "step": 19560 }, { "epoch": 2.36, "learning_rate": 2.3572633100457724e-05, "loss": 0.5092, "step": 19570 }, { "epoch": 2.36, "learning_rate": 2.358467839074922e-05, "loss": 0.4897, "step": 19580 }, { "epoch": 2.36, "learning_rate": 2.3596723681040714e-05, "loss": 0.449, "step": 19590 }, { "epoch": 2.36, "learning_rate": 2.360876897133221e-05, "loss": 0.2216, "step": 19600 }, { "epoch": 2.36, "learning_rate": 2.3620814261623704e-05, "loss": 0.2319, "step": 19610 }, { "epoch": 2.36, "learning_rate": 2.3632859551915203e-05, "loss": 0.3157, "step": 19620 }, { "epoch": 2.36, "learning_rate": 2.3644904842206698e-05, "loss": 0.2008, "step": 19630 }, { "epoch": 2.37, "learning_rate": 2.3656950132498197e-05, "loss": 0.2138, "step": 19640 }, { "epoch": 2.37, "learning_rate": 2.3668995422789692e-05, "loss": 0.2873, "step": 19650 }, { "epoch": 2.37, "learning_rate": 2.3681040713081187e-05, "loss": 0.1801, "step": 19660 }, { "epoch": 2.37, "learning_rate": 2.3693086003372682e-05, "loss": 0.116, "step": 19670 }, { "epoch": 2.37, "learning_rate": 2.3705131293664177e-05, "loss": 0.4723, "step": 19680 }, { "epoch": 2.37, "learning_rate": 2.3717176583955676e-05, "loss": 0.4795, "step": 19690 }, { "epoch": 2.37, "learning_rate": 2.372922187424717e-05, "loss": 0.2466, "step": 19700 }, { "epoch": 2.37, "learning_rate": 2.3741267164538666e-05, "loss": 0.3965, "step": 19710 }, { "epoch": 2.38, "learning_rate": 2.375331245483016e-05, "loss": 0.1324, "step": 19720 }, { "epoch": 2.38, "learning_rate": 2.3765357745121657e-05, "loss": 0.1808, "step": 19730 }, { "epoch": 2.38, "learning_rate": 2.3777403035413155e-05, "loss": 0.277, "step": 19740 }, { "epoch": 2.38, "learning_rate": 2.378944832570465e-05, "loss": 0.3511, "step": 19750 }, { "epoch": 2.38, "learning_rate": 2.3801493615996146e-05, "loss": 0.1964, "step": 19760 }, { "epoch": 2.38, "learning_rate": 2.3813538906287644e-05, "loss": 0.1887, "step": 19770 }, { "epoch": 2.38, "learning_rate": 2.382558419657914e-05, "loss": 0.4115, "step": 19780 }, { "epoch": 2.38, "learning_rate": 2.3837629486870635e-05, "loss": 0.2176, "step": 19790 }, { "epoch": 2.38, "learning_rate": 2.3849674777162133e-05, "loss": 0.3131, "step": 19800 }, { "epoch": 2.39, "learning_rate": 2.3861720067453628e-05, "loss": 0.2453, "step": 19810 }, { "epoch": 2.39, "learning_rate": 2.3873765357745123e-05, "loss": 0.3278, "step": 19820 }, { "epoch": 2.39, "learning_rate": 2.388581064803662e-05, "loss": 0.1994, "step": 19830 }, { "epoch": 2.39, "learning_rate": 2.3897855938328114e-05, "loss": 0.3938, "step": 19840 }, { "epoch": 2.39, "learning_rate": 2.390990122861961e-05, "loss": 0.5587, "step": 19850 }, { "epoch": 2.39, "learning_rate": 2.3921946518911108e-05, "loss": 0.2673, "step": 19860 }, { "epoch": 2.39, "learning_rate": 2.3933991809202603e-05, "loss": 0.1382, "step": 19870 }, { "epoch": 2.39, "learning_rate": 2.3946037099494098e-05, "loss": 0.2567, "step": 19880 }, { "epoch": 2.4, "learning_rate": 2.3958082389785593e-05, "loss": 0.2311, "step": 19890 }, { "epoch": 2.4, "learning_rate": 2.397012768007709e-05, "loss": 0.1598, "step": 19900 }, { "epoch": 2.4, "learning_rate": 2.3982172970368587e-05, "loss": 0.2893, "step": 19910 }, { "epoch": 2.4, "learning_rate": 2.3994218260660085e-05, "loss": 0.3223, "step": 19920 }, { "epoch": 2.4, "learning_rate": 2.400626355095158e-05, "loss": 0.4007, "step": 19930 }, { "epoch": 2.4, "learning_rate": 2.4018308841243076e-05, "loss": 0.3541, "step": 19940 }, { "epoch": 2.4, "learning_rate": 2.403035413153457e-05, "loss": 0.5569, "step": 19950 }, { "epoch": 2.4, "learning_rate": 2.4042399421826066e-05, "loss": 0.1709, "step": 19960 }, { "epoch": 2.41, "learning_rate": 2.4054444712117565e-05, "loss": 0.3553, "step": 19970 }, { "epoch": 2.41, "learning_rate": 2.406649000240906e-05, "loss": 0.419, "step": 19980 }, { "epoch": 2.41, "learning_rate": 2.4078535292700555e-05, "loss": 0.223, "step": 19990 }, { "epoch": 2.41, "learning_rate": 2.409058058299205e-05, "loss": 0.2623, "step": 20000 }, { "epoch": 2.41, "learning_rate": 2.4102625873283546e-05, "loss": 0.2627, "step": 20010 }, { "epoch": 2.41, "learning_rate": 2.411467116357504e-05, "loss": 0.3253, "step": 20020 }, { "epoch": 2.41, "learning_rate": 2.412671645386654e-05, "loss": 0.2556, "step": 20030 }, { "epoch": 2.41, "learning_rate": 2.4138761744158038e-05, "loss": 0.0949, "step": 20040 }, { "epoch": 2.41, "learning_rate": 2.4150807034449533e-05, "loss": 0.2357, "step": 20050 }, { "epoch": 2.42, "learning_rate": 2.4162852324741028e-05, "loss": 0.3031, "step": 20060 }, { "epoch": 2.42, "learning_rate": 2.4174897615032523e-05, "loss": 0.3499, "step": 20070 }, { "epoch": 2.42, "learning_rate": 2.418694290532402e-05, "loss": 0.2368, "step": 20080 }, { "epoch": 2.42, "learning_rate": 2.4198988195615517e-05, "loss": 0.2478, "step": 20090 }, { "epoch": 2.42, "learning_rate": 2.4211033485907012e-05, "loss": 0.2183, "step": 20100 }, { "epoch": 2.42, "learning_rate": 2.4223078776198508e-05, "loss": 0.2982, "step": 20110 }, { "epoch": 2.42, "learning_rate": 2.4235124066490003e-05, "loss": 0.3106, "step": 20120 }, { "epoch": 2.42, "learning_rate": 2.4247169356781498e-05, "loss": 0.2505, "step": 20130 }, { "epoch": 2.43, "learning_rate": 2.4259214647072993e-05, "loss": 0.4763, "step": 20140 }, { "epoch": 2.43, "learning_rate": 2.427125993736449e-05, "loss": 0.3278, "step": 20150 }, { "epoch": 2.43, "learning_rate": 2.4283305227655987e-05, "loss": 0.244, "step": 20160 }, { "epoch": 2.43, "learning_rate": 2.4295350517947485e-05, "loss": 0.4553, "step": 20170 }, { "epoch": 2.43, "learning_rate": 2.430739580823898e-05, "loss": 0.4193, "step": 20180 }, { "epoch": 2.43, "learning_rate": 2.4319441098530476e-05, "loss": 0.2976, "step": 20190 }, { "epoch": 2.43, "learning_rate": 2.433148638882197e-05, "loss": 0.5022, "step": 20200 }, { "epoch": 2.43, "learning_rate": 2.434353167911347e-05, "loss": 0.3763, "step": 20210 }, { "epoch": 2.44, "learning_rate": 2.4355576969404965e-05, "loss": 0.2371, "step": 20220 }, { "epoch": 2.44, "learning_rate": 2.436762225969646e-05, "loss": 0.3035, "step": 20230 }, { "epoch": 2.44, "learning_rate": 2.4379667549987955e-05, "loss": 0.2506, "step": 20240 }, { "epoch": 2.44, "learning_rate": 2.439171284027945e-05, "loss": 0.1691, "step": 20250 }, { "epoch": 2.44, "learning_rate": 2.440375813057095e-05, "loss": 0.2119, "step": 20260 }, { "epoch": 2.44, "learning_rate": 2.4415803420862444e-05, "loss": 0.2203, "step": 20270 }, { "epoch": 2.44, "learning_rate": 2.442784871115394e-05, "loss": 0.2862, "step": 20280 }, { "epoch": 2.44, "learning_rate": 2.4439894001445434e-05, "loss": 0.202, "step": 20290 }, { "epoch": 2.45, "learning_rate": 2.4451939291736933e-05, "loss": 0.512, "step": 20300 }, { "epoch": 2.45, "learning_rate": 2.4463984582028428e-05, "loss": 0.4032, "step": 20310 }, { "epoch": 2.45, "learning_rate": 2.4476029872319923e-05, "loss": 0.4375, "step": 20320 }, { "epoch": 2.45, "learning_rate": 2.4488075162611422e-05, "loss": 0.4582, "step": 20330 }, { "epoch": 2.45, "learning_rate": 2.4500120452902917e-05, "loss": 0.3347, "step": 20340 }, { "epoch": 2.45, "learning_rate": 2.4512165743194412e-05, "loss": 0.2841, "step": 20350 }, { "epoch": 2.45, "learning_rate": 2.4524211033485908e-05, "loss": 0.4269, "step": 20360 }, { "epoch": 2.45, "learning_rate": 2.4536256323777403e-05, "loss": 0.2332, "step": 20370 }, { "epoch": 2.45, "learning_rate": 2.45483016140689e-05, "loss": 0.4699, "step": 20380 }, { "epoch": 2.46, "learning_rate": 2.4560346904360396e-05, "loss": 0.2231, "step": 20390 }, { "epoch": 2.46, "learning_rate": 2.457239219465189e-05, "loss": 0.3149, "step": 20400 }, { "epoch": 2.46, "learning_rate": 2.4584437484943387e-05, "loss": 0.2881, "step": 20410 }, { "epoch": 2.46, "learning_rate": 2.4596482775234882e-05, "loss": 0.4093, "step": 20420 }, { "epoch": 2.46, "learning_rate": 2.460852806552638e-05, "loss": 0.248, "step": 20430 }, { "epoch": 2.46, "learning_rate": 2.462057335581788e-05, "loss": 0.3611, "step": 20440 }, { "epoch": 2.46, "learning_rate": 2.4632618646109374e-05, "loss": 0.2147, "step": 20450 }, { "epoch": 2.46, "learning_rate": 2.464466393640087e-05, "loss": 0.3283, "step": 20460 }, { "epoch": 2.47, "learning_rate": 2.4656709226692365e-05, "loss": 0.0916, "step": 20470 }, { "epoch": 2.47, "learning_rate": 2.466875451698386e-05, "loss": 0.1911, "step": 20480 }, { "epoch": 2.47, "learning_rate": 2.4680799807275355e-05, "loss": 0.3472, "step": 20490 }, { "epoch": 2.47, "learning_rate": 2.4692845097566854e-05, "loss": 0.2637, "step": 20500 }, { "epoch": 2.47, "learning_rate": 2.470489038785835e-05, "loss": 0.1025, "step": 20510 }, { "epoch": 2.47, "learning_rate": 2.4716935678149844e-05, "loss": 0.377, "step": 20520 }, { "epoch": 2.47, "learning_rate": 2.472898096844134e-05, "loss": 0.3237, "step": 20530 }, { "epoch": 2.47, "learning_rate": 2.4741026258732834e-05, "loss": 0.2445, "step": 20540 }, { "epoch": 2.48, "learning_rate": 2.4753071549024333e-05, "loss": 0.3494, "step": 20550 }, { "epoch": 2.48, "learning_rate": 2.476511683931583e-05, "loss": 0.2109, "step": 20560 }, { "epoch": 2.48, "learning_rate": 2.4777162129607327e-05, "loss": 0.2756, "step": 20570 }, { "epoch": 2.48, "learning_rate": 2.4789207419898822e-05, "loss": 0.1373, "step": 20580 }, { "epoch": 2.48, "learning_rate": 2.4801252710190317e-05, "loss": 0.2192, "step": 20590 }, { "epoch": 2.48, "learning_rate": 2.4813298000481812e-05, "loss": 0.1628, "step": 20600 }, { "epoch": 2.48, "learning_rate": 2.4825343290773307e-05, "loss": 0.1605, "step": 20610 }, { "epoch": 2.48, "learning_rate": 2.4837388581064806e-05, "loss": 0.6493, "step": 20620 }, { "epoch": 2.48, "learning_rate": 2.48494338713563e-05, "loss": 0.1737, "step": 20630 }, { "epoch": 2.49, "learning_rate": 2.4861479161647796e-05, "loss": 0.068, "step": 20640 }, { "epoch": 2.49, "learning_rate": 2.487352445193929e-05, "loss": 0.0615, "step": 20650 }, { "epoch": 2.49, "learning_rate": 2.4885569742230787e-05, "loss": 0.2926, "step": 20660 }, { "epoch": 2.49, "learning_rate": 2.4897615032522285e-05, "loss": 0.5142, "step": 20670 }, { "epoch": 2.49, "learning_rate": 2.490966032281378e-05, "loss": 0.5157, "step": 20680 }, { "epoch": 2.49, "learning_rate": 2.492170561310528e-05, "loss": 0.2207, "step": 20690 }, { "epoch": 2.49, "learning_rate": 2.4933750903396774e-05, "loss": 0.259, "step": 20700 }, { "epoch": 2.49, "learning_rate": 2.494579619368827e-05, "loss": 0.2244, "step": 20710 }, { "epoch": 2.5, "learning_rate": 2.4957841483979765e-05, "loss": 0.2879, "step": 20720 }, { "epoch": 2.5, "learning_rate": 2.4969886774271263e-05, "loss": 0.2618, "step": 20730 }, { "epoch": 2.5, "learning_rate": 2.498193206456276e-05, "loss": 0.1757, "step": 20740 }, { "epoch": 2.5, "learning_rate": 2.4993977354854254e-05, "loss": 0.2613, "step": 20750 }, { "epoch": 2.5, "learning_rate": 2.500602264514575e-05, "loss": 0.2259, "step": 20760 }, { "epoch": 2.5, "learning_rate": 2.5018067935437244e-05, "loss": 0.1728, "step": 20770 }, { "epoch": 2.5, "learning_rate": 2.503011322572874e-05, "loss": 0.2653, "step": 20780 }, { "epoch": 2.5, "learning_rate": 2.5042158516020238e-05, "loss": 0.2818, "step": 20790 }, { "epoch": 2.51, "learning_rate": 2.5054203806311733e-05, "loss": 0.4594, "step": 20800 }, { "epoch": 2.51, "learning_rate": 2.5066249096603228e-05, "loss": 0.2846, "step": 20810 }, { "epoch": 2.51, "learning_rate": 2.5078294386894723e-05, "loss": 0.2171, "step": 20820 }, { "epoch": 2.51, "learning_rate": 2.509033967718622e-05, "loss": 0.3856, "step": 20830 }, { "epoch": 2.51, "learning_rate": 2.5102384967477717e-05, "loss": 0.2612, "step": 20840 }, { "epoch": 2.51, "learning_rate": 2.5114430257769212e-05, "loss": 0.4043, "step": 20850 }, { "epoch": 2.51, "learning_rate": 2.5126475548060707e-05, "loss": 0.4001, "step": 20860 }, { "epoch": 2.51, "learning_rate": 2.5138520838352203e-05, "loss": 0.1688, "step": 20870 }, { "epoch": 2.51, "learning_rate": 2.5150566128643698e-05, "loss": 0.2486, "step": 20880 }, { "epoch": 2.52, "learning_rate": 2.51626114189352e-05, "loss": 0.4157, "step": 20890 }, { "epoch": 2.52, "learning_rate": 2.5174656709226695e-05, "loss": 0.1169, "step": 20900 }, { "epoch": 2.52, "learning_rate": 2.5186701999518194e-05, "loss": 0.2877, "step": 20910 }, { "epoch": 2.52, "learning_rate": 2.519874728980969e-05, "loss": 0.3148, "step": 20920 }, { "epoch": 2.52, "learning_rate": 2.5210792580101184e-05, "loss": 0.3448, "step": 20930 }, { "epoch": 2.52, "learning_rate": 2.522283787039268e-05, "loss": 0.4345, "step": 20940 }, { "epoch": 2.52, "learning_rate": 2.5234883160684174e-05, "loss": 0.3308, "step": 20950 }, { "epoch": 2.52, "learning_rate": 2.524692845097567e-05, "loss": 0.3733, "step": 20960 }, { "epoch": 2.53, "learning_rate": 2.5258973741267168e-05, "loss": 0.2372, "step": 20970 }, { "epoch": 2.53, "learning_rate": 2.5271019031558663e-05, "loss": 0.3075, "step": 20980 }, { "epoch": 2.53, "learning_rate": 2.528306432185016e-05, "loss": 0.3181, "step": 20990 }, { "epoch": 2.53, "learning_rate": 2.5295109612141654e-05, "loss": 0.2813, "step": 21000 }, { "epoch": 2.53, "learning_rate": 2.530715490243315e-05, "loss": 0.5037, "step": 21010 }, { "epoch": 2.53, "learning_rate": 2.5319200192724647e-05, "loss": 0.1678, "step": 21020 }, { "epoch": 2.53, "learning_rate": 2.5331245483016143e-05, "loss": 0.2546, "step": 21030 }, { "epoch": 2.53, "learning_rate": 2.5343290773307638e-05, "loss": 0.258, "step": 21040 }, { "epoch": 2.54, "learning_rate": 2.5355336063599133e-05, "loss": 0.4372, "step": 21050 }, { "epoch": 2.54, "learning_rate": 2.5367381353890628e-05, "loss": 0.1636, "step": 21060 }, { "epoch": 2.54, "learning_rate": 2.5379426644182123e-05, "loss": 0.2584, "step": 21070 }, { "epoch": 2.54, "learning_rate": 2.5391471934473622e-05, "loss": 0.1621, "step": 21080 }, { "epoch": 2.54, "learning_rate": 2.5403517224765117e-05, "loss": 0.2406, "step": 21090 }, { "epoch": 2.54, "learning_rate": 2.5415562515056612e-05, "loss": 0.2137, "step": 21100 }, { "epoch": 2.54, "learning_rate": 2.5427607805348107e-05, "loss": 0.3372, "step": 21110 }, { "epoch": 2.54, "learning_rate": 2.5439653095639603e-05, "loss": 0.1659, "step": 21120 }, { "epoch": 2.55, "learning_rate": 2.54516983859311e-05, "loss": 0.2762, "step": 21130 }, { "epoch": 2.55, "learning_rate": 2.5463743676222596e-05, "loss": 0.309, "step": 21140 }, { "epoch": 2.55, "learning_rate": 2.5475788966514098e-05, "loss": 0.2992, "step": 21150 }, { "epoch": 2.55, "learning_rate": 2.5487834256805593e-05, "loss": 0.3256, "step": 21160 }, { "epoch": 2.55, "learning_rate": 2.549987954709709e-05, "loss": 0.2369, "step": 21170 }, { "epoch": 2.55, "learning_rate": 2.5511924837388584e-05, "loss": 0.3958, "step": 21180 }, { "epoch": 2.55, "learning_rate": 2.552397012768008e-05, "loss": 0.3664, "step": 21190 }, { "epoch": 2.55, "learning_rate": 2.5536015417971578e-05, "loss": 0.2497, "step": 21200 }, { "epoch": 2.55, "learning_rate": 2.5548060708263073e-05, "loss": 0.1988, "step": 21210 }, { "epoch": 2.56, "learning_rate": 2.5560105998554568e-05, "loss": 0.2674, "step": 21220 }, { "epoch": 2.56, "learning_rate": 2.5572151288846063e-05, "loss": 0.3218, "step": 21230 }, { "epoch": 2.56, "learning_rate": 2.558419657913756e-05, "loss": 0.5888, "step": 21240 }, { "epoch": 2.56, "learning_rate": 2.5596241869429054e-05, "loss": 0.4599, "step": 21250 }, { "epoch": 2.56, "learning_rate": 2.5608287159720552e-05, "loss": 0.2878, "step": 21260 }, { "epoch": 2.56, "learning_rate": 2.5620332450012047e-05, "loss": 0.207, "step": 21270 }, { "epoch": 2.56, "learning_rate": 2.5632377740303542e-05, "loss": 0.2566, "step": 21280 }, { "epoch": 2.56, "learning_rate": 2.5644423030595038e-05, "loss": 0.3024, "step": 21290 }, { "epoch": 2.57, "learning_rate": 2.5656468320886533e-05, "loss": 0.4768, "step": 21300 }, { "epoch": 2.57, "learning_rate": 2.566851361117803e-05, "loss": 0.1432, "step": 21310 }, { "epoch": 2.57, "learning_rate": 2.5680558901469527e-05, "loss": 0.3281, "step": 21320 }, { "epoch": 2.57, "learning_rate": 2.5692604191761022e-05, "loss": 0.3313, "step": 21330 }, { "epoch": 2.57, "learning_rate": 2.5704649482052517e-05, "loss": 0.2419, "step": 21340 }, { "epoch": 2.57, "learning_rate": 2.5716694772344012e-05, "loss": 0.4625, "step": 21350 }, { "epoch": 2.57, "learning_rate": 2.572874006263551e-05, "loss": 0.2497, "step": 21360 }, { "epoch": 2.57, "learning_rate": 2.5740785352927006e-05, "loss": 0.2966, "step": 21370 }, { "epoch": 2.58, "learning_rate": 2.57528306432185e-05, "loss": 0.4213, "step": 21380 }, { "epoch": 2.58, "learning_rate": 2.5764875933509996e-05, "loss": 0.0535, "step": 21390 }, { "epoch": 2.58, "learning_rate": 2.577692122380149e-05, "loss": 0.3164, "step": 21400 }, { "epoch": 2.58, "learning_rate": 2.5788966514092987e-05, "loss": 0.3348, "step": 21410 }, { "epoch": 2.58, "learning_rate": 2.580101180438449e-05, "loss": 0.3176, "step": 21420 }, { "epoch": 2.58, "learning_rate": 2.5813057094675984e-05, "loss": 0.254, "step": 21430 }, { "epoch": 2.58, "learning_rate": 2.5825102384967482e-05, "loss": 0.2852, "step": 21440 }, { "epoch": 2.58, "learning_rate": 2.5837147675258978e-05, "loss": 0.4033, "step": 21450 }, { "epoch": 2.58, "learning_rate": 2.5849192965550473e-05, "loss": 0.409, "step": 21460 }, { "epoch": 2.59, "learning_rate": 2.5861238255841968e-05, "loss": 0.7264, "step": 21470 }, { "epoch": 2.59, "learning_rate": 2.5873283546133463e-05, "loss": 0.354, "step": 21480 }, { "epoch": 2.59, "learning_rate": 2.588532883642496e-05, "loss": 0.2336, "step": 21490 }, { "epoch": 2.59, "learning_rate": 2.5897374126716457e-05, "loss": 0.1809, "step": 21500 }, { "epoch": 2.59, "learning_rate": 2.5909419417007952e-05, "loss": 0.2313, "step": 21510 }, { "epoch": 2.59, "learning_rate": 2.5921464707299447e-05, "loss": 0.3703, "step": 21520 }, { "epoch": 2.59, "learning_rate": 2.5933509997590942e-05, "loss": 0.2312, "step": 21530 }, { "epoch": 2.59, "learning_rate": 2.5945555287882438e-05, "loss": 0.1573, "step": 21540 }, { "epoch": 2.6, "learning_rate": 2.5957600578173936e-05, "loss": 0.2092, "step": 21550 }, { "epoch": 2.6, "learning_rate": 2.596964586846543e-05, "loss": 0.2513, "step": 21560 }, { "epoch": 2.6, "learning_rate": 2.5981691158756927e-05, "loss": 0.2505, "step": 21570 }, { "epoch": 2.6, "learning_rate": 2.5993736449048422e-05, "loss": 0.3417, "step": 21580 }, { "epoch": 2.6, "learning_rate": 2.6005781739339917e-05, "loss": 0.3816, "step": 21590 }, { "epoch": 2.6, "learning_rate": 2.6017827029631415e-05, "loss": 0.1399, "step": 21600 }, { "epoch": 2.6, "learning_rate": 2.602987231992291e-05, "loss": 0.2792, "step": 21610 }, { "epoch": 2.6, "learning_rate": 2.6041917610214406e-05, "loss": 0.2773, "step": 21620 }, { "epoch": 2.61, "learning_rate": 2.60539629005059e-05, "loss": 0.147, "step": 21630 }, { "epoch": 2.61, "learning_rate": 2.6066008190797396e-05, "loss": 0.2646, "step": 21640 }, { "epoch": 2.61, "learning_rate": 2.6078053481088895e-05, "loss": 0.1882, "step": 21650 }, { "epoch": 2.61, "learning_rate": 2.609009877138039e-05, "loss": 0.1214, "step": 21660 }, { "epoch": 2.61, "learning_rate": 2.6102144061671885e-05, "loss": 0.4093, "step": 21670 }, { "epoch": 2.61, "learning_rate": 2.6114189351963387e-05, "loss": 0.1405, "step": 21680 }, { "epoch": 2.61, "learning_rate": 2.6126234642254882e-05, "loss": 0.2164, "step": 21690 }, { "epoch": 2.61, "learning_rate": 2.6138279932546377e-05, "loss": 0.4823, "step": 21700 }, { "epoch": 2.61, "learning_rate": 2.6150325222837873e-05, "loss": 0.2866, "step": 21710 }, { "epoch": 2.62, "learning_rate": 2.6162370513129368e-05, "loss": 0.2162, "step": 21720 }, { "epoch": 2.62, "learning_rate": 2.6174415803420866e-05, "loss": 0.2746, "step": 21730 }, { "epoch": 2.62, "learning_rate": 2.618646109371236e-05, "loss": 0.2423, "step": 21740 }, { "epoch": 2.62, "learning_rate": 2.6198506384003857e-05, "loss": 0.2615, "step": 21750 }, { "epoch": 2.62, "learning_rate": 2.6210551674295352e-05, "loss": 0.3858, "step": 21760 }, { "epoch": 2.62, "learning_rate": 2.6222596964586847e-05, "loss": 0.2437, "step": 21770 }, { "epoch": 2.62, "learning_rate": 2.6234642254878346e-05, "loss": 0.1717, "step": 21780 }, { "epoch": 2.62, "learning_rate": 2.624668754516984e-05, "loss": 0.2712, "step": 21790 }, { "epoch": 2.63, "learning_rate": 2.6258732835461336e-05, "loss": 0.1748, "step": 21800 }, { "epoch": 2.63, "learning_rate": 2.627077812575283e-05, "loss": 0.296, "step": 21810 }, { "epoch": 2.63, "learning_rate": 2.6282823416044326e-05, "loss": 0.201, "step": 21820 }, { "epoch": 2.63, "learning_rate": 2.6294868706335825e-05, "loss": 0.2119, "step": 21830 }, { "epoch": 2.63, "learning_rate": 2.630691399662732e-05, "loss": 0.4415, "step": 21840 }, { "epoch": 2.63, "learning_rate": 2.6318959286918815e-05, "loss": 0.3328, "step": 21850 }, { "epoch": 2.63, "learning_rate": 2.633100457721031e-05, "loss": 0.156, "step": 21860 }, { "epoch": 2.63, "learning_rate": 2.6343049867501806e-05, "loss": 0.3419, "step": 21870 }, { "epoch": 2.64, "learning_rate": 2.63550951577933e-05, "loss": 0.2542, "step": 21880 }, { "epoch": 2.64, "learning_rate": 2.63671404480848e-05, "loss": 0.1836, "step": 21890 }, { "epoch": 2.64, "learning_rate": 2.6379185738376295e-05, "loss": 0.3249, "step": 21900 }, { "epoch": 2.64, "learning_rate": 2.639123102866779e-05, "loss": 0.4331, "step": 21910 }, { "epoch": 2.64, "learning_rate": 2.6403276318959285e-05, "loss": 0.2408, "step": 21920 }, { "epoch": 2.64, "learning_rate": 2.641532160925078e-05, "loss": 0.317, "step": 21930 }, { "epoch": 2.64, "learning_rate": 2.6427366899542282e-05, "loss": 0.3323, "step": 21940 }, { "epoch": 2.64, "learning_rate": 2.6439412189833777e-05, "loss": 0.3771, "step": 21950 }, { "epoch": 2.64, "learning_rate": 2.6451457480125276e-05, "loss": 0.3542, "step": 21960 }, { "epoch": 2.65, "learning_rate": 2.646350277041677e-05, "loss": 0.2114, "step": 21970 }, { "epoch": 2.65, "learning_rate": 2.6475548060708266e-05, "loss": 0.2272, "step": 21980 }, { "epoch": 2.65, "learning_rate": 2.648759335099976e-05, "loss": 0.3801, "step": 21990 }, { "epoch": 2.65, "learning_rate": 2.6499638641291257e-05, "loss": 0.3304, "step": 22000 }, { "epoch": 2.65, "learning_rate": 2.6511683931582755e-05, "loss": 0.527, "step": 22010 }, { "epoch": 2.65, "learning_rate": 2.652372922187425e-05, "loss": 0.3428, "step": 22020 }, { "epoch": 2.65, "learning_rate": 2.6535774512165746e-05, "loss": 0.2774, "step": 22030 }, { "epoch": 2.65, "learning_rate": 2.654781980245724e-05, "loss": 0.2585, "step": 22040 }, { "epoch": 2.66, "learning_rate": 2.6559865092748736e-05, "loss": 0.3432, "step": 22050 }, { "epoch": 2.66, "learning_rate": 2.657191038304023e-05, "loss": 0.3698, "step": 22060 }, { "epoch": 2.66, "learning_rate": 2.658395567333173e-05, "loss": 0.169, "step": 22070 }, { "epoch": 2.66, "learning_rate": 2.6596000963623225e-05, "loss": 0.3153, "step": 22080 }, { "epoch": 2.66, "learning_rate": 2.660804625391472e-05, "loss": 0.4378, "step": 22090 }, { "epoch": 2.66, "learning_rate": 2.6620091544206215e-05, "loss": 0.3538, "step": 22100 }, { "epoch": 2.66, "learning_rate": 2.663213683449771e-05, "loss": 0.5005, "step": 22110 }, { "epoch": 2.66, "learning_rate": 2.664418212478921e-05, "loss": 0.2574, "step": 22120 }, { "epoch": 2.67, "learning_rate": 2.6656227415080704e-05, "loss": 0.4529, "step": 22130 }, { "epoch": 2.67, "learning_rate": 2.66682727053722e-05, "loss": 0.3321, "step": 22140 }, { "epoch": 2.67, "learning_rate": 2.6680317995663695e-05, "loss": 0.259, "step": 22150 }, { "epoch": 2.67, "learning_rate": 2.669236328595519e-05, "loss": 0.2119, "step": 22160 }, { "epoch": 2.67, "learning_rate": 2.6704408576246685e-05, "loss": 0.4118, "step": 22170 }, { "epoch": 2.67, "learning_rate": 2.6716453866538184e-05, "loss": 0.2888, "step": 22180 }, { "epoch": 2.67, "learning_rate": 2.672849915682968e-05, "loss": 0.3953, "step": 22190 }, { "epoch": 2.67, "learning_rate": 2.674054444712118e-05, "loss": 0.3275, "step": 22200 }, { "epoch": 2.68, "learning_rate": 2.6752589737412676e-05, "loss": 0.2427, "step": 22210 }, { "epoch": 2.68, "learning_rate": 2.676463502770417e-05, "loss": 0.1397, "step": 22220 }, { "epoch": 2.68, "learning_rate": 2.6776680317995666e-05, "loss": 0.1517, "step": 22230 }, { "epoch": 2.68, "learning_rate": 2.678872560828716e-05, "loss": 0.0927, "step": 22240 }, { "epoch": 2.68, "learning_rate": 2.680077089857866e-05, "loss": 0.3317, "step": 22250 }, { "epoch": 2.68, "learning_rate": 2.6812816188870155e-05, "loss": 0.3775, "step": 22260 }, { "epoch": 2.68, "learning_rate": 2.682486147916165e-05, "loss": 0.1634, "step": 22270 }, { "epoch": 2.68, "learning_rate": 2.6836906769453146e-05, "loss": 0.1831, "step": 22280 }, { "epoch": 2.68, "learning_rate": 2.684895205974464e-05, "loss": 0.2767, "step": 22290 }, { "epoch": 2.69, "learning_rate": 2.686099735003614e-05, "loss": 0.199, "step": 22300 }, { "epoch": 2.69, "learning_rate": 2.6873042640327635e-05, "loss": 0.2277, "step": 22310 }, { "epoch": 2.69, "learning_rate": 2.688508793061913e-05, "loss": 0.4781, "step": 22320 }, { "epoch": 2.69, "learning_rate": 2.6897133220910625e-05, "loss": 0.1999, "step": 22330 }, { "epoch": 2.69, "learning_rate": 2.690917851120212e-05, "loss": 0.4113, "step": 22340 }, { "epoch": 2.69, "learning_rate": 2.6921223801493615e-05, "loss": 0.1297, "step": 22350 }, { "epoch": 2.69, "learning_rate": 2.6933269091785114e-05, "loss": 0.5356, "step": 22360 }, { "epoch": 2.69, "learning_rate": 2.694531438207661e-05, "loss": 0.1349, "step": 22370 }, { "epoch": 2.7, "learning_rate": 2.6957359672368104e-05, "loss": 0.1879, "step": 22380 }, { "epoch": 2.7, "learning_rate": 2.69694049626596e-05, "loss": 0.2681, "step": 22390 }, { "epoch": 2.7, "learning_rate": 2.6981450252951095e-05, "loss": 0.2987, "step": 22400 }, { "epoch": 2.7, "learning_rate": 2.6993495543242593e-05, "loss": 0.278, "step": 22410 }, { "epoch": 2.7, "learning_rate": 2.700554083353409e-05, "loss": 0.1949, "step": 22420 }, { "epoch": 2.7, "learning_rate": 2.7017586123825584e-05, "loss": 0.2063, "step": 22430 }, { "epoch": 2.7, "learning_rate": 2.702963141411708e-05, "loss": 0.5082, "step": 22440 }, { "epoch": 2.7, "learning_rate": 2.7041676704408574e-05, "loss": 0.5487, "step": 22450 }, { "epoch": 2.71, "learning_rate": 2.7053721994700073e-05, "loss": 0.2249, "step": 22460 }, { "epoch": 2.71, "learning_rate": 2.706576728499157e-05, "loss": 0.2472, "step": 22470 }, { "epoch": 2.71, "learning_rate": 2.707781257528307e-05, "loss": 0.2717, "step": 22480 }, { "epoch": 2.71, "learning_rate": 2.7089857865574565e-05, "loss": 0.2413, "step": 22490 }, { "epoch": 2.71, "learning_rate": 2.710190315586606e-05, "loss": 0.3517, "step": 22500 }, { "epoch": 2.71, "learning_rate": 2.7113948446157555e-05, "loss": 0.1782, "step": 22510 }, { "epoch": 2.71, "learning_rate": 2.712599373644905e-05, "loss": 0.3212, "step": 22520 }, { "epoch": 2.71, "learning_rate": 2.7138039026740546e-05, "loss": 0.4954, "step": 22530 }, { "epoch": 2.71, "learning_rate": 2.7150084317032044e-05, "loss": 0.2772, "step": 22540 }, { "epoch": 2.72, "learning_rate": 2.716212960732354e-05, "loss": 0.2065, "step": 22550 }, { "epoch": 2.72, "learning_rate": 2.7174174897615035e-05, "loss": 0.3337, "step": 22560 }, { "epoch": 2.72, "learning_rate": 2.718622018790653e-05, "loss": 0.2453, "step": 22570 }, { "epoch": 2.72, "learning_rate": 2.7198265478198025e-05, "loss": 0.1573, "step": 22580 }, { "epoch": 2.72, "learning_rate": 2.7210310768489523e-05, "loss": 0.138, "step": 22590 }, { "epoch": 2.72, "learning_rate": 2.722235605878102e-05, "loss": 0.5015, "step": 22600 }, { "epoch": 2.72, "learning_rate": 2.7234401349072514e-05, "loss": 0.4585, "step": 22610 }, { "epoch": 2.72, "learning_rate": 2.724644663936401e-05, "loss": 0.3155, "step": 22620 }, { "epoch": 2.73, "learning_rate": 2.7258491929655504e-05, "loss": 0.169, "step": 22630 }, { "epoch": 2.73, "learning_rate": 2.7270537219947e-05, "loss": 0.3197, "step": 22640 }, { "epoch": 2.73, "learning_rate": 2.7282582510238498e-05, "loss": 0.216, "step": 22650 }, { "epoch": 2.73, "learning_rate": 2.7294627800529993e-05, "loss": 0.3279, "step": 22660 }, { "epoch": 2.73, "learning_rate": 2.730667309082149e-05, "loss": 0.2055, "step": 22670 }, { "epoch": 2.73, "learning_rate": 2.7318718381112984e-05, "loss": 0.2216, "step": 22680 }, { "epoch": 2.73, "learning_rate": 2.733076367140448e-05, "loss": 0.332, "step": 22690 }, { "epoch": 2.73, "learning_rate": 2.7342808961695977e-05, "loss": 0.5532, "step": 22700 }, { "epoch": 2.74, "learning_rate": 2.7354854251987472e-05, "loss": 0.4094, "step": 22710 }, { "epoch": 2.74, "learning_rate": 2.7366899542278968e-05, "loss": 0.1958, "step": 22720 }, { "epoch": 2.74, "learning_rate": 2.737894483257047e-05, "loss": 0.2323, "step": 22730 }, { "epoch": 2.74, "learning_rate": 2.7390990122861965e-05, "loss": 0.4287, "step": 22740 }, { "epoch": 2.74, "learning_rate": 2.740303541315346e-05, "loss": 0.4302, "step": 22750 }, { "epoch": 2.74, "learning_rate": 2.7415080703444955e-05, "loss": 0.2446, "step": 22760 }, { "epoch": 2.74, "learning_rate": 2.7427125993736454e-05, "loss": 0.4307, "step": 22770 }, { "epoch": 2.74, "learning_rate": 2.743917128402795e-05, "loss": 0.2247, "step": 22780 }, { "epoch": 2.74, "learning_rate": 2.7451216574319444e-05, "loss": 0.2067, "step": 22790 }, { "epoch": 2.75, "learning_rate": 2.746326186461094e-05, "loss": 0.3797, "step": 22800 }, { "epoch": 2.75, "learning_rate": 2.7475307154902434e-05, "loss": 0.2439, "step": 22810 }, { "epoch": 2.75, "learning_rate": 2.748735244519393e-05, "loss": 0.2485, "step": 22820 }, { "epoch": 2.75, "learning_rate": 2.7499397735485428e-05, "loss": 0.4136, "step": 22830 }, { "epoch": 2.75, "learning_rate": 2.7511443025776923e-05, "loss": 0.1345, "step": 22840 }, { "epoch": 2.75, "learning_rate": 2.752348831606842e-05, "loss": 0.3076, "step": 22850 }, { "epoch": 2.75, "learning_rate": 2.7535533606359914e-05, "loss": 0.3709, "step": 22860 }, { "epoch": 2.75, "learning_rate": 2.754757889665141e-05, "loss": 0.2915, "step": 22870 }, { "epoch": 2.76, "learning_rate": 2.7559624186942908e-05, "loss": 0.226, "step": 22880 }, { "epoch": 2.76, "learning_rate": 2.7571669477234403e-05, "loss": 0.5767, "step": 22890 }, { "epoch": 2.76, "learning_rate": 2.7583714767525898e-05, "loss": 0.1771, "step": 22900 }, { "epoch": 2.76, "learning_rate": 2.7595760057817393e-05, "loss": 0.2592, "step": 22910 }, { "epoch": 2.76, "learning_rate": 2.7607805348108888e-05, "loss": 0.3167, "step": 22920 }, { "epoch": 2.76, "learning_rate": 2.7619850638400387e-05, "loss": 0.3876, "step": 22930 }, { "epoch": 2.76, "learning_rate": 2.7631895928691882e-05, "loss": 0.2299, "step": 22940 }, { "epoch": 2.76, "learning_rate": 2.7643941218983377e-05, "loss": 0.2058, "step": 22950 }, { "epoch": 2.77, "learning_rate": 2.7655986509274872e-05, "loss": 0.3383, "step": 22960 }, { "epoch": 2.77, "learning_rate": 2.7668031799566368e-05, "loss": 0.0507, "step": 22970 }, { "epoch": 2.77, "learning_rate": 2.7680077089857863e-05, "loss": 0.389, "step": 22980 }, { "epoch": 2.77, "learning_rate": 2.7692122380149365e-05, "loss": 0.3891, "step": 22990 }, { "epoch": 2.77, "learning_rate": 2.770416767044086e-05, "loss": 0.1275, "step": 23000 }, { "epoch": 2.77, "learning_rate": 2.771621296073236e-05, "loss": 0.2209, "step": 23010 }, { "epoch": 2.77, "learning_rate": 2.7728258251023854e-05, "loss": 0.4105, "step": 23020 }, { "epoch": 2.77, "learning_rate": 2.774030354131535e-05, "loss": 0.1448, "step": 23030 }, { "epoch": 2.78, "learning_rate": 2.7752348831606844e-05, "loss": 0.3497, "step": 23040 }, { "epoch": 2.78, "learning_rate": 2.776439412189834e-05, "loss": 0.4104, "step": 23050 }, { "epoch": 2.78, "learning_rate": 2.7776439412189838e-05, "loss": 0.3292, "step": 23060 }, { "epoch": 2.78, "learning_rate": 2.7788484702481333e-05, "loss": 0.4962, "step": 23070 }, { "epoch": 2.78, "learning_rate": 2.7800529992772828e-05, "loss": 0.2761, "step": 23080 }, { "epoch": 2.78, "learning_rate": 2.7812575283064323e-05, "loss": 0.2209, "step": 23090 }, { "epoch": 2.78, "learning_rate": 2.782462057335582e-05, "loss": 0.4074, "step": 23100 }, { "epoch": 2.78, "learning_rate": 2.7836665863647314e-05, "loss": 0.368, "step": 23110 }, { "epoch": 2.78, "learning_rate": 2.7848711153938812e-05, "loss": 0.1944, "step": 23120 }, { "epoch": 2.79, "learning_rate": 2.7860756444230308e-05, "loss": 0.2146, "step": 23130 }, { "epoch": 2.79, "learning_rate": 2.7872801734521803e-05, "loss": 0.3026, "step": 23140 }, { "epoch": 2.79, "learning_rate": 2.7884847024813298e-05, "loss": 0.2399, "step": 23150 }, { "epoch": 2.79, "learning_rate": 2.7896892315104793e-05, "loss": 0.2011, "step": 23160 }, { "epoch": 2.79, "learning_rate": 2.790893760539629e-05, "loss": 0.3207, "step": 23170 }, { "epoch": 2.79, "learning_rate": 2.7920982895687787e-05, "loss": 0.3043, "step": 23180 }, { "epoch": 2.79, "learning_rate": 2.7933028185979282e-05, "loss": 0.1911, "step": 23190 }, { "epoch": 2.79, "learning_rate": 2.7945073476270777e-05, "loss": 0.4578, "step": 23200 }, { "epoch": 2.8, "learning_rate": 2.7957118766562272e-05, "loss": 0.2688, "step": 23210 }, { "epoch": 2.8, "learning_rate": 2.796916405685377e-05, "loss": 0.197, "step": 23220 }, { "epoch": 2.8, "learning_rate": 2.7981209347145266e-05, "loss": 0.2193, "step": 23230 }, { "epoch": 2.8, "learning_rate": 2.799325463743676e-05, "loss": 0.2328, "step": 23240 }, { "epoch": 2.8, "learning_rate": 2.8005299927728257e-05, "loss": 0.2603, "step": 23250 }, { "epoch": 2.8, "learning_rate": 2.801734521801976e-05, "loss": 0.4764, "step": 23260 }, { "epoch": 2.8, "learning_rate": 2.8029390508311254e-05, "loss": 0.241, "step": 23270 }, { "epoch": 2.8, "learning_rate": 2.804143579860275e-05, "loss": 0.2272, "step": 23280 }, { "epoch": 2.81, "learning_rate": 2.8053481088894244e-05, "loss": 0.3157, "step": 23290 }, { "epoch": 2.81, "learning_rate": 2.8065526379185743e-05, "loss": 0.1755, "step": 23300 }, { "epoch": 2.81, "learning_rate": 2.8077571669477238e-05, "loss": 0.2747, "step": 23310 }, { "epoch": 2.81, "learning_rate": 2.8089616959768733e-05, "loss": 0.3466, "step": 23320 }, { "epoch": 2.81, "learning_rate": 2.8101662250060228e-05, "loss": 0.3747, "step": 23330 }, { "epoch": 2.81, "learning_rate": 2.8113707540351723e-05, "loss": 0.4825, "step": 23340 }, { "epoch": 2.81, "learning_rate": 2.8125752830643222e-05, "loss": 0.2095, "step": 23350 }, { "epoch": 2.81, "learning_rate": 2.8137798120934717e-05, "loss": 0.2131, "step": 23360 }, { "epoch": 2.81, "learning_rate": 2.8149843411226212e-05, "loss": 0.444, "step": 23370 }, { "epoch": 2.82, "learning_rate": 2.8161888701517707e-05, "loss": 0.248, "step": 23380 }, { "epoch": 2.82, "learning_rate": 2.8173933991809203e-05, "loss": 0.2881, "step": 23390 }, { "epoch": 2.82, "learning_rate": 2.81859792821007e-05, "loss": 0.3376, "step": 23400 }, { "epoch": 2.82, "learning_rate": 2.8198024572392196e-05, "loss": 0.3045, "step": 23410 }, { "epoch": 2.82, "learning_rate": 2.821006986268369e-05, "loss": 0.2366, "step": 23420 }, { "epoch": 2.82, "learning_rate": 2.8222115152975187e-05, "loss": 0.3722, "step": 23430 }, { "epoch": 2.82, "learning_rate": 2.8234160443266682e-05, "loss": 0.1959, "step": 23440 }, { "epoch": 2.82, "learning_rate": 2.8246205733558177e-05, "loss": 0.3344, "step": 23450 }, { "epoch": 2.83, "learning_rate": 2.8258251023849676e-05, "loss": 0.4158, "step": 23460 }, { "epoch": 2.83, "learning_rate": 2.827029631414117e-05, "loss": 0.1405, "step": 23470 }, { "epoch": 2.83, "learning_rate": 2.8282341604432666e-05, "loss": 0.1122, "step": 23480 }, { "epoch": 2.83, "learning_rate": 2.829438689472416e-05, "loss": 0.5645, "step": 23490 }, { "epoch": 2.83, "learning_rate": 2.8306432185015656e-05, "loss": 0.221, "step": 23500 }, { "epoch": 2.83, "learning_rate": 2.8318477475307155e-05, "loss": 0.3729, "step": 23510 }, { "epoch": 2.83, "learning_rate": 2.8330522765598654e-05, "loss": 0.2177, "step": 23520 }, { "epoch": 2.83, "learning_rate": 2.8342568055890152e-05, "loss": 0.3059, "step": 23530 }, { "epoch": 2.84, "learning_rate": 2.8354613346181647e-05, "loss": 0.3359, "step": 23540 }, { "epoch": 2.84, "learning_rate": 2.8366658636473143e-05, "loss": 0.3591, "step": 23550 }, { "epoch": 2.84, "learning_rate": 2.8378703926764638e-05, "loss": 0.1704, "step": 23560 }, { "epoch": 2.84, "learning_rate": 2.8390749217056133e-05, "loss": 0.1867, "step": 23570 }, { "epoch": 2.84, "learning_rate": 2.840279450734763e-05, "loss": 0.3245, "step": 23580 }, { "epoch": 2.84, "learning_rate": 2.8414839797639127e-05, "loss": 0.4672, "step": 23590 }, { "epoch": 2.84, "learning_rate": 2.8426885087930622e-05, "loss": 0.4372, "step": 23600 }, { "epoch": 2.84, "learning_rate": 2.8438930378222117e-05, "loss": 0.3602, "step": 23610 }, { "epoch": 2.84, "learning_rate": 2.8450975668513612e-05, "loss": 0.2653, "step": 23620 }, { "epoch": 2.85, "learning_rate": 2.8463020958805107e-05, "loss": 0.323, "step": 23630 }, { "epoch": 2.85, "learning_rate": 2.8475066249096606e-05, "loss": 0.3061, "step": 23640 }, { "epoch": 2.85, "learning_rate": 2.84871115393881e-05, "loss": 0.3244, "step": 23650 }, { "epoch": 2.85, "learning_rate": 2.8499156829679596e-05, "loss": 0.3713, "step": 23660 }, { "epoch": 2.85, "learning_rate": 2.851120211997109e-05, "loss": 0.4599, "step": 23670 }, { "epoch": 2.85, "learning_rate": 2.8523247410262587e-05, "loss": 0.3721, "step": 23680 }, { "epoch": 2.85, "learning_rate": 2.8535292700554085e-05, "loss": 0.3501, "step": 23690 }, { "epoch": 2.85, "learning_rate": 2.854733799084558e-05, "loss": 0.2259, "step": 23700 }, { "epoch": 2.86, "learning_rate": 2.8559383281137076e-05, "loss": 0.2451, "step": 23710 }, { "epoch": 2.86, "learning_rate": 2.857142857142857e-05, "loss": 0.2708, "step": 23720 }, { "epoch": 2.86, "learning_rate": 2.8583473861720066e-05, "loss": 0.1206, "step": 23730 }, { "epoch": 2.86, "learning_rate": 2.859551915201156e-05, "loss": 0.3059, "step": 23740 }, { "epoch": 2.86, "learning_rate": 2.860756444230306e-05, "loss": 0.3561, "step": 23750 }, { "epoch": 2.86, "learning_rate": 2.8619609732594555e-05, "loss": 0.4388, "step": 23760 }, { "epoch": 2.86, "learning_rate": 2.863165502288605e-05, "loss": 0.3176, "step": 23770 }, { "epoch": 2.86, "learning_rate": 2.8643700313177552e-05, "loss": 0.3199, "step": 23780 }, { "epoch": 2.87, "learning_rate": 2.8655745603469047e-05, "loss": 0.2251, "step": 23790 }, { "epoch": 2.87, "learning_rate": 2.8667790893760543e-05, "loss": 0.2392, "step": 23800 }, { "epoch": 2.87, "learning_rate": 2.8679836184052038e-05, "loss": 0.254, "step": 23810 }, { "epoch": 2.87, "learning_rate": 2.8691881474343536e-05, "loss": 0.157, "step": 23820 }, { "epoch": 2.87, "learning_rate": 2.870392676463503e-05, "loss": 0.2289, "step": 23830 }, { "epoch": 2.87, "learning_rate": 2.8715972054926527e-05, "loss": 0.154, "step": 23840 }, { "epoch": 2.87, "learning_rate": 2.8728017345218022e-05, "loss": 0.2883, "step": 23850 }, { "epoch": 2.87, "learning_rate": 2.8740062635509517e-05, "loss": 0.3291, "step": 23860 }, { "epoch": 2.88, "learning_rate": 2.8752107925801016e-05, "loss": 0.3571, "step": 23870 }, { "epoch": 2.88, "learning_rate": 2.876415321609251e-05, "loss": 0.1982, "step": 23880 }, { "epoch": 2.88, "learning_rate": 2.8776198506384006e-05, "loss": 0.3095, "step": 23890 }, { "epoch": 2.88, "learning_rate": 2.87882437966755e-05, "loss": 0.2637, "step": 23900 }, { "epoch": 2.88, "learning_rate": 2.8800289086966996e-05, "loss": 0.5958, "step": 23910 }, { "epoch": 2.88, "learning_rate": 2.881233437725849e-05, "loss": 0.1342, "step": 23920 }, { "epoch": 2.88, "learning_rate": 2.882437966754999e-05, "loss": 0.1502, "step": 23930 }, { "epoch": 2.88, "learning_rate": 2.8836424957841485e-05, "loss": 0.341, "step": 23940 }, { "epoch": 2.88, "learning_rate": 2.884847024813298e-05, "loss": 0.3126, "step": 23950 }, { "epoch": 2.89, "learning_rate": 2.8860515538424476e-05, "loss": 0.5042, "step": 23960 }, { "epoch": 2.89, "learning_rate": 2.887256082871597e-05, "loss": 0.318, "step": 23970 }, { "epoch": 2.89, "learning_rate": 2.888460611900747e-05, "loss": 0.3203, "step": 23980 }, { "epoch": 2.89, "learning_rate": 2.8896651409298965e-05, "loss": 0.2613, "step": 23990 }, { "epoch": 2.89, "learning_rate": 2.890869669959046e-05, "loss": 0.4611, "step": 24000 }, { "epoch": 2.89, "learning_rate": 2.8920741989881955e-05, "loss": 0.3859, "step": 24010 }, { "epoch": 2.89, "learning_rate": 2.893278728017345e-05, "loss": 0.2085, "step": 24020 }, { "epoch": 2.89, "learning_rate": 2.894483257046495e-05, "loss": 0.2968, "step": 24030 }, { "epoch": 2.9, "learning_rate": 2.8956877860756447e-05, "loss": 0.2486, "step": 24040 }, { "epoch": 2.9, "learning_rate": 2.8968923151047946e-05, "loss": 0.2532, "step": 24050 }, { "epoch": 2.9, "learning_rate": 2.898096844133944e-05, "loss": 0.363, "step": 24060 }, { "epoch": 2.9, "learning_rate": 2.8993013731630936e-05, "loss": 0.2028, "step": 24070 }, { "epoch": 2.9, "learning_rate": 2.900505902192243e-05, "loss": 0.159, "step": 24080 }, { "epoch": 2.9, "learning_rate": 2.9017104312213927e-05, "loss": 0.156, "step": 24090 }, { "epoch": 2.9, "learning_rate": 2.9029149602505422e-05, "loss": 0.2031, "step": 24100 }, { "epoch": 2.9, "learning_rate": 2.904119489279692e-05, "loss": 0.3288, "step": 24110 }, { "epoch": 2.91, "learning_rate": 2.9053240183088416e-05, "loss": 0.3644, "step": 24120 }, { "epoch": 2.91, "learning_rate": 2.906528547337991e-05, "loss": 0.1941, "step": 24130 }, { "epoch": 2.91, "learning_rate": 2.9077330763671406e-05, "loss": 0.2012, "step": 24140 }, { "epoch": 2.91, "learning_rate": 2.90893760539629e-05, "loss": 0.319, "step": 24150 }, { "epoch": 2.91, "learning_rate": 2.91014213442544e-05, "loss": 0.3506, "step": 24160 }, { "epoch": 2.91, "learning_rate": 2.9113466634545895e-05, "loss": 0.3147, "step": 24170 }, { "epoch": 2.91, "learning_rate": 2.912551192483739e-05, "loss": 0.4684, "step": 24180 }, { "epoch": 2.91, "learning_rate": 2.9137557215128885e-05, "loss": 0.0931, "step": 24190 }, { "epoch": 2.91, "learning_rate": 2.914960250542038e-05, "loss": 0.3831, "step": 24200 }, { "epoch": 2.92, "learning_rate": 2.9161647795711876e-05, "loss": 0.1619, "step": 24210 }, { "epoch": 2.92, "learning_rate": 2.9173693086003374e-05, "loss": 0.3145, "step": 24220 }, { "epoch": 2.92, "learning_rate": 2.918573837629487e-05, "loss": 0.3001, "step": 24230 }, { "epoch": 2.92, "learning_rate": 2.9197783666586365e-05, "loss": 0.1481, "step": 24240 }, { "epoch": 2.92, "learning_rate": 2.920982895687786e-05, "loss": 0.2407, "step": 24250 }, { "epoch": 2.92, "learning_rate": 2.9221874247169355e-05, "loss": 0.4357, "step": 24260 }, { "epoch": 2.92, "learning_rate": 2.9233919537460853e-05, "loss": 0.345, "step": 24270 }, { "epoch": 2.92, "learning_rate": 2.924596482775235e-05, "loss": 0.1391, "step": 24280 }, { "epoch": 2.93, "learning_rate": 2.9258010118043844e-05, "loss": 0.1861, "step": 24290 }, { "epoch": 2.93, "learning_rate": 2.927005540833534e-05, "loss": 0.2042, "step": 24300 }, { "epoch": 2.93, "learning_rate": 2.928210069862684e-05, "loss": 0.4824, "step": 24310 }, { "epoch": 2.93, "learning_rate": 2.9294145988918336e-05, "loss": 0.3685, "step": 24320 }, { "epoch": 2.93, "learning_rate": 2.930619127920983e-05, "loss": 0.1789, "step": 24330 }, { "epoch": 2.93, "learning_rate": 2.931823656950133e-05, "loss": 0.1879, "step": 24340 }, { "epoch": 2.93, "learning_rate": 2.9330281859792825e-05, "loss": 0.1064, "step": 24350 }, { "epoch": 2.93, "learning_rate": 2.934232715008432e-05, "loss": 0.0305, "step": 24360 }, { "epoch": 2.94, "learning_rate": 2.9354372440375815e-05, "loss": 0.2212, "step": 24370 }, { "epoch": 2.94, "learning_rate": 2.936641773066731e-05, "loss": 0.2533, "step": 24380 }, { "epoch": 2.94, "learning_rate": 2.9378463020958806e-05, "loss": 0.3723, "step": 24390 }, { "epoch": 2.94, "learning_rate": 2.9390508311250304e-05, "loss": 0.3442, "step": 24400 }, { "epoch": 2.94, "learning_rate": 2.94025536015418e-05, "loss": 0.4811, "step": 24410 }, { "epoch": 2.94, "learning_rate": 2.9414598891833295e-05, "loss": 0.2306, "step": 24420 }, { "epoch": 2.94, "learning_rate": 2.942664418212479e-05, "loss": 0.2211, "step": 24430 }, { "epoch": 2.94, "learning_rate": 2.9438689472416285e-05, "loss": 0.169, "step": 24440 }, { "epoch": 2.94, "learning_rate": 2.9450734762707784e-05, "loss": 0.1488, "step": 24450 }, { "epoch": 2.95, "learning_rate": 2.946278005299928e-05, "loss": 0.0416, "step": 24460 }, { "epoch": 2.95, "learning_rate": 2.9474825343290774e-05, "loss": 0.2125, "step": 24470 }, { "epoch": 2.95, "learning_rate": 2.948687063358227e-05, "loss": 0.2843, "step": 24480 }, { "epoch": 2.95, "learning_rate": 2.9498915923873764e-05, "loss": 0.1981, "step": 24490 }, { "epoch": 2.95, "learning_rate": 2.9510961214165263e-05, "loss": 0.2135, "step": 24500 }, { "epoch": 2.95, "learning_rate": 2.9523006504456758e-05, "loss": 0.3554, "step": 24510 }, { "epoch": 2.95, "learning_rate": 2.9535051794748253e-05, "loss": 0.2508, "step": 24520 }, { "epoch": 2.95, "learning_rate": 2.954709708503975e-05, "loss": 0.1298, "step": 24530 }, { "epoch": 2.96, "learning_rate": 2.9559142375331244e-05, "loss": 0.1082, "step": 24540 }, { "epoch": 2.96, "learning_rate": 2.957118766562274e-05, "loss": 0.4061, "step": 24550 }, { "epoch": 2.96, "learning_rate": 2.9583232955914238e-05, "loss": 0.3149, "step": 24560 }, { "epoch": 2.96, "learning_rate": 2.9595278246205736e-05, "loss": 0.2459, "step": 24570 }, { "epoch": 2.96, "learning_rate": 2.9607323536497235e-05, "loss": 0.2732, "step": 24580 }, { "epoch": 2.96, "learning_rate": 2.961936882678873e-05, "loss": 0.2766, "step": 24590 }, { "epoch": 2.96, "learning_rate": 2.9631414117080225e-05, "loss": 0.4046, "step": 24600 }, { "epoch": 2.96, "learning_rate": 2.964345940737172e-05, "loss": 0.4159, "step": 24610 }, { "epoch": 2.97, "learning_rate": 2.9655504697663215e-05, "loss": 0.2421, "step": 24620 }, { "epoch": 2.97, "learning_rate": 2.9667549987954714e-05, "loss": 0.1116, "step": 24630 }, { "epoch": 2.97, "learning_rate": 2.967959527824621e-05, "loss": 0.2036, "step": 24640 }, { "epoch": 2.97, "learning_rate": 2.9691640568537704e-05, "loss": 0.1473, "step": 24650 }, { "epoch": 2.97, "learning_rate": 2.97036858588292e-05, "loss": 0.2274, "step": 24660 }, { "epoch": 2.97, "learning_rate": 2.9715731149120695e-05, "loss": 0.3042, "step": 24670 }, { "epoch": 2.97, "learning_rate": 2.972777643941219e-05, "loss": 0.2341, "step": 24680 }, { "epoch": 2.97, "learning_rate": 2.973982172970369e-05, "loss": 0.2255, "step": 24690 }, { "epoch": 2.98, "learning_rate": 2.9751867019995184e-05, "loss": 0.2568, "step": 24700 }, { "epoch": 2.98, "learning_rate": 2.976391231028668e-05, "loss": 0.2563, "step": 24710 }, { "epoch": 2.98, "learning_rate": 2.9775957600578174e-05, "loss": 0.215, "step": 24720 }, { "epoch": 2.98, "learning_rate": 2.978800289086967e-05, "loss": 0.3708, "step": 24730 }, { "epoch": 2.98, "learning_rate": 2.9800048181161168e-05, "loss": 0.1765, "step": 24740 }, { "epoch": 2.98, "learning_rate": 2.9812093471452663e-05, "loss": 0.1991, "step": 24750 }, { "epoch": 2.98, "learning_rate": 2.9824138761744158e-05, "loss": 0.398, "step": 24760 }, { "epoch": 2.98, "learning_rate": 2.9836184052035653e-05, "loss": 0.2089, "step": 24770 }, { "epoch": 2.98, "learning_rate": 2.984822934232715e-05, "loss": 0.3183, "step": 24780 }, { "epoch": 2.99, "learning_rate": 2.9860274632618647e-05, "loss": 0.2312, "step": 24790 }, { "epoch": 2.99, "learning_rate": 2.9872319922910142e-05, "loss": 0.2768, "step": 24800 }, { "epoch": 2.99, "learning_rate": 2.9884365213201637e-05, "loss": 0.236, "step": 24810 }, { "epoch": 2.99, "learning_rate": 2.9896410503493133e-05, "loss": 0.1127, "step": 24820 }, { "epoch": 2.99, "learning_rate": 2.9908455793784635e-05, "loss": 0.2971, "step": 24830 }, { "epoch": 2.99, "learning_rate": 2.992050108407613e-05, "loss": 0.1909, "step": 24840 }, { "epoch": 2.99, "learning_rate": 2.9932546374367625e-05, "loss": 0.3927, "step": 24850 }, { "epoch": 2.99, "learning_rate": 2.994459166465912e-05, "loss": 0.2993, "step": 24860 }, { "epoch": 3.0, "learning_rate": 2.995663695495062e-05, "loss": 0.1727, "step": 24870 }, { "epoch": 3.0, "learning_rate": 2.9968682245242114e-05, "loss": 0.336, "step": 24880 }, { "epoch": 3.0, "learning_rate": 2.998072753553361e-05, "loss": 0.3784, "step": 24890 }, { "epoch": 3.0, "learning_rate": 2.9992772825825104e-05, "loss": 0.5678, "step": 24900 }, { "epoch": 3.0, "eval_accuracy": { "accuracy": 0.9118940078289671 }, "eval_f1": { "f1": 0.9116597109413729 }, "eval_loss": 0.3114425837993622, "eval_precision": { "precision": 0.9165908158739848 }, "eval_recall": { "recall": 0.9116030141797212 }, "eval_runtime": 457.3219, "eval_samples_per_second": 36.309, "eval_steps_per_second": 18.156, "step": 24907 }, { "epoch": 3.0, "learning_rate": 3.00048181161166e-05, "loss": 0.2599, "step": 24910 }, { "epoch": 3.0, "learning_rate": 3.0016863406408098e-05, "loss": 0.172, "step": 24920 }, { "epoch": 3.0, "learning_rate": 3.0028908696699593e-05, "loss": 0.2137, "step": 24930 }, { "epoch": 3.0, "learning_rate": 3.004095398699109e-05, "loss": 0.4116, "step": 24940 }, { "epoch": 3.01, "learning_rate": 3.0052999277282584e-05, "loss": 0.3722, "step": 24950 }, { "epoch": 3.01, "learning_rate": 3.006504456757408e-05, "loss": 0.1935, "step": 24960 }, { "epoch": 3.01, "learning_rate": 3.0077089857865577e-05, "loss": 0.2207, "step": 24970 }, { "epoch": 3.01, "learning_rate": 3.0089135148157073e-05, "loss": 0.4092, "step": 24980 }, { "epoch": 3.01, "learning_rate": 3.0101180438448568e-05, "loss": 0.3625, "step": 24990 }, { "epoch": 3.01, "learning_rate": 3.0113225728740063e-05, "loss": 0.2725, "step": 25000 }, { "epoch": 3.01, "learning_rate": 3.0125271019031558e-05, "loss": 0.1696, "step": 25010 }, { "epoch": 3.01, "learning_rate": 3.0137316309323053e-05, "loss": 0.1985, "step": 25020 }, { "epoch": 3.01, "learning_rate": 3.0149361599614552e-05, "loss": 0.2278, "step": 25030 }, { "epoch": 3.02, "learning_rate": 3.0161406889906047e-05, "loss": 0.1984, "step": 25040 }, { "epoch": 3.02, "learning_rate": 3.0173452180197542e-05, "loss": 0.4152, "step": 25050 }, { "epoch": 3.02, "learning_rate": 3.0185497470489037e-05, "loss": 0.3541, "step": 25060 }, { "epoch": 3.02, "learning_rate": 3.0197542760780533e-05, "loss": 0.1387, "step": 25070 }, { "epoch": 3.02, "learning_rate": 3.020958805107203e-05, "loss": 0.2596, "step": 25080 }, { "epoch": 3.02, "learning_rate": 3.022163334136353e-05, "loss": 0.1854, "step": 25090 }, { "epoch": 3.02, "learning_rate": 3.023367863165503e-05, "loss": 0.2492, "step": 25100 }, { "epoch": 3.02, "learning_rate": 3.0245723921946524e-05, "loss": 0.2725, "step": 25110 }, { "epoch": 3.03, "learning_rate": 3.025776921223802e-05, "loss": 0.2586, "step": 25120 }, { "epoch": 3.03, "learning_rate": 3.0269814502529514e-05, "loss": 0.2255, "step": 25130 }, { "epoch": 3.03, "learning_rate": 3.028185979282101e-05, "loss": 0.1305, "step": 25140 }, { "epoch": 3.03, "learning_rate": 3.0293905083112504e-05, "loss": 0.3667, "step": 25150 }, { "epoch": 3.03, "learning_rate": 3.0305950373404003e-05, "loss": 0.1553, "step": 25160 }, { "epoch": 3.03, "learning_rate": 3.0317995663695498e-05, "loss": 0.1278, "step": 25170 }, { "epoch": 3.03, "learning_rate": 3.0330040953986993e-05, "loss": 0.2466, "step": 25180 }, { "epoch": 3.03, "learning_rate": 3.034208624427849e-05, "loss": 0.2929, "step": 25190 }, { "epoch": 3.04, "learning_rate": 3.0354131534569984e-05, "loss": 0.2995, "step": 25200 }, { "epoch": 3.04, "learning_rate": 3.0366176824861482e-05, "loss": 0.2917, "step": 25210 }, { "epoch": 3.04, "learning_rate": 3.0378222115152977e-05, "loss": 0.1022, "step": 25220 }, { "epoch": 3.04, "learning_rate": 3.0390267405444473e-05, "loss": 0.3321, "step": 25230 }, { "epoch": 3.04, "learning_rate": 3.0402312695735968e-05, "loss": 0.36, "step": 25240 }, { "epoch": 3.04, "learning_rate": 3.0414357986027463e-05, "loss": 0.1658, "step": 25250 }, { "epoch": 3.04, "learning_rate": 3.042640327631896e-05, "loss": 0.2034, "step": 25260 }, { "epoch": 3.04, "learning_rate": 3.0438448566610457e-05, "loss": 0.1107, "step": 25270 }, { "epoch": 3.04, "learning_rate": 3.0450493856901952e-05, "loss": 0.3025, "step": 25280 }, { "epoch": 3.05, "learning_rate": 3.0462539147193447e-05, "loss": 0.2878, "step": 25290 }, { "epoch": 3.05, "learning_rate": 3.0474584437484942e-05, "loss": 0.4318, "step": 25300 }, { "epoch": 3.05, "learning_rate": 3.0486629727776437e-05, "loss": 0.351, "step": 25310 }, { "epoch": 3.05, "learning_rate": 3.0498675018067936e-05, "loss": 0.266, "step": 25320 }, { "epoch": 3.05, "learning_rate": 3.051072030835943e-05, "loss": 0.4601, "step": 25330 }, { "epoch": 3.05, "learning_rate": 3.0522765598650926e-05, "loss": 0.2249, "step": 25340 }, { "epoch": 3.05, "learning_rate": 3.053481088894242e-05, "loss": 0.2655, "step": 25350 }, { "epoch": 3.05, "learning_rate": 3.0546856179233923e-05, "loss": 0.1564, "step": 25360 }, { "epoch": 3.06, "learning_rate": 3.055890146952542e-05, "loss": 0.22, "step": 25370 }, { "epoch": 3.06, "learning_rate": 3.0570946759816914e-05, "loss": 0.2266, "step": 25380 }, { "epoch": 3.06, "learning_rate": 3.058299205010841e-05, "loss": 0.2388, "step": 25390 }, { "epoch": 3.06, "learning_rate": 3.0595037340399904e-05, "loss": 0.5174, "step": 25400 }, { "epoch": 3.06, "learning_rate": 3.06070826306914e-05, "loss": 0.2869, "step": 25410 }, { "epoch": 3.06, "learning_rate": 3.06191279209829e-05, "loss": 0.1731, "step": 25420 }, { "epoch": 3.06, "learning_rate": 3.0631173211274397e-05, "loss": 0.2053, "step": 25430 }, { "epoch": 3.06, "learning_rate": 3.064321850156589e-05, "loss": 0.3669, "step": 25440 }, { "epoch": 3.07, "learning_rate": 3.065526379185739e-05, "loss": 0.3826, "step": 25450 }, { "epoch": 3.07, "learning_rate": 3.066730908214888e-05, "loss": 0.2083, "step": 25460 }, { "epoch": 3.07, "learning_rate": 3.067935437244038e-05, "loss": 0.3435, "step": 25470 }, { "epoch": 3.07, "learning_rate": 3.069139966273187e-05, "loss": 0.2688, "step": 25480 }, { "epoch": 3.07, "learning_rate": 3.070344495302337e-05, "loss": 0.3383, "step": 25490 }, { "epoch": 3.07, "learning_rate": 3.071549024331486e-05, "loss": 0.3739, "step": 25500 }, { "epoch": 3.07, "learning_rate": 3.072753553360636e-05, "loss": 0.2326, "step": 25510 }, { "epoch": 3.07, "learning_rate": 3.073958082389786e-05, "loss": 0.2474, "step": 25520 }, { "epoch": 3.07, "learning_rate": 3.0751626114189355e-05, "loss": 0.2917, "step": 25530 }, { "epoch": 3.08, "learning_rate": 3.076367140448085e-05, "loss": 0.2756, "step": 25540 }, { "epoch": 3.08, "learning_rate": 3.0775716694772346e-05, "loss": 0.2438, "step": 25550 }, { "epoch": 3.08, "learning_rate": 3.078776198506384e-05, "loss": 0.168, "step": 25560 }, { "epoch": 3.08, "learning_rate": 3.0799807275355336e-05, "loss": 0.2376, "step": 25570 }, { "epoch": 3.08, "learning_rate": 3.081185256564683e-05, "loss": 0.2065, "step": 25580 }, { "epoch": 3.08, "learning_rate": 3.0823897855938326e-05, "loss": 0.4056, "step": 25590 }, { "epoch": 3.08, "learning_rate": 3.083594314622982e-05, "loss": 0.2064, "step": 25600 }, { "epoch": 3.08, "learning_rate": 3.084798843652132e-05, "loss": 0.1649, "step": 25610 }, { "epoch": 3.09, "learning_rate": 3.086003372681282e-05, "loss": 0.2391, "step": 25620 }, { "epoch": 3.09, "learning_rate": 3.0872079017104314e-05, "loss": 0.1636, "step": 25630 }, { "epoch": 3.09, "learning_rate": 3.088412430739581e-05, "loss": 0.4071, "step": 25640 }, { "epoch": 3.09, "learning_rate": 3.089616959768731e-05, "loss": 0.0904, "step": 25650 }, { "epoch": 3.09, "learning_rate": 3.0908214887978806e-05, "loss": 0.2451, "step": 25660 }, { "epoch": 3.09, "learning_rate": 3.09202601782703e-05, "loss": 0.3351, "step": 25670 }, { "epoch": 3.09, "learning_rate": 3.0932305468561797e-05, "loss": 0.3635, "step": 25680 }, { "epoch": 3.09, "learning_rate": 3.094435075885329e-05, "loss": 0.3487, "step": 25690 }, { "epoch": 3.1, "learning_rate": 3.095639604914479e-05, "loss": 0.4103, "step": 25700 }, { "epoch": 3.1, "learning_rate": 3.096844133943628e-05, "loss": 0.2533, "step": 25710 }, { "epoch": 3.1, "learning_rate": 3.098048662972778e-05, "loss": 0.2963, "step": 25720 }, { "epoch": 3.1, "learning_rate": 3.099253192001927e-05, "loss": 0.239, "step": 25730 }, { "epoch": 3.1, "learning_rate": 3.100457721031077e-05, "loss": 0.1837, "step": 25740 }, { "epoch": 3.1, "learning_rate": 3.101662250060226e-05, "loss": 0.3309, "step": 25750 }, { "epoch": 3.1, "learning_rate": 3.1028667790893765e-05, "loss": 0.4147, "step": 25760 }, { "epoch": 3.1, "learning_rate": 3.104071308118526e-05, "loss": 0.274, "step": 25770 }, { "epoch": 3.11, "learning_rate": 3.1052758371476755e-05, "loss": 0.3612, "step": 25780 }, { "epoch": 3.11, "learning_rate": 3.106480366176825e-05, "loss": 0.2133, "step": 25790 }, { "epoch": 3.11, "learning_rate": 3.1076848952059746e-05, "loss": 0.0694, "step": 25800 }, { "epoch": 3.11, "learning_rate": 3.108889424235124e-05, "loss": 0.3588, "step": 25810 }, { "epoch": 3.11, "learning_rate": 3.1100939532642736e-05, "loss": 0.2257, "step": 25820 }, { "epoch": 3.11, "learning_rate": 3.111298482293423e-05, "loss": 0.2314, "step": 25830 }, { "epoch": 3.11, "learning_rate": 3.1125030113225726e-05, "loss": 0.2875, "step": 25840 }, { "epoch": 3.11, "learning_rate": 3.113707540351722e-05, "loss": 0.1578, "step": 25850 }, { "epoch": 3.11, "learning_rate": 3.114912069380872e-05, "loss": 0.2069, "step": 25860 }, { "epoch": 3.12, "learning_rate": 3.116116598410022e-05, "loss": 0.375, "step": 25870 }, { "epoch": 3.12, "learning_rate": 3.1173211274391714e-05, "loss": 0.1165, "step": 25880 }, { "epoch": 3.12, "learning_rate": 3.1185256564683216e-05, "loss": 0.2349, "step": 25890 }, { "epoch": 3.12, "learning_rate": 3.119730185497471e-05, "loss": 0.2425, "step": 25900 }, { "epoch": 3.12, "learning_rate": 3.1209347145266206e-05, "loss": 0.4854, "step": 25910 }, { "epoch": 3.12, "learning_rate": 3.12213924355577e-05, "loss": 0.272, "step": 25920 }, { "epoch": 3.12, "learning_rate": 3.1233437725849196e-05, "loss": 0.0181, "step": 25930 }, { "epoch": 3.12, "learning_rate": 3.124548301614069e-05, "loss": 0.3008, "step": 25940 }, { "epoch": 3.13, "learning_rate": 3.125752830643219e-05, "loss": 0.1838, "step": 25950 }, { "epoch": 3.13, "learning_rate": 3.126957359672368e-05, "loss": 0.0766, "step": 25960 }, { "epoch": 3.13, "learning_rate": 3.128161888701518e-05, "loss": 0.1983, "step": 25970 }, { "epoch": 3.13, "learning_rate": 3.129366417730667e-05, "loss": 0.4983, "step": 25980 }, { "epoch": 3.13, "learning_rate": 3.1305709467598174e-05, "loss": 0.4632, "step": 25990 }, { "epoch": 3.13, "learning_rate": 3.131775475788967e-05, "loss": 0.3192, "step": 26000 }, { "epoch": 3.13, "learning_rate": 3.1329800048181165e-05, "loss": 0.3183, "step": 26010 }, { "epoch": 3.13, "learning_rate": 3.134184533847266e-05, "loss": 0.3288, "step": 26020 }, { "epoch": 3.14, "learning_rate": 3.1353890628764155e-05, "loss": 0.084, "step": 26030 }, { "epoch": 3.14, "learning_rate": 3.136593591905565e-05, "loss": 0.3194, "step": 26040 }, { "epoch": 3.14, "learning_rate": 3.1377981209347145e-05, "loss": 0.2384, "step": 26050 }, { "epoch": 3.14, "learning_rate": 3.139002649963864e-05, "loss": 0.2473, "step": 26060 }, { "epoch": 3.14, "learning_rate": 3.1402071789930136e-05, "loss": 0.1829, "step": 26070 }, { "epoch": 3.14, "learning_rate": 3.141411708022163e-05, "loss": 0.2635, "step": 26080 }, { "epoch": 3.14, "learning_rate": 3.1426162370513126e-05, "loss": 0.3213, "step": 26090 }, { "epoch": 3.14, "learning_rate": 3.143820766080463e-05, "loss": 0.3113, "step": 26100 }, { "epoch": 3.14, "learning_rate": 3.145025295109612e-05, "loss": 0.459, "step": 26110 }, { "epoch": 3.15, "learning_rate": 3.146229824138762e-05, "loss": 0.2854, "step": 26120 }, { "epoch": 3.15, "learning_rate": 3.1474343531679114e-05, "loss": 0.4353, "step": 26130 }, { "epoch": 3.15, "learning_rate": 3.148638882197061e-05, "loss": 0.2821, "step": 26140 }, { "epoch": 3.15, "learning_rate": 3.149843411226211e-05, "loss": 0.3122, "step": 26150 }, { "epoch": 3.15, "learning_rate": 3.1510479402553606e-05, "loss": 0.2357, "step": 26160 }, { "epoch": 3.15, "learning_rate": 3.15225246928451e-05, "loss": 0.4204, "step": 26170 }, { "epoch": 3.15, "learning_rate": 3.1534569983136596e-05, "loss": 0.0572, "step": 26180 }, { "epoch": 3.15, "learning_rate": 3.154661527342809e-05, "loss": 0.1571, "step": 26190 }, { "epoch": 3.16, "learning_rate": 3.155866056371959e-05, "loss": 0.0867, "step": 26200 }, { "epoch": 3.16, "learning_rate": 3.157070585401108e-05, "loss": 0.3343, "step": 26210 }, { "epoch": 3.16, "learning_rate": 3.158275114430258e-05, "loss": 0.3134, "step": 26220 }, { "epoch": 3.16, "learning_rate": 3.159479643459408e-05, "loss": 0.1975, "step": 26230 }, { "epoch": 3.16, "learning_rate": 3.1606841724885574e-05, "loss": 0.1028, "step": 26240 }, { "epoch": 3.16, "learning_rate": 3.161888701517707e-05, "loss": 0.2928, "step": 26250 }, { "epoch": 3.16, "learning_rate": 3.1630932305468565e-05, "loss": 0.3859, "step": 26260 }, { "epoch": 3.16, "learning_rate": 3.164297759576006e-05, "loss": 0.3256, "step": 26270 }, { "epoch": 3.17, "learning_rate": 3.1655022886051555e-05, "loss": 0.2488, "step": 26280 }, { "epoch": 3.17, "learning_rate": 3.166706817634305e-05, "loss": 0.1517, "step": 26290 }, { "epoch": 3.17, "learning_rate": 3.1679113466634545e-05, "loss": 0.1971, "step": 26300 }, { "epoch": 3.17, "learning_rate": 3.169115875692604e-05, "loss": 0.2443, "step": 26310 }, { "epoch": 3.17, "learning_rate": 3.1703204047217536e-05, "loss": 0.3108, "step": 26320 }, { "epoch": 3.17, "learning_rate": 3.171524933750903e-05, "loss": 0.3149, "step": 26330 }, { "epoch": 3.17, "learning_rate": 3.172729462780053e-05, "loss": 0.2452, "step": 26340 }, { "epoch": 3.17, "learning_rate": 3.173933991809203e-05, "loss": 0.2322, "step": 26350 }, { "epoch": 3.17, "learning_rate": 3.175138520838352e-05, "loss": 0.2111, "step": 26360 }, { "epoch": 3.18, "learning_rate": 3.176343049867502e-05, "loss": 0.155, "step": 26370 }, { "epoch": 3.18, "learning_rate": 3.1775475788966514e-05, "loss": 0.2402, "step": 26380 }, { "epoch": 3.18, "learning_rate": 3.178752107925801e-05, "loss": 0.3156, "step": 26390 }, { "epoch": 3.18, "learning_rate": 3.1799566369549504e-05, "loss": 0.1624, "step": 26400 }, { "epoch": 3.18, "learning_rate": 3.1811611659841006e-05, "loss": 0.2086, "step": 26410 }, { "epoch": 3.18, "learning_rate": 3.18236569501325e-05, "loss": 0.134, "step": 26420 }, { "epoch": 3.18, "learning_rate": 3.1835702240423996e-05, "loss": 0.1561, "step": 26430 }, { "epoch": 3.18, "learning_rate": 3.184774753071549e-05, "loss": 0.1846, "step": 26440 }, { "epoch": 3.19, "learning_rate": 3.185979282100699e-05, "loss": 0.4856, "step": 26450 }, { "epoch": 3.19, "learning_rate": 3.187183811129849e-05, "loss": 0.2791, "step": 26460 }, { "epoch": 3.19, "learning_rate": 3.1883883401589984e-05, "loss": 0.3095, "step": 26470 }, { "epoch": 3.19, "learning_rate": 3.189592869188148e-05, "loss": 0.2822, "step": 26480 }, { "epoch": 3.19, "learning_rate": 3.1907973982172974e-05, "loss": 0.1731, "step": 26490 }, { "epoch": 3.19, "learning_rate": 3.192001927246447e-05, "loss": 0.2702, "step": 26500 }, { "epoch": 3.19, "learning_rate": 3.1932064562755965e-05, "loss": 0.1237, "step": 26510 }, { "epoch": 3.19, "learning_rate": 3.194410985304746e-05, "loss": 0.2848, "step": 26520 }, { "epoch": 3.2, "learning_rate": 3.1956155143338955e-05, "loss": 0.2309, "step": 26530 }, { "epoch": 3.2, "learning_rate": 3.196820043363045e-05, "loss": 0.4189, "step": 26540 }, { "epoch": 3.2, "learning_rate": 3.1980245723921945e-05, "loss": 0.323, "step": 26550 }, { "epoch": 3.2, "learning_rate": 3.199229101421344e-05, "loss": 0.3852, "step": 26560 }, { "epoch": 3.2, "learning_rate": 3.200433630450494e-05, "loss": 0.0965, "step": 26570 }, { "epoch": 3.2, "learning_rate": 3.201638159479644e-05, "loss": 0.15, "step": 26580 }, { "epoch": 3.2, "learning_rate": 3.202842688508793e-05, "loss": 0.2596, "step": 26590 }, { "epoch": 3.2, "learning_rate": 3.204047217537943e-05, "loss": 0.2513, "step": 26600 }, { "epoch": 3.21, "learning_rate": 3.205251746567092e-05, "loss": 0.4017, "step": 26610 }, { "epoch": 3.21, "learning_rate": 3.206456275596242e-05, "loss": 0.2846, "step": 26620 }, { "epoch": 3.21, "learning_rate": 3.2076608046253914e-05, "loss": 0.1983, "step": 26630 }, { "epoch": 3.21, "learning_rate": 3.208865333654541e-05, "loss": 0.1891, "step": 26640 }, { "epoch": 3.21, "learning_rate": 3.2100698626836904e-05, "loss": 0.1694, "step": 26650 }, { "epoch": 3.21, "learning_rate": 3.21127439171284e-05, "loss": 0.2307, "step": 26660 }, { "epoch": 3.21, "learning_rate": 3.21247892074199e-05, "loss": 0.3475, "step": 26670 }, { "epoch": 3.21, "learning_rate": 3.2136834497711396e-05, "loss": 0.2109, "step": 26680 }, { "epoch": 3.21, "learning_rate": 3.214887978800289e-05, "loss": 0.1501, "step": 26690 }, { "epoch": 3.22, "learning_rate": 3.2160925078294393e-05, "loss": 0.3511, "step": 26700 }, { "epoch": 3.22, "learning_rate": 3.217297036858589e-05, "loss": 0.1537, "step": 26710 }, { "epoch": 3.22, "learning_rate": 3.2185015658877384e-05, "loss": 0.4354, "step": 26720 }, { "epoch": 3.22, "learning_rate": 3.219706094916888e-05, "loss": 0.2124, "step": 26730 }, { "epoch": 3.22, "learning_rate": 3.2209106239460374e-05, "loss": 0.3478, "step": 26740 }, { "epoch": 3.22, "learning_rate": 3.222115152975187e-05, "loss": 0.3381, "step": 26750 }, { "epoch": 3.22, "learning_rate": 3.2233196820043365e-05, "loss": 0.0724, "step": 26760 }, { "epoch": 3.22, "learning_rate": 3.224524211033486e-05, "loss": 0.317, "step": 26770 }, { "epoch": 3.23, "learning_rate": 3.2257287400626355e-05, "loss": 0.5549, "step": 26780 }, { "epoch": 3.23, "learning_rate": 3.226933269091785e-05, "loss": 0.2279, "step": 26790 }, { "epoch": 3.23, "learning_rate": 3.2281377981209345e-05, "loss": 0.2565, "step": 26800 }, { "epoch": 3.23, "learning_rate": 3.229342327150085e-05, "loss": 0.1639, "step": 26810 }, { "epoch": 3.23, "learning_rate": 3.230546856179234e-05, "loss": 0.2193, "step": 26820 }, { "epoch": 3.23, "learning_rate": 3.231751385208384e-05, "loss": 0.5281, "step": 26830 }, { "epoch": 3.23, "learning_rate": 3.232955914237533e-05, "loss": 0.2905, "step": 26840 }, { "epoch": 3.23, "learning_rate": 3.234160443266683e-05, "loss": 0.2982, "step": 26850 }, { "epoch": 3.24, "learning_rate": 3.235364972295832e-05, "loss": 0.4296, "step": 26860 }, { "epoch": 3.24, "learning_rate": 3.236569501324982e-05, "loss": 0.2324, "step": 26870 }, { "epoch": 3.24, "learning_rate": 3.2377740303541314e-05, "loss": 0.2498, "step": 26880 }, { "epoch": 3.24, "learning_rate": 3.238978559383281e-05, "loss": 0.4992, "step": 26890 }, { "epoch": 3.24, "learning_rate": 3.2401830884124304e-05, "loss": 0.2781, "step": 26900 }, { "epoch": 3.24, "learning_rate": 3.2413876174415806e-05, "loss": 0.6001, "step": 26910 }, { "epoch": 3.24, "learning_rate": 3.24259214647073e-05, "loss": 0.1741, "step": 26920 }, { "epoch": 3.24, "learning_rate": 3.24379667549988e-05, "loss": 0.3282, "step": 26930 }, { "epoch": 3.24, "learning_rate": 3.24500120452903e-05, "loss": 0.2494, "step": 26940 }, { "epoch": 3.25, "learning_rate": 3.2462057335581793e-05, "loss": 0.3007, "step": 26950 }, { "epoch": 3.25, "learning_rate": 3.247410262587329e-05, "loss": 0.2339, "step": 26960 }, { "epoch": 3.25, "learning_rate": 3.2486147916164784e-05, "loss": 0.5343, "step": 26970 }, { "epoch": 3.25, "learning_rate": 3.249819320645628e-05, "loss": 0.3372, "step": 26980 }, { "epoch": 3.25, "learning_rate": 3.2510238496747774e-05, "loss": 0.333, "step": 26990 }, { "epoch": 3.25, "learning_rate": 3.252228378703927e-05, "loss": 0.1817, "step": 27000 }, { "epoch": 3.25, "learning_rate": 3.2534329077330765e-05, "loss": 0.2449, "step": 27010 }, { "epoch": 3.25, "learning_rate": 3.254637436762226e-05, "loss": 0.196, "step": 27020 }, { "epoch": 3.26, "learning_rate": 3.2558419657913755e-05, "loss": 0.1557, "step": 27030 }, { "epoch": 3.26, "learning_rate": 3.257046494820526e-05, "loss": 0.2312, "step": 27040 }, { "epoch": 3.26, "learning_rate": 3.258251023849675e-05, "loss": 0.2446, "step": 27050 }, { "epoch": 3.26, "learning_rate": 3.259455552878825e-05, "loss": 0.3231, "step": 27060 }, { "epoch": 3.26, "learning_rate": 3.260660081907974e-05, "loss": 0.3948, "step": 27070 }, { "epoch": 3.26, "learning_rate": 3.261864610937124e-05, "loss": 0.195, "step": 27080 }, { "epoch": 3.26, "learning_rate": 3.263069139966273e-05, "loss": 0.1357, "step": 27090 }, { "epoch": 3.26, "learning_rate": 3.264273668995423e-05, "loss": 0.3025, "step": 27100 }, { "epoch": 3.27, "learning_rate": 3.265478198024572e-05, "loss": 0.1533, "step": 27110 }, { "epoch": 3.27, "learning_rate": 3.266682727053722e-05, "loss": 0.1883, "step": 27120 }, { "epoch": 3.27, "learning_rate": 3.2678872560828714e-05, "loss": 0.4334, "step": 27130 }, { "epoch": 3.27, "learning_rate": 3.269091785112021e-05, "loss": 0.3685, "step": 27140 }, { "epoch": 3.27, "learning_rate": 3.270296314141171e-05, "loss": 0.2211, "step": 27150 }, { "epoch": 3.27, "learning_rate": 3.2715008431703206e-05, "loss": 0.3349, "step": 27160 }, { "epoch": 3.27, "learning_rate": 3.27270537219947e-05, "loss": 0.2777, "step": 27170 }, { "epoch": 3.27, "learning_rate": 3.2739099012286196e-05, "loss": 0.2564, "step": 27180 }, { "epoch": 3.27, "learning_rate": 3.275114430257769e-05, "loss": 0.1833, "step": 27190 }, { "epoch": 3.28, "learning_rate": 3.276318959286919e-05, "loss": 0.2043, "step": 27200 }, { "epoch": 3.28, "learning_rate": 3.277523488316069e-05, "loss": 0.1586, "step": 27210 }, { "epoch": 3.28, "learning_rate": 3.2787280173452184e-05, "loss": 0.2566, "step": 27220 }, { "epoch": 3.28, "learning_rate": 3.279932546374368e-05, "loss": 0.3144, "step": 27230 }, { "epoch": 3.28, "learning_rate": 3.2811370754035174e-05, "loss": 0.3006, "step": 27240 }, { "epoch": 3.28, "learning_rate": 3.282341604432667e-05, "loss": 0.1209, "step": 27250 }, { "epoch": 3.28, "learning_rate": 3.2835461334618164e-05, "loss": 0.2154, "step": 27260 }, { "epoch": 3.28, "learning_rate": 3.284750662490966e-05, "loss": 0.2126, "step": 27270 }, { "epoch": 3.29, "learning_rate": 3.285955191520116e-05, "loss": 0.2153, "step": 27280 }, { "epoch": 3.29, "learning_rate": 3.287159720549266e-05, "loss": 0.3973, "step": 27290 }, { "epoch": 3.29, "learning_rate": 3.288364249578415e-05, "loss": 0.2594, "step": 27300 }, { "epoch": 3.29, "learning_rate": 3.289568778607565e-05, "loss": 0.3626, "step": 27310 }, { "epoch": 3.29, "learning_rate": 3.290773307636714e-05, "loss": 0.2522, "step": 27320 }, { "epoch": 3.29, "learning_rate": 3.291977836665864e-05, "loss": 0.2456, "step": 27330 }, { "epoch": 3.29, "learning_rate": 3.293182365695013e-05, "loss": 0.3565, "step": 27340 }, { "epoch": 3.29, "learning_rate": 3.294386894724163e-05, "loss": 0.1523, "step": 27350 }, { "epoch": 3.3, "learning_rate": 3.295591423753312e-05, "loss": 0.1415, "step": 27360 }, { "epoch": 3.3, "learning_rate": 3.296795952782462e-05, "loss": 0.195, "step": 27370 }, { "epoch": 3.3, "learning_rate": 3.298000481811612e-05, "loss": 0.4392, "step": 27380 }, { "epoch": 3.3, "learning_rate": 3.2992050108407615e-05, "loss": 0.2469, "step": 27390 }, { "epoch": 3.3, "learning_rate": 3.300409539869911e-05, "loss": 0.2366, "step": 27400 }, { "epoch": 3.3, "learning_rate": 3.3016140688990606e-05, "loss": 0.2001, "step": 27410 }, { "epoch": 3.3, "learning_rate": 3.30281859792821e-05, "loss": 0.3769, "step": 27420 }, { "epoch": 3.3, "learning_rate": 3.3040231269573596e-05, "loss": 0.2706, "step": 27430 }, { "epoch": 3.31, "learning_rate": 3.305227655986509e-05, "loss": 0.1299, "step": 27440 }, { "epoch": 3.31, "learning_rate": 3.3064321850156587e-05, "loss": 0.3105, "step": 27450 }, { "epoch": 3.31, "learning_rate": 3.307636714044809e-05, "loss": 0.3287, "step": 27460 }, { "epoch": 3.31, "learning_rate": 3.3088412430739584e-05, "loss": 0.2702, "step": 27470 }, { "epoch": 3.31, "learning_rate": 3.310045772103108e-05, "loss": 0.3644, "step": 27480 }, { "epoch": 3.31, "learning_rate": 3.3112503011322574e-05, "loss": 0.1838, "step": 27490 }, { "epoch": 3.31, "learning_rate": 3.312454830161407e-05, "loss": 0.2314, "step": 27500 }, { "epoch": 3.31, "learning_rate": 3.313659359190557e-05, "loss": 0.2208, "step": 27510 }, { "epoch": 3.31, "learning_rate": 3.3148638882197066e-05, "loss": 0.2811, "step": 27520 }, { "epoch": 3.32, "learning_rate": 3.316068417248856e-05, "loss": 0.3014, "step": 27530 }, { "epoch": 3.32, "learning_rate": 3.317272946278006e-05, "loss": 0.2912, "step": 27540 }, { "epoch": 3.32, "learning_rate": 3.318477475307155e-05, "loss": 0.4619, "step": 27550 }, { "epoch": 3.32, "learning_rate": 3.319682004336305e-05, "loss": 0.3145, "step": 27560 }, { "epoch": 3.32, "learning_rate": 3.320886533365454e-05, "loss": 0.3264, "step": 27570 }, { "epoch": 3.32, "learning_rate": 3.322091062394604e-05, "loss": 0.1826, "step": 27580 }, { "epoch": 3.32, "learning_rate": 3.323295591423753e-05, "loss": 0.2336, "step": 27590 }, { "epoch": 3.32, "learning_rate": 3.324500120452903e-05, "loss": 0.1922, "step": 27600 }, { "epoch": 3.33, "learning_rate": 3.325704649482052e-05, "loss": 0.2336, "step": 27610 }, { "epoch": 3.33, "learning_rate": 3.3269091785112025e-05, "loss": 0.1899, "step": 27620 }, { "epoch": 3.33, "learning_rate": 3.328113707540352e-05, "loss": 0.3204, "step": 27630 }, { "epoch": 3.33, "learning_rate": 3.3293182365695015e-05, "loss": 0.3708, "step": 27640 }, { "epoch": 3.33, "learning_rate": 3.330522765598651e-05, "loss": 0.2505, "step": 27650 }, { "epoch": 3.33, "learning_rate": 3.3317272946278006e-05, "loss": 0.2448, "step": 27660 }, { "epoch": 3.33, "learning_rate": 3.33293182365695e-05, "loss": 0.3365, "step": 27670 }, { "epoch": 3.33, "learning_rate": 3.3341363526860996e-05, "loss": 0.5215, "step": 27680 }, { "epoch": 3.34, "learning_rate": 3.335340881715249e-05, "loss": 0.1472, "step": 27690 }, { "epoch": 3.34, "learning_rate": 3.3365454107443986e-05, "loss": 0.1775, "step": 27700 }, { "epoch": 3.34, "learning_rate": 3.337749939773548e-05, "loss": 0.2579, "step": 27710 }, { "epoch": 3.34, "learning_rate": 3.3389544688026984e-05, "loss": 0.1481, "step": 27720 }, { "epoch": 3.34, "learning_rate": 3.340158997831848e-05, "loss": 0.3572, "step": 27730 }, { "epoch": 3.34, "learning_rate": 3.3413635268609974e-05, "loss": 0.2743, "step": 27740 }, { "epoch": 3.34, "learning_rate": 3.3425680558901476e-05, "loss": 0.1853, "step": 27750 }, { "epoch": 3.34, "learning_rate": 3.343772584919297e-05, "loss": 0.163, "step": 27760 }, { "epoch": 3.34, "learning_rate": 3.3449771139484466e-05, "loss": 0.2779, "step": 27770 }, { "epoch": 3.35, "learning_rate": 3.346181642977596e-05, "loss": 0.2821, "step": 27780 }, { "epoch": 3.35, "learning_rate": 3.347386172006746e-05, "loss": 0.2313, "step": 27790 }, { "epoch": 3.35, "learning_rate": 3.348590701035895e-05, "loss": 0.079, "step": 27800 }, { "epoch": 3.35, "learning_rate": 3.349795230065045e-05, "loss": 0.1402, "step": 27810 }, { "epoch": 3.35, "learning_rate": 3.350999759094194e-05, "loss": 0.315, "step": 27820 }, { "epoch": 3.35, "learning_rate": 3.352204288123344e-05, "loss": 0.3868, "step": 27830 }, { "epoch": 3.35, "learning_rate": 3.353408817152493e-05, "loss": 0.2866, "step": 27840 }, { "epoch": 3.35, "learning_rate": 3.3546133461816435e-05, "loss": 0.2879, "step": 27850 }, { "epoch": 3.36, "learning_rate": 3.355817875210793e-05, "loss": 0.3291, "step": 27860 }, { "epoch": 3.36, "learning_rate": 3.3570224042399425e-05, "loss": 0.3301, "step": 27870 }, { "epoch": 3.36, "learning_rate": 3.358226933269092e-05, "loss": 0.2164, "step": 27880 }, { "epoch": 3.36, "learning_rate": 3.3594314622982415e-05, "loss": 0.3495, "step": 27890 }, { "epoch": 3.36, "learning_rate": 3.360635991327391e-05, "loss": 0.2831, "step": 27900 }, { "epoch": 3.36, "learning_rate": 3.3618405203565406e-05, "loss": 0.2365, "step": 27910 }, { "epoch": 3.36, "learning_rate": 3.36304504938569e-05, "loss": 0.2632, "step": 27920 }, { "epoch": 3.36, "learning_rate": 3.3642495784148396e-05, "loss": 0.2833, "step": 27930 }, { "epoch": 3.37, "learning_rate": 3.365454107443989e-05, "loss": 0.4187, "step": 27940 }, { "epoch": 3.37, "learning_rate": 3.3666586364731386e-05, "loss": 0.1767, "step": 27950 }, { "epoch": 3.37, "learning_rate": 3.367863165502289e-05, "loss": 0.2488, "step": 27960 }, { "epoch": 3.37, "learning_rate": 3.3690676945314384e-05, "loss": 0.3396, "step": 27970 }, { "epoch": 3.37, "learning_rate": 3.3702722235605886e-05, "loss": 0.2273, "step": 27980 }, { "epoch": 3.37, "learning_rate": 3.371476752589738e-05, "loss": 0.2951, "step": 27990 }, { "epoch": 3.37, "learning_rate": 3.3726812816188876e-05, "loss": 0.1065, "step": 28000 }, { "epoch": 3.37, "learning_rate": 3.373885810648037e-05, "loss": 0.3347, "step": 28010 }, { "epoch": 3.37, "learning_rate": 3.3750903396771866e-05, "loss": 0.2958, "step": 28020 }, { "epoch": 3.38, "learning_rate": 3.376294868706336e-05, "loss": 0.4251, "step": 28030 }, { "epoch": 3.38, "learning_rate": 3.377499397735486e-05, "loss": 0.3955, "step": 28040 }, { "epoch": 3.38, "learning_rate": 3.378703926764635e-05, "loss": 0.1451, "step": 28050 }, { "epoch": 3.38, "learning_rate": 3.379908455793785e-05, "loss": 0.3612, "step": 28060 }, { "epoch": 3.38, "learning_rate": 3.381112984822934e-05, "loss": 0.212, "step": 28070 }, { "epoch": 3.38, "learning_rate": 3.382317513852084e-05, "loss": 0.3269, "step": 28080 }, { "epoch": 3.38, "learning_rate": 3.383522042881234e-05, "loss": 0.0841, "step": 28090 }, { "epoch": 3.38, "learning_rate": 3.3847265719103835e-05, "loss": 0.4192, "step": 28100 }, { "epoch": 3.39, "learning_rate": 3.385931100939533e-05, "loss": 0.2288, "step": 28110 }, { "epoch": 3.39, "learning_rate": 3.3871356299686825e-05, "loss": 0.2554, "step": 28120 }, { "epoch": 3.39, "learning_rate": 3.388340158997832e-05, "loss": 0.2873, "step": 28130 }, { "epoch": 3.39, "learning_rate": 3.3895446880269815e-05, "loss": 0.4357, "step": 28140 }, { "epoch": 3.39, "learning_rate": 3.390749217056131e-05, "loss": 0.4127, "step": 28150 }, { "epoch": 3.39, "learning_rate": 3.3919537460852806e-05, "loss": 0.17, "step": 28160 }, { "epoch": 3.39, "learning_rate": 3.39315827511443e-05, "loss": 0.1578, "step": 28170 }, { "epoch": 3.39, "learning_rate": 3.3943628041435796e-05, "loss": 0.4238, "step": 28180 }, { "epoch": 3.4, "learning_rate": 3.395567333172729e-05, "loss": 0.3858, "step": 28190 }, { "epoch": 3.4, "learning_rate": 3.396771862201879e-05, "loss": 0.2309, "step": 28200 }, { "epoch": 3.4, "learning_rate": 3.397976391231029e-05, "loss": 0.3369, "step": 28210 }, { "epoch": 3.4, "learning_rate": 3.3991809202601784e-05, "loss": 0.2925, "step": 28220 }, { "epoch": 3.4, "learning_rate": 3.400385449289328e-05, "loss": 0.3534, "step": 28230 }, { "epoch": 3.4, "learning_rate": 3.4015899783184774e-05, "loss": 0.1134, "step": 28240 }, { "epoch": 3.4, "learning_rate": 3.4027945073476276e-05, "loss": 0.1432, "step": 28250 }, { "epoch": 3.4, "learning_rate": 3.403999036376777e-05, "loss": 0.2639, "step": 28260 }, { "epoch": 3.4, "learning_rate": 3.4052035654059266e-05, "loss": 0.3689, "step": 28270 }, { "epoch": 3.41, "learning_rate": 3.406408094435076e-05, "loss": 0.2787, "step": 28280 }, { "epoch": 3.41, "learning_rate": 3.407612623464226e-05, "loss": 0.2192, "step": 28290 }, { "epoch": 3.41, "learning_rate": 3.408817152493375e-05, "loss": 0.4711, "step": 28300 }, { "epoch": 3.41, "learning_rate": 3.410021681522525e-05, "loss": 0.2256, "step": 28310 }, { "epoch": 3.41, "learning_rate": 3.411226210551675e-05, "loss": 0.2276, "step": 28320 }, { "epoch": 3.41, "learning_rate": 3.4124307395808244e-05, "loss": 0.3108, "step": 28330 }, { "epoch": 3.41, "learning_rate": 3.413635268609974e-05, "loss": 0.248, "step": 28340 }, { "epoch": 3.41, "learning_rate": 3.4148397976391235e-05, "loss": 0.2026, "step": 28350 }, { "epoch": 3.42, "learning_rate": 3.416044326668273e-05, "loss": 0.3229, "step": 28360 }, { "epoch": 3.42, "learning_rate": 3.4172488556974225e-05, "loss": 0.1863, "step": 28370 }, { "epoch": 3.42, "learning_rate": 3.418453384726572e-05, "loss": 0.1905, "step": 28380 }, { "epoch": 3.42, "learning_rate": 3.4196579137557215e-05, "loss": 0.1956, "step": 28390 }, { "epoch": 3.42, "learning_rate": 3.420862442784871e-05, "loss": 0.2965, "step": 28400 }, { "epoch": 3.42, "learning_rate": 3.4220669718140206e-05, "loss": 0.2003, "step": 28410 }, { "epoch": 3.42, "learning_rate": 3.42327150084317e-05, "loss": 0.246, "step": 28420 }, { "epoch": 3.42, "learning_rate": 3.42447602987232e-05, "loss": 0.2369, "step": 28430 }, { "epoch": 3.43, "learning_rate": 3.42568055890147e-05, "loss": 0.1804, "step": 28440 }, { "epoch": 3.43, "learning_rate": 3.426885087930619e-05, "loss": 0.2908, "step": 28450 }, { "epoch": 3.43, "learning_rate": 3.428089616959769e-05, "loss": 0.2236, "step": 28460 }, { "epoch": 3.43, "learning_rate": 3.4292941459889184e-05, "loss": 0.2773, "step": 28470 }, { "epoch": 3.43, "learning_rate": 3.430498675018068e-05, "loss": 0.4637, "step": 28480 }, { "epoch": 3.43, "learning_rate": 3.4317032040472174e-05, "loss": 0.2316, "step": 28490 }, { "epoch": 3.43, "learning_rate": 3.432907733076367e-05, "loss": 0.2739, "step": 28500 }, { "epoch": 3.43, "learning_rate": 3.434112262105517e-05, "loss": 0.3636, "step": 28510 }, { "epoch": 3.44, "learning_rate": 3.4353167911346666e-05, "loss": 0.2554, "step": 28520 }, { "epoch": 3.44, "learning_rate": 3.436521320163816e-05, "loss": 0.2611, "step": 28530 }, { "epoch": 3.44, "learning_rate": 3.4377258491929657e-05, "loss": 0.1838, "step": 28540 }, { "epoch": 3.44, "learning_rate": 3.438930378222115e-05, "loss": 0.2369, "step": 28550 }, { "epoch": 3.44, "learning_rate": 3.4401349072512654e-05, "loss": 0.1403, "step": 28560 }, { "epoch": 3.44, "learning_rate": 3.441339436280415e-05, "loss": 0.1892, "step": 28570 }, { "epoch": 3.44, "learning_rate": 3.4425439653095644e-05, "loss": 0.1982, "step": 28580 }, { "epoch": 3.44, "learning_rate": 3.443748494338714e-05, "loss": 0.1611, "step": 28590 }, { "epoch": 3.44, "learning_rate": 3.4449530233678634e-05, "loss": 0.1947, "step": 28600 }, { "epoch": 3.45, "learning_rate": 3.446157552397013e-05, "loss": 0.3007, "step": 28610 }, { "epoch": 3.45, "learning_rate": 3.4473620814261625e-05, "loss": 0.4272, "step": 28620 }, { "epoch": 3.45, "learning_rate": 3.448566610455312e-05, "loss": 0.2701, "step": 28630 }, { "epoch": 3.45, "learning_rate": 3.4497711394844615e-05, "loss": 0.3036, "step": 28640 }, { "epoch": 3.45, "learning_rate": 3.450975668513611e-05, "loss": 0.5418, "step": 28650 }, { "epoch": 3.45, "learning_rate": 3.452180197542761e-05, "loss": 0.2001, "step": 28660 }, { "epoch": 3.45, "learning_rate": 3.453384726571911e-05, "loss": 0.3192, "step": 28670 }, { "epoch": 3.45, "learning_rate": 3.45458925560106e-05, "loss": 0.2101, "step": 28680 }, { "epoch": 3.46, "learning_rate": 3.45579378463021e-05, "loss": 0.1474, "step": 28690 }, { "epoch": 3.46, "learning_rate": 3.456998313659359e-05, "loss": 0.2524, "step": 28700 }, { "epoch": 3.46, "learning_rate": 3.458202842688509e-05, "loss": 0.161, "step": 28710 }, { "epoch": 3.46, "learning_rate": 3.4594073717176583e-05, "loss": 0.2108, "step": 28720 }, { "epoch": 3.46, "learning_rate": 3.460611900746808e-05, "loss": 0.1493, "step": 28730 }, { "epoch": 3.46, "learning_rate": 3.4618164297759574e-05, "loss": 0.4295, "step": 28740 }, { "epoch": 3.46, "learning_rate": 3.463020958805107e-05, "loss": 0.4098, "step": 28750 }, { "epoch": 3.46, "learning_rate": 3.4642254878342564e-05, "loss": 0.1491, "step": 28760 }, { "epoch": 3.47, "learning_rate": 3.4654300168634066e-05, "loss": 0.2773, "step": 28770 }, { "epoch": 3.47, "learning_rate": 3.466634545892556e-05, "loss": 0.3226, "step": 28780 }, { "epoch": 3.47, "learning_rate": 3.467839074921706e-05, "loss": 0.3866, "step": 28790 }, { "epoch": 3.47, "learning_rate": 3.469043603950856e-05, "loss": 0.2606, "step": 28800 }, { "epoch": 3.47, "learning_rate": 3.4702481329800054e-05, "loss": 0.2448, "step": 28810 }, { "epoch": 3.47, "learning_rate": 3.471452662009155e-05, "loss": 0.5117, "step": 28820 }, { "epoch": 3.47, "learning_rate": 3.4726571910383044e-05, "loss": 0.2296, "step": 28830 }, { "epoch": 3.47, "learning_rate": 3.473861720067454e-05, "loss": 0.221, "step": 28840 }, { "epoch": 3.47, "learning_rate": 3.4750662490966034e-05, "loss": 0.1589, "step": 28850 }, { "epoch": 3.48, "learning_rate": 3.476270778125753e-05, "loss": 0.1813, "step": 28860 }, { "epoch": 3.48, "learning_rate": 3.4774753071549025e-05, "loss": 0.2542, "step": 28870 }, { "epoch": 3.48, "learning_rate": 3.478679836184052e-05, "loss": 0.3602, "step": 28880 }, { "epoch": 3.48, "learning_rate": 3.4798843652132015e-05, "loss": 0.5525, "step": 28890 }, { "epoch": 3.48, "learning_rate": 3.481088894242352e-05, "loss": 0.3918, "step": 28900 }, { "epoch": 3.48, "learning_rate": 3.482293423271501e-05, "loss": 0.426, "step": 28910 }, { "epoch": 3.48, "learning_rate": 3.483497952300651e-05, "loss": 0.1926, "step": 28920 }, { "epoch": 3.48, "learning_rate": 3.4847024813298e-05, "loss": 0.2345, "step": 28930 }, { "epoch": 3.49, "learning_rate": 3.48590701035895e-05, "loss": 0.409, "step": 28940 }, { "epoch": 3.49, "learning_rate": 3.487111539388099e-05, "loss": 0.4043, "step": 28950 }, { "epoch": 3.49, "learning_rate": 3.488316068417249e-05, "loss": 0.0697, "step": 28960 }, { "epoch": 3.49, "learning_rate": 3.4895205974463983e-05, "loss": 0.3538, "step": 28970 }, { "epoch": 3.49, "learning_rate": 3.490725126475548e-05, "loss": 0.3459, "step": 28980 }, { "epoch": 3.49, "learning_rate": 3.4919296555046974e-05, "loss": 0.3013, "step": 28990 }, { "epoch": 3.49, "learning_rate": 3.493134184533847e-05, "loss": 0.3845, "step": 29000 }, { "epoch": 3.49, "learning_rate": 3.494338713562997e-05, "loss": 0.2159, "step": 29010 }, { "epoch": 3.5, "learning_rate": 3.4955432425921466e-05, "loss": 0.1297, "step": 29020 }, { "epoch": 3.5, "learning_rate": 3.496747771621296e-05, "loss": 0.0949, "step": 29030 }, { "epoch": 3.5, "learning_rate": 3.497952300650446e-05, "loss": 0.1753, "step": 29040 }, { "epoch": 3.5, "learning_rate": 3.499156829679596e-05, "loss": 0.1532, "step": 29050 }, { "epoch": 3.5, "learning_rate": 3.5003613587087454e-05, "loss": 0.2305, "step": 29060 }, { "epoch": 3.5, "learning_rate": 3.501565887737895e-05, "loss": 0.2291, "step": 29070 }, { "epoch": 3.5, "learning_rate": 3.5027704167670444e-05, "loss": 0.2113, "step": 29080 }, { "epoch": 3.5, "learning_rate": 3.503974945796194e-05, "loss": 0.377, "step": 29090 }, { "epoch": 3.5, "learning_rate": 3.5051794748253434e-05, "loss": 0.2532, "step": 29100 }, { "epoch": 3.51, "learning_rate": 3.506384003854493e-05, "loss": 0.1015, "step": 29110 }, { "epoch": 3.51, "learning_rate": 3.5075885328836425e-05, "loss": 0.2461, "step": 29120 }, { "epoch": 3.51, "learning_rate": 3.508793061912793e-05, "loss": 0.2566, "step": 29130 }, { "epoch": 3.51, "learning_rate": 3.509997590941942e-05, "loss": 0.3046, "step": 29140 }, { "epoch": 3.51, "learning_rate": 3.511202119971092e-05, "loss": 0.2855, "step": 29150 }, { "epoch": 3.51, "learning_rate": 3.512406649000241e-05, "loss": 0.4425, "step": 29160 }, { "epoch": 3.51, "learning_rate": 3.513611178029391e-05, "loss": 0.1984, "step": 29170 }, { "epoch": 3.51, "learning_rate": 3.51481570705854e-05, "loss": 0.4098, "step": 29180 }, { "epoch": 3.52, "learning_rate": 3.51602023608769e-05, "loss": 0.3805, "step": 29190 }, { "epoch": 3.52, "learning_rate": 3.517224765116839e-05, "loss": 0.158, "step": 29200 }, { "epoch": 3.52, "learning_rate": 3.518429294145989e-05, "loss": 0.3621, "step": 29210 }, { "epoch": 3.52, "learning_rate": 3.519633823175138e-05, "loss": 0.3789, "step": 29220 }, { "epoch": 3.52, "learning_rate": 3.520838352204288e-05, "loss": 0.3018, "step": 29230 }, { "epoch": 3.52, "learning_rate": 3.522042881233438e-05, "loss": 0.2386, "step": 29240 }, { "epoch": 3.52, "learning_rate": 3.5232474102625876e-05, "loss": 0.2217, "step": 29250 }, { "epoch": 3.52, "learning_rate": 3.524451939291737e-05, "loss": 0.0859, "step": 29260 }, { "epoch": 3.53, "learning_rate": 3.5256564683208866e-05, "loss": 0.2963, "step": 29270 }, { "epoch": 3.53, "learning_rate": 3.526860997350036e-05, "loss": 0.1147, "step": 29280 }, { "epoch": 3.53, "learning_rate": 3.5280655263791856e-05, "loss": 0.3466, "step": 29290 }, { "epoch": 3.53, "learning_rate": 3.529270055408336e-05, "loss": 0.2558, "step": 29300 }, { "epoch": 3.53, "learning_rate": 3.5304745844374854e-05, "loss": 0.1643, "step": 29310 }, { "epoch": 3.53, "learning_rate": 3.531679113466635e-05, "loss": 0.2388, "step": 29320 }, { "epoch": 3.53, "learning_rate": 3.5328836424957844e-05, "loss": 0.4024, "step": 29330 }, { "epoch": 3.53, "learning_rate": 3.534088171524934e-05, "loss": 0.1953, "step": 29340 }, { "epoch": 3.54, "learning_rate": 3.5352927005540834e-05, "loss": 0.2426, "step": 29350 }, { "epoch": 3.54, "learning_rate": 3.536497229583233e-05, "loss": 0.3099, "step": 29360 }, { "epoch": 3.54, "learning_rate": 3.537701758612383e-05, "loss": 0.1474, "step": 29370 }, { "epoch": 3.54, "learning_rate": 3.538906287641533e-05, "loss": 0.332, "step": 29380 }, { "epoch": 3.54, "learning_rate": 3.540110816670682e-05, "loss": 0.2334, "step": 29390 }, { "epoch": 3.54, "learning_rate": 3.541315345699832e-05, "loss": 0.2899, "step": 29400 }, { "epoch": 3.54, "learning_rate": 3.542519874728981e-05, "loss": 0.2405, "step": 29410 }, { "epoch": 3.54, "learning_rate": 3.543724403758131e-05, "loss": 0.0439, "step": 29420 }, { "epoch": 3.54, "learning_rate": 3.54492893278728e-05, "loss": 0.2572, "step": 29430 }, { "epoch": 3.55, "learning_rate": 3.54613346181643e-05, "loss": 0.2024, "step": 29440 }, { "epoch": 3.55, "learning_rate": 3.547337990845579e-05, "loss": 0.2211, "step": 29450 }, { "epoch": 3.55, "learning_rate": 3.548542519874729e-05, "loss": 0.1838, "step": 29460 }, { "epoch": 3.55, "learning_rate": 3.549747048903878e-05, "loss": 0.3312, "step": 29470 }, { "epoch": 3.55, "learning_rate": 3.5509515779330285e-05, "loss": 0.2681, "step": 29480 }, { "epoch": 3.55, "learning_rate": 3.552156106962178e-05, "loss": 0.3029, "step": 29490 }, { "epoch": 3.55, "learning_rate": 3.5533606359913276e-05, "loss": 0.1685, "step": 29500 }, { "epoch": 3.55, "learning_rate": 3.554565165020477e-05, "loss": 0.1895, "step": 29510 }, { "epoch": 3.56, "learning_rate": 3.5557696940496266e-05, "loss": 0.1539, "step": 29520 }, { "epoch": 3.56, "learning_rate": 3.556974223078776e-05, "loss": 0.2321, "step": 29530 }, { "epoch": 3.56, "learning_rate": 3.5581787521079256e-05, "loss": 0.396, "step": 29540 }, { "epoch": 3.56, "learning_rate": 3.559383281137075e-05, "loss": 0.1805, "step": 29550 }, { "epoch": 3.56, "learning_rate": 3.5605878101662254e-05, "loss": 0.0878, "step": 29560 }, { "epoch": 3.56, "learning_rate": 3.561792339195375e-05, "loss": 0.3098, "step": 29570 }, { "epoch": 3.56, "learning_rate": 3.5629968682245244e-05, "loss": 0.3556, "step": 29580 }, { "epoch": 3.56, "learning_rate": 3.564201397253674e-05, "loss": 0.5092, "step": 29590 }, { "epoch": 3.57, "learning_rate": 3.565405926282824e-05, "loss": 0.371, "step": 29600 }, { "epoch": 3.57, "learning_rate": 3.5666104553119736e-05, "loss": 0.278, "step": 29610 }, { "epoch": 3.57, "learning_rate": 3.567814984341123e-05, "loss": 0.2133, "step": 29620 }, { "epoch": 3.57, "learning_rate": 3.5690195133702727e-05, "loss": 0.136, "step": 29630 }, { "epoch": 3.57, "learning_rate": 3.570224042399422e-05, "loss": 0.2947, "step": 29640 }, { "epoch": 3.57, "learning_rate": 3.571428571428572e-05, "loss": 0.1043, "step": 29650 }, { "epoch": 3.57, "learning_rate": 3.572633100457721e-05, "loss": 0.3384, "step": 29660 }, { "epoch": 3.57, "learning_rate": 3.573837629486871e-05, "loss": 0.151, "step": 29670 }, { "epoch": 3.57, "learning_rate": 3.57504215851602e-05, "loss": 0.3201, "step": 29680 }, { "epoch": 3.58, "learning_rate": 3.57624668754517e-05, "loss": 0.3953, "step": 29690 }, { "epoch": 3.58, "learning_rate": 3.577451216574319e-05, "loss": 0.1301, "step": 29700 }, { "epoch": 3.58, "learning_rate": 3.5786557456034695e-05, "loss": 0.1684, "step": 29710 }, { "epoch": 3.58, "learning_rate": 3.579860274632619e-05, "loss": 0.1028, "step": 29720 }, { "epoch": 3.58, "learning_rate": 3.5810648036617685e-05, "loss": 0.1464, "step": 29730 }, { "epoch": 3.58, "learning_rate": 3.582269332690918e-05, "loss": 0.5429, "step": 29740 }, { "epoch": 3.58, "learning_rate": 3.5834738617200676e-05, "loss": 0.1933, "step": 29750 }, { "epoch": 3.58, "learning_rate": 3.584678390749217e-05, "loss": 0.3122, "step": 29760 }, { "epoch": 3.59, "learning_rate": 3.5858829197783666e-05, "loss": 0.2116, "step": 29770 }, { "epoch": 3.59, "learning_rate": 3.587087448807516e-05, "loss": 0.349, "step": 29780 }, { "epoch": 3.59, "learning_rate": 3.5882919778366656e-05, "loss": 0.3734, "step": 29790 }, { "epoch": 3.59, "learning_rate": 3.589496506865815e-05, "loss": 0.036, "step": 29800 }, { "epoch": 3.59, "learning_rate": 3.590701035894965e-05, "loss": 0.33, "step": 29810 }, { "epoch": 3.59, "learning_rate": 3.591905564924115e-05, "loss": 0.2557, "step": 29820 }, { "epoch": 3.59, "learning_rate": 3.5931100939532644e-05, "loss": 0.3023, "step": 29830 }, { "epoch": 3.59, "learning_rate": 3.5943146229824146e-05, "loss": 0.4156, "step": 29840 }, { "epoch": 3.6, "learning_rate": 3.595519152011564e-05, "loss": 0.332, "step": 29850 }, { "epoch": 3.6, "learning_rate": 3.5967236810407136e-05, "loss": 0.0274, "step": 29860 }, { "epoch": 3.6, "learning_rate": 3.597928210069863e-05, "loss": 0.1758, "step": 29870 }, { "epoch": 3.6, "learning_rate": 3.5991327390990127e-05, "loss": 0.143, "step": 29880 }, { "epoch": 3.6, "learning_rate": 3.600337268128162e-05, "loss": 0.1809, "step": 29890 }, { "epoch": 3.6, "learning_rate": 3.601541797157312e-05, "loss": 0.1966, "step": 29900 }, { "epoch": 3.6, "learning_rate": 3.602746326186461e-05, "loss": 0.2268, "step": 29910 }, { "epoch": 3.6, "learning_rate": 3.603950855215611e-05, "loss": 0.2262, "step": 29920 }, { "epoch": 3.6, "learning_rate": 3.60515538424476e-05, "loss": 0.2277, "step": 29930 }, { "epoch": 3.61, "learning_rate": 3.60635991327391e-05, "loss": 0.3233, "step": 29940 }, { "epoch": 3.61, "learning_rate": 3.60756444230306e-05, "loss": 0.2062, "step": 29950 }, { "epoch": 3.61, "learning_rate": 3.6087689713322095e-05, "loss": 0.1506, "step": 29960 }, { "epoch": 3.61, "learning_rate": 3.609973500361359e-05, "loss": 0.1795, "step": 29970 }, { "epoch": 3.61, "learning_rate": 3.6111780293905085e-05, "loss": 0.324, "step": 29980 }, { "epoch": 3.61, "learning_rate": 3.612382558419658e-05, "loss": 0.2705, "step": 29990 }, { "epoch": 3.61, "learning_rate": 3.6135870874488076e-05, "loss": 0.2606, "step": 30000 }, { "epoch": 3.61, "learning_rate": 3.614791616477957e-05, "loss": 0.2471, "step": 30010 }, { "epoch": 3.62, "learning_rate": 3.6159961455071066e-05, "loss": 0.275, "step": 30020 }, { "epoch": 3.62, "learning_rate": 3.617200674536256e-05, "loss": 0.2872, "step": 30030 }, { "epoch": 3.62, "learning_rate": 3.6184052035654056e-05, "loss": 0.1818, "step": 30040 }, { "epoch": 3.62, "learning_rate": 3.619609732594556e-05, "loss": 0.289, "step": 30050 }, { "epoch": 3.62, "learning_rate": 3.6208142616237053e-05, "loss": 0.3251, "step": 30060 }, { "epoch": 3.62, "learning_rate": 3.622018790652855e-05, "loss": 0.5788, "step": 30070 }, { "epoch": 3.62, "learning_rate": 3.6232233196820044e-05, "loss": 0.1556, "step": 30080 }, { "epoch": 3.62, "learning_rate": 3.6244278487111546e-05, "loss": 0.1435, "step": 30090 }, { "epoch": 3.63, "learning_rate": 3.625632377740304e-05, "loss": 0.4737, "step": 30100 }, { "epoch": 3.63, "learning_rate": 3.6268369067694536e-05, "loss": 0.2713, "step": 30110 }, { "epoch": 3.63, "learning_rate": 3.628041435798603e-05, "loss": 0.2977, "step": 30120 }, { "epoch": 3.63, "learning_rate": 3.6292459648277526e-05, "loss": 0.3495, "step": 30130 }, { "epoch": 3.63, "learning_rate": 3.630450493856902e-05, "loss": 0.3357, "step": 30140 }, { "epoch": 3.63, "learning_rate": 3.631655022886052e-05, "loss": 0.2426, "step": 30150 }, { "epoch": 3.63, "learning_rate": 3.632859551915201e-05, "loss": 0.2325, "step": 30160 }, { "epoch": 3.63, "learning_rate": 3.634064080944351e-05, "loss": 0.2328, "step": 30170 }, { "epoch": 3.64, "learning_rate": 3.635268609973501e-05, "loss": 0.2438, "step": 30180 }, { "epoch": 3.64, "learning_rate": 3.6364731390026504e-05, "loss": 0.2695, "step": 30190 }, { "epoch": 3.64, "learning_rate": 3.6376776680318e-05, "loss": 0.2466, "step": 30200 }, { "epoch": 3.64, "learning_rate": 3.6388821970609495e-05, "loss": 0.3575, "step": 30210 }, { "epoch": 3.64, "learning_rate": 3.640086726090099e-05, "loss": 0.2356, "step": 30220 }, { "epoch": 3.64, "learning_rate": 3.6412912551192485e-05, "loss": 0.2355, "step": 30230 }, { "epoch": 3.64, "learning_rate": 3.642495784148398e-05, "loss": 0.2285, "step": 30240 }, { "epoch": 3.64, "learning_rate": 3.6437003131775475e-05, "loss": 0.1745, "step": 30250 }, { "epoch": 3.64, "learning_rate": 3.644904842206697e-05, "loss": 0.3633, "step": 30260 }, { "epoch": 3.65, "learning_rate": 3.6461093712358466e-05, "loss": 0.0721, "step": 30270 }, { "epoch": 3.65, "learning_rate": 3.647313900264996e-05, "loss": 0.1836, "step": 30280 }, { "epoch": 3.65, "learning_rate": 3.648518429294146e-05, "loss": 0.2849, "step": 30290 }, { "epoch": 3.65, "learning_rate": 3.649722958323296e-05, "loss": 0.3982, "step": 30300 }, { "epoch": 3.65, "learning_rate": 3.650927487352445e-05, "loss": 0.3111, "step": 30310 }, { "epoch": 3.65, "learning_rate": 3.652132016381595e-05, "loss": 0.3179, "step": 30320 }, { "epoch": 3.65, "learning_rate": 3.6533365454107444e-05, "loss": 0.2384, "step": 30330 }, { "epoch": 3.65, "learning_rate": 3.654541074439894e-05, "loss": 0.1421, "step": 30340 }, { "epoch": 3.66, "learning_rate": 3.655745603469044e-05, "loss": 0.3572, "step": 30350 }, { "epoch": 3.66, "learning_rate": 3.6569501324981936e-05, "loss": 0.2889, "step": 30360 }, { "epoch": 3.66, "learning_rate": 3.658154661527343e-05, "loss": 0.4068, "step": 30370 }, { "epoch": 3.66, "learning_rate": 3.6593591905564926e-05, "loss": 0.2564, "step": 30380 }, { "epoch": 3.66, "learning_rate": 3.660563719585642e-05, "loss": 0.3677, "step": 30390 }, { "epoch": 3.66, "learning_rate": 3.661768248614792e-05, "loss": 0.2067, "step": 30400 }, { "epoch": 3.66, "learning_rate": 3.662972777643941e-05, "loss": 0.2491, "step": 30410 }, { "epoch": 3.66, "learning_rate": 3.6641773066730914e-05, "loss": 0.1803, "step": 30420 }, { "epoch": 3.67, "learning_rate": 3.665381835702241e-05, "loss": 0.301, "step": 30430 }, { "epoch": 3.67, "learning_rate": 3.6665863647313904e-05, "loss": 0.3228, "step": 30440 }, { "epoch": 3.67, "learning_rate": 3.66779089376054e-05, "loss": 0.3324, "step": 30450 }, { "epoch": 3.67, "learning_rate": 3.6689954227896895e-05, "loss": 0.191, "step": 30460 }, { "epoch": 3.67, "learning_rate": 3.670199951818839e-05, "loss": 0.1301, "step": 30470 }, { "epoch": 3.67, "learning_rate": 3.6714044808479885e-05, "loss": 0.2735, "step": 30480 }, { "epoch": 3.67, "learning_rate": 3.672609009877138e-05, "loss": 0.1805, "step": 30490 }, { "epoch": 3.67, "learning_rate": 3.6738135389062875e-05, "loss": 0.2997, "step": 30500 }, { "epoch": 3.67, "learning_rate": 3.675018067935437e-05, "loss": 0.1745, "step": 30510 }, { "epoch": 3.68, "learning_rate": 3.676222596964587e-05, "loss": 0.3216, "step": 30520 }, { "epoch": 3.68, "learning_rate": 3.677427125993737e-05, "loss": 0.3147, "step": 30530 }, { "epoch": 3.68, "learning_rate": 3.678631655022886e-05, "loss": 0.241, "step": 30540 }, { "epoch": 3.68, "learning_rate": 3.679836184052036e-05, "loss": 0.1335, "step": 30550 }, { "epoch": 3.68, "learning_rate": 3.681040713081185e-05, "loss": 0.1245, "step": 30560 }, { "epoch": 3.68, "learning_rate": 3.682245242110335e-05, "loss": 0.2514, "step": 30570 }, { "epoch": 3.68, "learning_rate": 3.6834497711394844e-05, "loss": 0.269, "step": 30580 }, { "epoch": 3.68, "learning_rate": 3.684654300168634e-05, "loss": 0.2751, "step": 30590 }, { "epoch": 3.69, "learning_rate": 3.6858588291977834e-05, "loss": 0.2758, "step": 30600 }, { "epoch": 3.69, "learning_rate": 3.6870633582269336e-05, "loss": 0.2884, "step": 30610 }, { "epoch": 3.69, "learning_rate": 3.688267887256083e-05, "loss": 0.4988, "step": 30620 }, { "epoch": 3.69, "learning_rate": 3.6894724162852326e-05, "loss": 0.3114, "step": 30630 }, { "epoch": 3.69, "learning_rate": 3.690676945314382e-05, "loss": 0.1234, "step": 30640 }, { "epoch": 3.69, "learning_rate": 3.6918814743435324e-05, "loss": 0.3002, "step": 30650 }, { "epoch": 3.69, "learning_rate": 3.693086003372682e-05, "loss": 0.2148, "step": 30660 }, { "epoch": 3.69, "learning_rate": 3.6942905324018314e-05, "loss": 0.3237, "step": 30670 }, { "epoch": 3.7, "learning_rate": 3.695495061430981e-05, "loss": 0.1988, "step": 30680 }, { "epoch": 3.7, "learning_rate": 3.6966995904601304e-05, "loss": 0.3415, "step": 30690 }, { "epoch": 3.7, "learning_rate": 3.69790411948928e-05, "loss": 0.1558, "step": 30700 }, { "epoch": 3.7, "learning_rate": 3.6991086485184295e-05, "loss": 0.2451, "step": 30710 }, { "epoch": 3.7, "learning_rate": 3.700313177547579e-05, "loss": 0.3001, "step": 30720 }, { "epoch": 3.7, "learning_rate": 3.7015177065767285e-05, "loss": 0.3589, "step": 30730 }, { "epoch": 3.7, "learning_rate": 3.702722235605878e-05, "loss": 0.262, "step": 30740 }, { "epoch": 3.7, "learning_rate": 3.7039267646350275e-05, "loss": 0.21, "step": 30750 }, { "epoch": 3.7, "learning_rate": 3.705131293664178e-05, "loss": 0.2191, "step": 30760 }, { "epoch": 3.71, "learning_rate": 3.706335822693327e-05, "loss": 0.2786, "step": 30770 }, { "epoch": 3.71, "learning_rate": 3.707540351722477e-05, "loss": 0.3213, "step": 30780 }, { "epoch": 3.71, "learning_rate": 3.708744880751626e-05, "loss": 0.4135, "step": 30790 }, { "epoch": 3.71, "learning_rate": 3.709949409780776e-05, "loss": 0.1375, "step": 30800 }, { "epoch": 3.71, "learning_rate": 3.711153938809925e-05, "loss": 0.156, "step": 30810 }, { "epoch": 3.71, "learning_rate": 3.712358467839075e-05, "loss": 0.2564, "step": 30820 }, { "epoch": 3.71, "learning_rate": 3.7135629968682244e-05, "loss": 0.1209, "step": 30830 }, { "epoch": 3.71, "learning_rate": 3.714767525897374e-05, "loss": 0.4383, "step": 30840 }, { "epoch": 3.72, "learning_rate": 3.7159720549265234e-05, "loss": 0.231, "step": 30850 }, { "epoch": 3.72, "learning_rate": 3.717176583955673e-05, "loss": 0.322, "step": 30860 }, { "epoch": 3.72, "learning_rate": 3.718381112984823e-05, "loss": 0.5875, "step": 30870 }, { "epoch": 3.72, "learning_rate": 3.7195856420139726e-05, "loss": 0.2961, "step": 30880 }, { "epoch": 3.72, "learning_rate": 3.720790171043123e-05, "loss": 0.3705, "step": 30890 }, { "epoch": 3.72, "learning_rate": 3.7219947000722724e-05, "loss": 0.3186, "step": 30900 }, { "epoch": 3.72, "learning_rate": 3.723199229101422e-05, "loss": 0.1865, "step": 30910 }, { "epoch": 3.72, "learning_rate": 3.7244037581305714e-05, "loss": 0.4117, "step": 30920 }, { "epoch": 3.73, "learning_rate": 3.725608287159721e-05, "loss": 0.3478, "step": 30930 }, { "epoch": 3.73, "learning_rate": 3.7268128161888704e-05, "loss": 0.2881, "step": 30940 }, { "epoch": 3.73, "learning_rate": 3.72801734521802e-05, "loss": 0.1268, "step": 30950 }, { "epoch": 3.73, "learning_rate": 3.7292218742471695e-05, "loss": 0.1755, "step": 30960 }, { "epoch": 3.73, "learning_rate": 3.730426403276319e-05, "loss": 0.0614, "step": 30970 }, { "epoch": 3.73, "learning_rate": 3.7316309323054685e-05, "loss": 0.1117, "step": 30980 }, { "epoch": 3.73, "learning_rate": 3.732835461334619e-05, "loss": 0.4885, "step": 30990 }, { "epoch": 3.73, "learning_rate": 3.734039990363768e-05, "loss": 0.475, "step": 31000 }, { "epoch": 3.74, "learning_rate": 3.735244519392918e-05, "loss": 0.3494, "step": 31010 }, { "epoch": 3.74, "learning_rate": 3.736449048422067e-05, "loss": 0.1653, "step": 31020 }, { "epoch": 3.74, "learning_rate": 3.737653577451217e-05, "loss": 0.2319, "step": 31030 }, { "epoch": 3.74, "learning_rate": 3.738858106480366e-05, "loss": 0.3022, "step": 31040 }, { "epoch": 3.74, "learning_rate": 3.740062635509516e-05, "loss": 0.2353, "step": 31050 }, { "epoch": 3.74, "learning_rate": 3.741267164538665e-05, "loss": 0.1884, "step": 31060 }, { "epoch": 3.74, "learning_rate": 3.742471693567815e-05, "loss": 0.1914, "step": 31070 }, { "epoch": 3.74, "learning_rate": 3.7436762225969644e-05, "loss": 0.1656, "step": 31080 }, { "epoch": 3.74, "learning_rate": 3.744880751626114e-05, "loss": 0.1695, "step": 31090 }, { "epoch": 3.75, "learning_rate": 3.746085280655264e-05, "loss": 0.2556, "step": 31100 }, { "epoch": 3.75, "learning_rate": 3.7472898096844136e-05, "loss": 0.2783, "step": 31110 }, { "epoch": 3.75, "learning_rate": 3.748494338713563e-05, "loss": 0.1514, "step": 31120 }, { "epoch": 3.75, "learning_rate": 3.7496988677427126e-05, "loss": 0.1685, "step": 31130 }, { "epoch": 3.75, "learning_rate": 3.750903396771863e-05, "loss": 0.1317, "step": 31140 }, { "epoch": 3.75, "learning_rate": 3.7521079258010123e-05, "loss": 0.3363, "step": 31150 }, { "epoch": 3.75, "learning_rate": 3.753312454830162e-05, "loss": 0.1205, "step": 31160 }, { "epoch": 3.75, "learning_rate": 3.7545169838593114e-05, "loss": 0.3478, "step": 31170 }, { "epoch": 3.76, "learning_rate": 3.755721512888461e-05, "loss": 0.2595, "step": 31180 }, { "epoch": 3.76, "learning_rate": 3.7569260419176104e-05, "loss": 0.1813, "step": 31190 }, { "epoch": 3.76, "learning_rate": 3.75813057094676e-05, "loss": 0.5207, "step": 31200 }, { "epoch": 3.76, "learning_rate": 3.7593350999759095e-05, "loss": 0.2503, "step": 31210 }, { "epoch": 3.76, "learning_rate": 3.760539629005059e-05, "loss": 0.1479, "step": 31220 }, { "epoch": 3.76, "learning_rate": 3.761744158034209e-05, "loss": 0.1361, "step": 31230 }, { "epoch": 3.76, "learning_rate": 3.762948687063359e-05, "loss": 0.4258, "step": 31240 }, { "epoch": 3.76, "learning_rate": 3.764153216092508e-05, "loss": 0.2277, "step": 31250 }, { "epoch": 3.77, "learning_rate": 3.765357745121658e-05, "loss": 0.294, "step": 31260 }, { "epoch": 3.77, "learning_rate": 3.766562274150807e-05, "loss": 0.4002, "step": 31270 }, { "epoch": 3.77, "learning_rate": 3.767766803179957e-05, "loss": 0.3469, "step": 31280 }, { "epoch": 3.77, "learning_rate": 3.768971332209106e-05, "loss": 0.3815, "step": 31290 }, { "epoch": 3.77, "learning_rate": 3.770175861238256e-05, "loss": 0.1233, "step": 31300 }, { "epoch": 3.77, "learning_rate": 3.771380390267405e-05, "loss": 0.4479, "step": 31310 }, { "epoch": 3.77, "learning_rate": 3.772584919296555e-05, "loss": 0.3598, "step": 31320 }, { "epoch": 3.77, "learning_rate": 3.7737894483257044e-05, "loss": 0.2303, "step": 31330 }, { "epoch": 3.77, "learning_rate": 3.7749939773548546e-05, "loss": 0.2205, "step": 31340 }, { "epoch": 3.78, "learning_rate": 3.776198506384004e-05, "loss": 0.4345, "step": 31350 }, { "epoch": 3.78, "learning_rate": 3.7774030354131536e-05, "loss": 0.2429, "step": 31360 }, { "epoch": 3.78, "learning_rate": 3.778607564442303e-05, "loss": 0.214, "step": 31370 }, { "epoch": 3.78, "learning_rate": 3.7798120934714526e-05, "loss": 0.1664, "step": 31380 }, { "epoch": 3.78, "learning_rate": 3.781016622500602e-05, "loss": 0.2048, "step": 31390 }, { "epoch": 3.78, "learning_rate": 3.7822211515297523e-05, "loss": 0.3724, "step": 31400 }, { "epoch": 3.78, "learning_rate": 3.783425680558902e-05, "loss": 0.1379, "step": 31410 }, { "epoch": 3.78, "learning_rate": 3.7846302095880514e-05, "loss": 0.3881, "step": 31420 }, { "epoch": 3.79, "learning_rate": 3.785834738617201e-05, "loss": 0.5259, "step": 31430 }, { "epoch": 3.79, "learning_rate": 3.7870392676463504e-05, "loss": 0.1808, "step": 31440 }, { "epoch": 3.79, "learning_rate": 3.7882437966755e-05, "loss": 0.2002, "step": 31450 }, { "epoch": 3.79, "learning_rate": 3.78944832570465e-05, "loss": 0.1679, "step": 31460 }, { "epoch": 3.79, "learning_rate": 3.7906528547337996e-05, "loss": 0.1672, "step": 31470 }, { "epoch": 3.79, "learning_rate": 3.791857383762949e-05, "loss": 0.2711, "step": 31480 }, { "epoch": 3.79, "learning_rate": 3.793061912792099e-05, "loss": 0.3224, "step": 31490 }, { "epoch": 3.79, "learning_rate": 3.794266441821248e-05, "loss": 0.2515, "step": 31500 }, { "epoch": 3.8, "learning_rate": 3.795470970850398e-05, "loss": 0.194, "step": 31510 }, { "epoch": 3.8, "learning_rate": 3.796675499879547e-05, "loss": 0.1304, "step": 31520 }, { "epoch": 3.8, "learning_rate": 3.797880028908697e-05, "loss": 0.2688, "step": 31530 }, { "epoch": 3.8, "learning_rate": 3.799084557937846e-05, "loss": 0.5799, "step": 31540 }, { "epoch": 3.8, "learning_rate": 3.800289086966996e-05, "loss": 0.2252, "step": 31550 }, { "epoch": 3.8, "learning_rate": 3.801493615996145e-05, "loss": 0.3822, "step": 31560 }, { "epoch": 3.8, "learning_rate": 3.8026981450252955e-05, "loss": 0.2039, "step": 31570 }, { "epoch": 3.8, "learning_rate": 3.803902674054445e-05, "loss": 0.4044, "step": 31580 }, { "epoch": 3.8, "learning_rate": 3.8051072030835945e-05, "loss": 0.4511, "step": 31590 }, { "epoch": 3.81, "learning_rate": 3.806311732112744e-05, "loss": 0.3262, "step": 31600 }, { "epoch": 3.81, "learning_rate": 3.8075162611418936e-05, "loss": 0.3387, "step": 31610 }, { "epoch": 3.81, "learning_rate": 3.808720790171043e-05, "loss": 0.2547, "step": 31620 }, { "epoch": 3.81, "learning_rate": 3.8099253192001926e-05, "loss": 0.3501, "step": 31630 }, { "epoch": 3.81, "learning_rate": 3.811129848229342e-05, "loss": 0.2675, "step": 31640 }, { "epoch": 3.81, "learning_rate": 3.8123343772584917e-05, "loss": 0.2941, "step": 31650 }, { "epoch": 3.81, "learning_rate": 3.813538906287642e-05, "loss": 0.5954, "step": 31660 }, { "epoch": 3.81, "learning_rate": 3.8147434353167914e-05, "loss": 0.2393, "step": 31670 }, { "epoch": 3.82, "learning_rate": 3.815947964345941e-05, "loss": 0.2527, "step": 31680 }, { "epoch": 3.82, "learning_rate": 3.8171524933750904e-05, "loss": 0.2628, "step": 31690 }, { "epoch": 3.82, "learning_rate": 3.8183570224042406e-05, "loss": 0.4984, "step": 31700 }, { "epoch": 3.82, "learning_rate": 3.81956155143339e-05, "loss": 0.3775, "step": 31710 }, { "epoch": 3.82, "learning_rate": 3.8207660804625396e-05, "loss": 0.2264, "step": 31720 }, { "epoch": 3.82, "learning_rate": 3.821970609491689e-05, "loss": 0.3325, "step": 31730 }, { "epoch": 3.82, "learning_rate": 3.823175138520839e-05, "loss": 0.2153, "step": 31740 }, { "epoch": 3.82, "learning_rate": 3.824379667549988e-05, "loss": 0.4226, "step": 31750 }, { "epoch": 3.83, "learning_rate": 3.825584196579138e-05, "loss": 0.2325, "step": 31760 }, { "epoch": 3.83, "learning_rate": 3.826788725608287e-05, "loss": 0.1181, "step": 31770 }, { "epoch": 3.83, "learning_rate": 3.827993254637437e-05, "loss": 0.4905, "step": 31780 }, { "epoch": 3.83, "learning_rate": 3.829197783666586e-05, "loss": 0.2614, "step": 31790 }, { "epoch": 3.83, "learning_rate": 3.830402312695736e-05, "loss": 0.2512, "step": 31800 }, { "epoch": 3.83, "learning_rate": 3.831606841724886e-05, "loss": 0.3443, "step": 31810 }, { "epoch": 3.83, "learning_rate": 3.8328113707540355e-05, "loss": 0.1375, "step": 31820 }, { "epoch": 3.83, "learning_rate": 3.834015899783185e-05, "loss": 0.2822, "step": 31830 }, { "epoch": 3.83, "learning_rate": 3.8352204288123345e-05, "loss": 0.3088, "step": 31840 }, { "epoch": 3.84, "learning_rate": 3.836424957841484e-05, "loss": 0.2784, "step": 31850 }, { "epoch": 3.84, "learning_rate": 3.8376294868706336e-05, "loss": 0.3122, "step": 31860 }, { "epoch": 3.84, "learning_rate": 3.838834015899783e-05, "loss": 0.1446, "step": 31870 }, { "epoch": 3.84, "learning_rate": 3.8400385449289326e-05, "loss": 0.3248, "step": 31880 }, { "epoch": 3.84, "learning_rate": 3.841243073958082e-05, "loss": 0.3632, "step": 31890 }, { "epoch": 3.84, "learning_rate": 3.8424476029872317e-05, "loss": 0.2204, "step": 31900 }, { "epoch": 3.84, "learning_rate": 3.843652132016382e-05, "loss": 0.2558, "step": 31910 }, { "epoch": 3.84, "learning_rate": 3.8448566610455314e-05, "loss": 0.214, "step": 31920 }, { "epoch": 3.85, "learning_rate": 3.8460611900746816e-05, "loss": 0.2772, "step": 31930 }, { "epoch": 3.85, "learning_rate": 3.847265719103831e-05, "loss": 0.2405, "step": 31940 }, { "epoch": 3.85, "learning_rate": 3.8484702481329806e-05, "loss": 0.3423, "step": 31950 }, { "epoch": 3.85, "learning_rate": 3.84967477716213e-05, "loss": 0.3093, "step": 31960 }, { "epoch": 3.85, "learning_rate": 3.8508793061912796e-05, "loss": 0.088, "step": 31970 }, { "epoch": 3.85, "learning_rate": 3.852083835220429e-05, "loss": 0.4126, "step": 31980 }, { "epoch": 3.85, "learning_rate": 3.853288364249579e-05, "loss": 0.3808, "step": 31990 }, { "epoch": 3.85, "learning_rate": 3.854492893278728e-05, "loss": 0.1786, "step": 32000 }, { "epoch": 3.86, "learning_rate": 3.855697422307878e-05, "loss": 0.1148, "step": 32010 }, { "epoch": 3.86, "learning_rate": 3.856901951337027e-05, "loss": 0.2674, "step": 32020 }, { "epoch": 3.86, "learning_rate": 3.858106480366177e-05, "loss": 0.3931, "step": 32030 }, { "epoch": 3.86, "learning_rate": 3.859311009395327e-05, "loss": 0.2202, "step": 32040 }, { "epoch": 3.86, "learning_rate": 3.8605155384244765e-05, "loss": 0.1917, "step": 32050 }, { "epoch": 3.86, "learning_rate": 3.861720067453626e-05, "loss": 0.2674, "step": 32060 }, { "epoch": 3.86, "learning_rate": 3.8629245964827755e-05, "loss": 0.335, "step": 32070 }, { "epoch": 3.86, "learning_rate": 3.864129125511925e-05, "loss": 0.3662, "step": 32080 }, { "epoch": 3.87, "learning_rate": 3.8653336545410745e-05, "loss": 0.2781, "step": 32090 }, { "epoch": 3.87, "learning_rate": 3.866538183570224e-05, "loss": 0.0613, "step": 32100 }, { "epoch": 3.87, "learning_rate": 3.8677427125993736e-05, "loss": 0.197, "step": 32110 }, { "epoch": 3.87, "learning_rate": 3.868947241628523e-05, "loss": 0.3122, "step": 32120 }, { "epoch": 3.87, "learning_rate": 3.8701517706576726e-05, "loss": 0.289, "step": 32130 }, { "epoch": 3.87, "learning_rate": 3.871356299686822e-05, "loss": 0.167, "step": 32140 }, { "epoch": 3.87, "learning_rate": 3.872560828715972e-05, "loss": 0.4499, "step": 32150 }, { "epoch": 3.87, "learning_rate": 3.873765357745122e-05, "loss": 0.2209, "step": 32160 }, { "epoch": 3.87, "learning_rate": 3.8749698867742714e-05, "loss": 0.315, "step": 32170 }, { "epoch": 3.88, "learning_rate": 3.876174415803421e-05, "loss": 0.2434, "step": 32180 }, { "epoch": 3.88, "learning_rate": 3.877378944832571e-05, "loss": 0.2975, "step": 32190 }, { "epoch": 3.88, "learning_rate": 3.8785834738617206e-05, "loss": 0.3039, "step": 32200 }, { "epoch": 3.88, "learning_rate": 3.87978800289087e-05, "loss": 0.2617, "step": 32210 }, { "epoch": 3.88, "learning_rate": 3.8809925319200196e-05, "loss": 0.0703, "step": 32220 }, { "epoch": 3.88, "learning_rate": 3.882197060949169e-05, "loss": 0.2077, "step": 32230 }, { "epoch": 3.88, "learning_rate": 3.883401589978319e-05, "loss": 0.4114, "step": 32240 }, { "epoch": 3.88, "learning_rate": 3.884606119007468e-05, "loss": 0.2079, "step": 32250 }, { "epoch": 3.89, "learning_rate": 3.885810648036618e-05, "loss": 0.2474, "step": 32260 }, { "epoch": 3.89, "learning_rate": 3.887015177065768e-05, "loss": 0.2822, "step": 32270 }, { "epoch": 3.89, "learning_rate": 3.8882197060949174e-05, "loss": 0.3251, "step": 32280 }, { "epoch": 3.89, "learning_rate": 3.889424235124067e-05, "loss": 0.4296, "step": 32290 }, { "epoch": 3.89, "learning_rate": 3.8906287641532165e-05, "loss": 0.4961, "step": 32300 }, { "epoch": 3.89, "learning_rate": 3.891833293182366e-05, "loss": 0.1191, "step": 32310 }, { "epoch": 3.89, "learning_rate": 3.8930378222115155e-05, "loss": 0.193, "step": 32320 }, { "epoch": 3.89, "learning_rate": 3.894242351240665e-05, "loss": 0.1235, "step": 32330 }, { "epoch": 3.9, "learning_rate": 3.8954468802698145e-05, "loss": 0.1821, "step": 32340 }, { "epoch": 3.9, "learning_rate": 3.896651409298964e-05, "loss": 0.3404, "step": 32350 }, { "epoch": 3.9, "learning_rate": 3.8978559383281136e-05, "loss": 0.2161, "step": 32360 }, { "epoch": 3.9, "learning_rate": 3.899060467357263e-05, "loss": 0.4441, "step": 32370 }, { "epoch": 3.9, "learning_rate": 3.900264996386413e-05, "loss": 0.4316, "step": 32380 }, { "epoch": 3.9, "learning_rate": 3.901469525415563e-05, "loss": 0.2905, "step": 32390 }, { "epoch": 3.9, "learning_rate": 3.902674054444712e-05, "loss": 0.159, "step": 32400 }, { "epoch": 3.9, "learning_rate": 3.903878583473862e-05, "loss": 0.2364, "step": 32410 }, { "epoch": 3.9, "learning_rate": 3.9050831125030114e-05, "loss": 0.1755, "step": 32420 }, { "epoch": 3.91, "learning_rate": 3.906287641532161e-05, "loss": 0.3848, "step": 32430 }, { "epoch": 3.91, "learning_rate": 3.9074921705613104e-05, "loss": 0.2705, "step": 32440 }, { "epoch": 3.91, "learning_rate": 3.9086966995904606e-05, "loss": 0.2744, "step": 32450 }, { "epoch": 3.91, "learning_rate": 3.90990122861961e-05, "loss": 0.4313, "step": 32460 }, { "epoch": 3.91, "learning_rate": 3.9111057576487596e-05, "loss": 0.1795, "step": 32470 }, { "epoch": 3.91, "learning_rate": 3.912310286677909e-05, "loss": 0.3415, "step": 32480 }, { "epoch": 3.91, "learning_rate": 3.913514815707059e-05, "loss": 0.311, "step": 32490 }, { "epoch": 3.91, "learning_rate": 3.914719344736208e-05, "loss": 0.2383, "step": 32500 }, { "epoch": 3.92, "learning_rate": 3.9159238737653584e-05, "loss": 0.194, "step": 32510 }, { "epoch": 3.92, "learning_rate": 3.917128402794508e-05, "loss": 0.3744, "step": 32520 }, { "epoch": 3.92, "learning_rate": 3.9183329318236574e-05, "loss": 0.2468, "step": 32530 }, { "epoch": 3.92, "learning_rate": 3.919537460852807e-05, "loss": 0.1223, "step": 32540 }, { "epoch": 3.92, "learning_rate": 3.9207419898819565e-05, "loss": 0.2399, "step": 32550 }, { "epoch": 3.92, "learning_rate": 3.921946518911106e-05, "loss": 0.1196, "step": 32560 }, { "epoch": 3.92, "learning_rate": 3.9231510479402555e-05, "loss": 0.2538, "step": 32570 }, { "epoch": 3.92, "learning_rate": 3.924355576969405e-05, "loss": 0.3124, "step": 32580 }, { "epoch": 3.93, "learning_rate": 3.9255601059985545e-05, "loss": 0.2125, "step": 32590 }, { "epoch": 3.93, "learning_rate": 3.926764635027704e-05, "loss": 0.2427, "step": 32600 }, { "epoch": 3.93, "learning_rate": 3.9279691640568536e-05, "loss": 0.1456, "step": 32610 }, { "epoch": 3.93, "learning_rate": 3.929173693086004e-05, "loss": 0.1592, "step": 32620 }, { "epoch": 3.93, "learning_rate": 3.930378222115153e-05, "loss": 0.3114, "step": 32630 }, { "epoch": 3.93, "learning_rate": 3.931582751144303e-05, "loss": 0.1138, "step": 32640 }, { "epoch": 3.93, "learning_rate": 3.932787280173452e-05, "loss": 0.3536, "step": 32650 }, { "epoch": 3.93, "learning_rate": 3.933991809202602e-05, "loss": 0.2036, "step": 32660 }, { "epoch": 3.93, "learning_rate": 3.9351963382317514e-05, "loss": 0.2999, "step": 32670 }, { "epoch": 3.94, "learning_rate": 3.936400867260901e-05, "loss": 0.2174, "step": 32680 }, { "epoch": 3.94, "learning_rate": 3.9376053962900504e-05, "loss": 0.1972, "step": 32690 }, { "epoch": 3.94, "learning_rate": 3.9388099253192e-05, "loss": 0.2826, "step": 32700 }, { "epoch": 3.94, "learning_rate": 3.94001445434835e-05, "loss": 0.3127, "step": 32710 }, { "epoch": 3.94, "learning_rate": 3.9412189833774996e-05, "loss": 0.2374, "step": 32720 }, { "epoch": 3.94, "learning_rate": 3.942423512406649e-05, "loss": 0.2799, "step": 32730 }, { "epoch": 3.94, "learning_rate": 3.943628041435799e-05, "loss": 0.3392, "step": 32740 }, { "epoch": 3.94, "learning_rate": 3.944832570464949e-05, "loss": 0.4945, "step": 32750 }, { "epoch": 3.95, "learning_rate": 3.9460370994940984e-05, "loss": 0.3363, "step": 32760 }, { "epoch": 3.95, "learning_rate": 3.947241628523248e-05, "loss": 0.3333, "step": 32770 }, { "epoch": 3.95, "learning_rate": 3.9484461575523974e-05, "loss": 0.2672, "step": 32780 }, { "epoch": 3.95, "learning_rate": 3.949650686581547e-05, "loss": 0.2658, "step": 32790 }, { "epoch": 3.95, "learning_rate": 3.9508552156106964e-05, "loss": 0.421, "step": 32800 }, { "epoch": 3.95, "learning_rate": 3.952059744639846e-05, "loss": 0.4374, "step": 32810 }, { "epoch": 3.95, "learning_rate": 3.9532642736689955e-05, "loss": 0.3412, "step": 32820 }, { "epoch": 3.95, "learning_rate": 3.954468802698145e-05, "loss": 0.4263, "step": 32830 }, { "epoch": 3.96, "learning_rate": 3.9556733317272945e-05, "loss": 0.4438, "step": 32840 }, { "epoch": 3.96, "learning_rate": 3.956877860756445e-05, "loss": 0.3669, "step": 32850 }, { "epoch": 3.96, "learning_rate": 3.958082389785594e-05, "loss": 0.1728, "step": 32860 }, { "epoch": 3.96, "learning_rate": 3.959286918814744e-05, "loss": 0.3205, "step": 32870 }, { "epoch": 3.96, "learning_rate": 3.960491447843893e-05, "loss": 0.3602, "step": 32880 }, { "epoch": 3.96, "learning_rate": 3.961695976873043e-05, "loss": 0.1639, "step": 32890 }, { "epoch": 3.96, "learning_rate": 3.962900505902192e-05, "loss": 0.306, "step": 32900 }, { "epoch": 3.96, "learning_rate": 3.964105034931342e-05, "loss": 0.1455, "step": 32910 }, { "epoch": 3.97, "learning_rate": 3.9653095639604913e-05, "loss": 0.05, "step": 32920 }, { "epoch": 3.97, "learning_rate": 3.966514092989641e-05, "loss": 0.1081, "step": 32930 }, { "epoch": 3.97, "learning_rate": 3.9677186220187904e-05, "loss": 0.2159, "step": 32940 }, { "epoch": 3.97, "learning_rate": 3.96892315104794e-05, "loss": 0.3165, "step": 32950 }, { "epoch": 3.97, "learning_rate": 3.97012768007709e-05, "loss": 0.2959, "step": 32960 }, { "epoch": 3.97, "learning_rate": 3.9713322091062396e-05, "loss": 0.0904, "step": 32970 }, { "epoch": 3.97, "learning_rate": 3.97253673813539e-05, "loss": 0.176, "step": 32980 }, { "epoch": 3.97, "learning_rate": 3.973741267164539e-05, "loss": 0.3635, "step": 32990 }, { "epoch": 3.97, "learning_rate": 3.974945796193689e-05, "loss": 0.2314, "step": 33000 }, { "epoch": 3.98, "learning_rate": 3.9761503252228384e-05, "loss": 0.4163, "step": 33010 }, { "epoch": 3.98, "learning_rate": 3.977354854251988e-05, "loss": 0.2576, "step": 33020 }, { "epoch": 3.98, "learning_rate": 3.9785593832811374e-05, "loss": 0.3855, "step": 33030 }, { "epoch": 3.98, "learning_rate": 3.979763912310287e-05, "loss": 0.2024, "step": 33040 }, { "epoch": 3.98, "learning_rate": 3.9809684413394364e-05, "loss": 0.2654, "step": 33050 }, { "epoch": 3.98, "learning_rate": 3.982172970368586e-05, "loss": 0.2568, "step": 33060 }, { "epoch": 3.98, "learning_rate": 3.9833774993977355e-05, "loss": 0.2786, "step": 33070 }, { "epoch": 3.98, "learning_rate": 3.984582028426885e-05, "loss": 0.1339, "step": 33080 }, { "epoch": 3.99, "learning_rate": 3.985786557456035e-05, "loss": 0.3138, "step": 33090 }, { "epoch": 3.99, "learning_rate": 3.986991086485185e-05, "loss": 0.2643, "step": 33100 }, { "epoch": 3.99, "learning_rate": 3.988195615514334e-05, "loss": 0.217, "step": 33110 }, { "epoch": 3.99, "learning_rate": 3.989400144543484e-05, "loss": 0.2583, "step": 33120 }, { "epoch": 3.99, "learning_rate": 3.990604673572633e-05, "loss": 0.3285, "step": 33130 }, { "epoch": 3.99, "learning_rate": 3.991809202601783e-05, "loss": 0.2812, "step": 33140 }, { "epoch": 3.99, "learning_rate": 3.993013731630932e-05, "loss": 0.2848, "step": 33150 }, { "epoch": 3.99, "learning_rate": 3.994218260660082e-05, "loss": 0.2948, "step": 33160 }, { "epoch": 4.0, "learning_rate": 3.9954227896892313e-05, "loss": 0.293, "step": 33170 }, { "epoch": 4.0, "learning_rate": 3.996627318718381e-05, "loss": 0.239, "step": 33180 }, { "epoch": 4.0, "learning_rate": 3.997831847747531e-05, "loss": 0.1979, "step": 33190 }, { "epoch": 4.0, "learning_rate": 3.9990363767766806e-05, "loss": 0.2219, "step": 33200 }, { "epoch": 4.0, "learning_rate": 4.00024090580583e-05, "loss": 0.09, "step": 33210 }, { "epoch": 4.0, "eval_accuracy": { "accuracy": 0.9270099367660344 }, "eval_f1": { "f1": 0.9272025877193879 }, "eval_loss": 0.27678197622299194, "eval_precision": { "precision": 0.9305221603080029 }, "eval_recall": { "recall": 0.9267551810236085 }, "eval_runtime": 429.0362, "eval_samples_per_second": 38.703, "eval_steps_per_second": 19.353, "step": 33210 }, { "epoch": 4.0, "learning_rate": 4.0014454348349796e-05, "loss": 0.2848, "step": 33220 }, { "epoch": 4.0, "learning_rate": 4.002649963864129e-05, "loss": 0.2293, "step": 33230 }, { "epoch": 4.0, "learning_rate": 4.003854492893279e-05, "loss": 0.1984, "step": 33240 }, { "epoch": 4.0, "learning_rate": 4.005059021922429e-05, "loss": 0.1779, "step": 33250 }, { "epoch": 4.01, "learning_rate": 4.0062635509515784e-05, "loss": 0.3088, "step": 33260 }, { "epoch": 4.01, "learning_rate": 4.007468079980728e-05, "loss": 0.2359, "step": 33270 }, { "epoch": 4.01, "learning_rate": 4.0086726090098774e-05, "loss": 0.1043, "step": 33280 }, { "epoch": 4.01, "learning_rate": 4.009877138039027e-05, "loss": 0.2482, "step": 33290 }, { "epoch": 4.01, "learning_rate": 4.0110816670681764e-05, "loss": 0.1613, "step": 33300 }, { "epoch": 4.01, "learning_rate": 4.012286196097326e-05, "loss": 0.2923, "step": 33310 }, { "epoch": 4.01, "learning_rate": 4.013490725126476e-05, "loss": 0.2288, "step": 33320 }, { "epoch": 4.01, "learning_rate": 4.014695254155626e-05, "loss": 0.3003, "step": 33330 }, { "epoch": 4.02, "learning_rate": 4.015899783184775e-05, "loss": 0.067, "step": 33340 }, { "epoch": 4.02, "learning_rate": 4.017104312213925e-05, "loss": 0.2218, "step": 33350 }, { "epoch": 4.02, "learning_rate": 4.018308841243074e-05, "loss": 0.3513, "step": 33360 }, { "epoch": 4.02, "learning_rate": 4.019513370272224e-05, "loss": 0.231, "step": 33370 }, { "epoch": 4.02, "learning_rate": 4.020717899301373e-05, "loss": 0.1881, "step": 33380 }, { "epoch": 4.02, "learning_rate": 4.021922428330523e-05, "loss": 0.1857, "step": 33390 }, { "epoch": 4.02, "learning_rate": 4.023126957359672e-05, "loss": 0.2366, "step": 33400 }, { "epoch": 4.02, "learning_rate": 4.024331486388822e-05, "loss": 0.3442, "step": 33410 }, { "epoch": 4.03, "learning_rate": 4.0255360154179713e-05, "loss": 0.235, "step": 33420 }, { "epoch": 4.03, "learning_rate": 4.0267405444471215e-05, "loss": 0.1487, "step": 33430 }, { "epoch": 4.03, "learning_rate": 4.027945073476271e-05, "loss": 0.217, "step": 33440 }, { "epoch": 4.03, "learning_rate": 4.0291496025054206e-05, "loss": 0.3399, "step": 33450 }, { "epoch": 4.03, "learning_rate": 4.03035413153457e-05, "loss": 0.1086, "step": 33460 }, { "epoch": 4.03, "learning_rate": 4.0315586605637196e-05, "loss": 0.1164, "step": 33470 }, { "epoch": 4.03, "learning_rate": 4.032763189592869e-05, "loss": 0.431, "step": 33480 }, { "epoch": 4.03, "learning_rate": 4.0339677186220186e-05, "loss": 0.1056, "step": 33490 }, { "epoch": 4.03, "learning_rate": 4.035172247651169e-05, "loss": 0.4112, "step": 33500 }, { "epoch": 4.04, "learning_rate": 4.0363767766803184e-05, "loss": 0.2826, "step": 33510 }, { "epoch": 4.04, "learning_rate": 4.037581305709468e-05, "loss": 0.3452, "step": 33520 }, { "epoch": 4.04, "learning_rate": 4.0387858347386174e-05, "loss": 0.1911, "step": 33530 }, { "epoch": 4.04, "learning_rate": 4.039990363767767e-05, "loss": 0.2294, "step": 33540 }, { "epoch": 4.04, "learning_rate": 4.0411948927969164e-05, "loss": 0.2684, "step": 33550 }, { "epoch": 4.04, "learning_rate": 4.0423994218260666e-05, "loss": 0.4026, "step": 33560 }, { "epoch": 4.04, "learning_rate": 4.043603950855216e-05, "loss": 0.3656, "step": 33570 }, { "epoch": 4.04, "learning_rate": 4.044808479884366e-05, "loss": 0.1778, "step": 33580 }, { "epoch": 4.05, "learning_rate": 4.046013008913515e-05, "loss": 0.3002, "step": 33590 }, { "epoch": 4.05, "learning_rate": 4.047217537942665e-05, "loss": 0.1779, "step": 33600 }, { "epoch": 4.05, "learning_rate": 4.048422066971814e-05, "loss": 0.1599, "step": 33610 }, { "epoch": 4.05, "learning_rate": 4.049626596000964e-05, "loss": 0.2857, "step": 33620 }, { "epoch": 4.05, "learning_rate": 4.050831125030113e-05, "loss": 0.2589, "step": 33630 }, { "epoch": 4.05, "learning_rate": 4.052035654059263e-05, "loss": 0.1421, "step": 33640 }, { "epoch": 4.05, "learning_rate": 4.053240183088412e-05, "loss": 0.3318, "step": 33650 }, { "epoch": 4.05, "learning_rate": 4.0544447121175625e-05, "loss": 0.1712, "step": 33660 }, { "epoch": 4.06, "learning_rate": 4.055649241146712e-05, "loss": 0.1877, "step": 33670 }, { "epoch": 4.06, "learning_rate": 4.0568537701758615e-05, "loss": 0.1467, "step": 33680 }, { "epoch": 4.06, "learning_rate": 4.058058299205011e-05, "loss": 0.1698, "step": 33690 }, { "epoch": 4.06, "learning_rate": 4.0592628282341606e-05, "loss": 0.4227, "step": 33700 }, { "epoch": 4.06, "learning_rate": 4.06046735726331e-05, "loss": 0.1164, "step": 33710 }, { "epoch": 4.06, "learning_rate": 4.0616718862924596e-05, "loss": 0.2135, "step": 33720 }, { "epoch": 4.06, "learning_rate": 4.062876415321609e-05, "loss": 0.3482, "step": 33730 }, { "epoch": 4.06, "learning_rate": 4.0640809443507586e-05, "loss": 0.2802, "step": 33740 }, { "epoch": 4.07, "learning_rate": 4.065285473379908e-05, "loss": 0.2943, "step": 33750 }, { "epoch": 4.07, "learning_rate": 4.0664900024090584e-05, "loss": 0.1267, "step": 33760 }, { "epoch": 4.07, "learning_rate": 4.067694531438208e-05, "loss": 0.3951, "step": 33770 }, { "epoch": 4.07, "learning_rate": 4.0688990604673574e-05, "loss": 0.1685, "step": 33780 }, { "epoch": 4.07, "learning_rate": 4.0701035894965076e-05, "loss": 0.2998, "step": 33790 }, { "epoch": 4.07, "learning_rate": 4.071308118525657e-05, "loss": 0.3728, "step": 33800 }, { "epoch": 4.07, "learning_rate": 4.0725126475548066e-05, "loss": 0.1915, "step": 33810 }, { "epoch": 4.07, "learning_rate": 4.073717176583956e-05, "loss": 0.2399, "step": 33820 }, { "epoch": 4.07, "learning_rate": 4.074921705613106e-05, "loss": 0.1379, "step": 33830 }, { "epoch": 4.08, "learning_rate": 4.076126234642255e-05, "loss": 0.2221, "step": 33840 }, { "epoch": 4.08, "learning_rate": 4.077330763671405e-05, "loss": 0.228, "step": 33850 }, { "epoch": 4.08, "learning_rate": 4.078535292700554e-05, "loss": 0.4477, "step": 33860 }, { "epoch": 4.08, "learning_rate": 4.079739821729704e-05, "loss": 0.2107, "step": 33870 }, { "epoch": 4.08, "learning_rate": 4.080944350758853e-05, "loss": 0.4021, "step": 33880 }, { "epoch": 4.08, "learning_rate": 4.082148879788003e-05, "loss": 0.2587, "step": 33890 }, { "epoch": 4.08, "learning_rate": 4.083353408817153e-05, "loss": 0.1665, "step": 33900 }, { "epoch": 4.08, "learning_rate": 4.0845579378463025e-05, "loss": 0.1526, "step": 33910 }, { "epoch": 4.09, "learning_rate": 4.085762466875452e-05, "loss": 0.2276, "step": 33920 }, { "epoch": 4.09, "learning_rate": 4.0869669959046015e-05, "loss": 0.1205, "step": 33930 }, { "epoch": 4.09, "learning_rate": 4.088171524933751e-05, "loss": 0.1753, "step": 33940 }, { "epoch": 4.09, "learning_rate": 4.0893760539629006e-05, "loss": 0.2157, "step": 33950 }, { "epoch": 4.09, "learning_rate": 4.09058058299205e-05, "loss": 0.1346, "step": 33960 }, { "epoch": 4.09, "learning_rate": 4.0917851120211996e-05, "loss": 0.2414, "step": 33970 }, { "epoch": 4.09, "learning_rate": 4.092989641050349e-05, "loss": 0.3214, "step": 33980 }, { "epoch": 4.09, "learning_rate": 4.0941941700794986e-05, "loss": 0.3031, "step": 33990 }, { "epoch": 4.1, "learning_rate": 4.095398699108648e-05, "loss": 0.3728, "step": 34000 }, { "epoch": 4.1, "learning_rate": 4.0966032281377984e-05, "loss": 0.1784, "step": 34010 }, { "epoch": 4.1, "learning_rate": 4.097807757166948e-05, "loss": 0.1688, "step": 34020 }, { "epoch": 4.1, "learning_rate": 4.099012286196098e-05, "loss": 0.2658, "step": 34030 }, { "epoch": 4.1, "learning_rate": 4.1002168152252476e-05, "loss": 0.1885, "step": 34040 }, { "epoch": 4.1, "learning_rate": 4.101421344254397e-05, "loss": 0.0878, "step": 34050 }, { "epoch": 4.1, "learning_rate": 4.1026258732835466e-05, "loss": 0.1709, "step": 34060 }, { "epoch": 4.1, "learning_rate": 4.103830402312696e-05, "loss": 0.2661, "step": 34070 }, { "epoch": 4.1, "learning_rate": 4.1050349313418457e-05, "loss": 0.1532, "step": 34080 }, { "epoch": 4.11, "learning_rate": 4.106239460370995e-05, "loss": 0.2694, "step": 34090 }, { "epoch": 4.11, "learning_rate": 4.107443989400145e-05, "loss": 0.2558, "step": 34100 }, { "epoch": 4.11, "learning_rate": 4.108648518429294e-05, "loss": 0.1314, "step": 34110 }, { "epoch": 4.11, "learning_rate": 4.109853047458444e-05, "loss": 0.2068, "step": 34120 }, { "epoch": 4.11, "learning_rate": 4.111057576487594e-05, "loss": 0.0472, "step": 34130 }, { "epoch": 4.11, "learning_rate": 4.1122621055167434e-05, "loss": 0.2623, "step": 34140 }, { "epoch": 4.11, "learning_rate": 4.113466634545893e-05, "loss": 0.3258, "step": 34150 }, { "epoch": 4.11, "learning_rate": 4.1146711635750425e-05, "loss": 0.1509, "step": 34160 }, { "epoch": 4.12, "learning_rate": 4.115875692604192e-05, "loss": 0.3047, "step": 34170 }, { "epoch": 4.12, "learning_rate": 4.1170802216333415e-05, "loss": 0.3142, "step": 34180 }, { "epoch": 4.12, "learning_rate": 4.118284750662491e-05, "loss": 0.3367, "step": 34190 }, { "epoch": 4.12, "learning_rate": 4.1194892796916406e-05, "loss": 0.2741, "step": 34200 }, { "epoch": 4.12, "learning_rate": 4.12069380872079e-05, "loss": 0.1541, "step": 34210 }, { "epoch": 4.12, "learning_rate": 4.1218983377499396e-05, "loss": 0.3188, "step": 34220 }, { "epoch": 4.12, "learning_rate": 4.123102866779089e-05, "loss": 0.0998, "step": 34230 }, { "epoch": 4.12, "learning_rate": 4.124307395808239e-05, "loss": 0.4016, "step": 34240 }, { "epoch": 4.13, "learning_rate": 4.125511924837389e-05, "loss": 0.2447, "step": 34250 }, { "epoch": 4.13, "learning_rate": 4.1267164538665383e-05, "loss": 0.2129, "step": 34260 }, { "epoch": 4.13, "learning_rate": 4.127920982895688e-05, "loss": 0.2318, "step": 34270 }, { "epoch": 4.13, "learning_rate": 4.1291255119248374e-05, "loss": 0.1819, "step": 34280 }, { "epoch": 4.13, "learning_rate": 4.1303300409539876e-05, "loss": 0.2334, "step": 34290 }, { "epoch": 4.13, "learning_rate": 4.131534569983137e-05, "loss": 0.1018, "step": 34300 }, { "epoch": 4.13, "learning_rate": 4.1327390990122866e-05, "loss": 0.2236, "step": 34310 }, { "epoch": 4.13, "learning_rate": 4.133943628041436e-05, "loss": 0.2631, "step": 34320 }, { "epoch": 4.13, "learning_rate": 4.1351481570705857e-05, "loss": 0.3035, "step": 34330 }, { "epoch": 4.14, "learning_rate": 4.136352686099735e-05, "loss": 0.1854, "step": 34340 }, { "epoch": 4.14, "learning_rate": 4.137557215128885e-05, "loss": 0.3571, "step": 34350 }, { "epoch": 4.14, "learning_rate": 4.138761744158034e-05, "loss": 0.2795, "step": 34360 }, { "epoch": 4.14, "learning_rate": 4.1399662731871844e-05, "loss": 0.5032, "step": 34370 }, { "epoch": 4.14, "learning_rate": 4.141170802216334e-05, "loss": 0.2605, "step": 34380 }, { "epoch": 4.14, "learning_rate": 4.1423753312454834e-05, "loss": 0.383, "step": 34390 }, { "epoch": 4.14, "learning_rate": 4.143579860274633e-05, "loss": 0.193, "step": 34400 }, { "epoch": 4.14, "learning_rate": 4.1447843893037825e-05, "loss": 0.226, "step": 34410 }, { "epoch": 4.15, "learning_rate": 4.145988918332932e-05, "loss": 0.2275, "step": 34420 }, { "epoch": 4.15, "learning_rate": 4.1471934473620815e-05, "loss": 0.344, "step": 34430 }, { "epoch": 4.15, "learning_rate": 4.148397976391231e-05, "loss": 0.2784, "step": 34440 }, { "epoch": 4.15, "learning_rate": 4.1496025054203806e-05, "loss": 0.1532, "step": 34450 }, { "epoch": 4.15, "learning_rate": 4.15080703444953e-05, "loss": 0.1556, "step": 34460 }, { "epoch": 4.15, "learning_rate": 4.1520115634786796e-05, "loss": 0.2741, "step": 34470 }, { "epoch": 4.15, "learning_rate": 4.15321609250783e-05, "loss": 0.2575, "step": 34480 }, { "epoch": 4.15, "learning_rate": 4.154420621536979e-05, "loss": 0.2658, "step": 34490 }, { "epoch": 4.16, "learning_rate": 4.155625150566129e-05, "loss": 0.1676, "step": 34500 }, { "epoch": 4.16, "learning_rate": 4.1568296795952783e-05, "loss": 0.2814, "step": 34510 }, { "epoch": 4.16, "learning_rate": 4.158034208624428e-05, "loss": 0.1741, "step": 34520 }, { "epoch": 4.16, "learning_rate": 4.1592387376535774e-05, "loss": 0.2742, "step": 34530 }, { "epoch": 4.16, "learning_rate": 4.160443266682727e-05, "loss": 0.1246, "step": 34540 }, { "epoch": 4.16, "learning_rate": 4.161647795711877e-05, "loss": 0.2753, "step": 34550 }, { "epoch": 4.16, "learning_rate": 4.1628523247410266e-05, "loss": 0.2132, "step": 34560 }, { "epoch": 4.16, "learning_rate": 4.164056853770176e-05, "loss": 0.1867, "step": 34570 }, { "epoch": 4.17, "learning_rate": 4.1652613827993256e-05, "loss": 0.0775, "step": 34580 }, { "epoch": 4.17, "learning_rate": 4.166465911828475e-05, "loss": 0.4127, "step": 34590 }, { "epoch": 4.17, "learning_rate": 4.1676704408576254e-05, "loss": 0.2134, "step": 34600 }, { "epoch": 4.17, "learning_rate": 4.168874969886775e-05, "loss": 0.1471, "step": 34610 }, { "epoch": 4.17, "learning_rate": 4.1700794989159244e-05, "loss": 0.273, "step": 34620 }, { "epoch": 4.17, "learning_rate": 4.171284027945074e-05, "loss": 0.3621, "step": 34630 }, { "epoch": 4.17, "learning_rate": 4.1724885569742234e-05, "loss": 0.3377, "step": 34640 }, { "epoch": 4.17, "learning_rate": 4.173693086003373e-05, "loss": 0.2689, "step": 34650 }, { "epoch": 4.17, "learning_rate": 4.1748976150325225e-05, "loss": 0.1868, "step": 34660 }, { "epoch": 4.18, "learning_rate": 4.176102144061672e-05, "loss": 0.2096, "step": 34670 }, { "epoch": 4.18, "learning_rate": 4.1773066730908215e-05, "loss": 0.3064, "step": 34680 }, { "epoch": 4.18, "learning_rate": 4.178511202119971e-05, "loss": 0.1884, "step": 34690 }, { "epoch": 4.18, "learning_rate": 4.1797157311491205e-05, "loss": 0.1012, "step": 34700 }, { "epoch": 4.18, "learning_rate": 4.180920260178271e-05, "loss": 0.287, "step": 34710 }, { "epoch": 4.18, "learning_rate": 4.18212478920742e-05, "loss": 0.2554, "step": 34720 }, { "epoch": 4.18, "learning_rate": 4.18332931823657e-05, "loss": 0.2213, "step": 34730 }, { "epoch": 4.18, "learning_rate": 4.184533847265719e-05, "loss": 0.2224, "step": 34740 }, { "epoch": 4.19, "learning_rate": 4.185738376294869e-05, "loss": 0.4572, "step": 34750 }, { "epoch": 4.19, "learning_rate": 4.186942905324018e-05, "loss": 0.3385, "step": 34760 }, { "epoch": 4.19, "learning_rate": 4.188147434353168e-05, "loss": 0.3803, "step": 34770 }, { "epoch": 4.19, "learning_rate": 4.1893519633823174e-05, "loss": 0.2067, "step": 34780 }, { "epoch": 4.19, "learning_rate": 4.190556492411467e-05, "loss": 0.2881, "step": 34790 }, { "epoch": 4.19, "learning_rate": 4.1917610214406164e-05, "loss": 0.2447, "step": 34800 }, { "epoch": 4.19, "learning_rate": 4.192965550469766e-05, "loss": 0.1968, "step": 34810 }, { "epoch": 4.19, "learning_rate": 4.194170079498916e-05, "loss": 0.4183, "step": 34820 }, { "epoch": 4.2, "learning_rate": 4.1953746085280656e-05, "loss": 0.1916, "step": 34830 }, { "epoch": 4.2, "learning_rate": 4.196579137557216e-05, "loss": 0.1375, "step": 34840 }, { "epoch": 4.2, "learning_rate": 4.1977836665863654e-05, "loss": 0.2471, "step": 34850 }, { "epoch": 4.2, "learning_rate": 4.198988195615515e-05, "loss": 0.1431, "step": 34860 }, { "epoch": 4.2, "learning_rate": 4.2001927246446644e-05, "loss": 0.1863, "step": 34870 }, { "epoch": 4.2, "learning_rate": 4.201397253673814e-05, "loss": 0.3069, "step": 34880 }, { "epoch": 4.2, "learning_rate": 4.2026017827029634e-05, "loss": 0.2522, "step": 34890 }, { "epoch": 4.2, "learning_rate": 4.203806311732113e-05, "loss": 0.2719, "step": 34900 }, { "epoch": 4.2, "learning_rate": 4.2050108407612625e-05, "loss": 0.3427, "step": 34910 }, { "epoch": 4.21, "learning_rate": 4.206215369790412e-05, "loss": 0.2643, "step": 34920 }, { "epoch": 4.21, "learning_rate": 4.2074198988195615e-05, "loss": 0.4355, "step": 34930 }, { "epoch": 4.21, "learning_rate": 4.208624427848711e-05, "loss": 0.3746, "step": 34940 }, { "epoch": 4.21, "learning_rate": 4.209828956877861e-05, "loss": 0.2947, "step": 34950 }, { "epoch": 4.21, "learning_rate": 4.211033485907011e-05, "loss": 0.1173, "step": 34960 }, { "epoch": 4.21, "learning_rate": 4.21223801493616e-05, "loss": 0.2151, "step": 34970 }, { "epoch": 4.21, "learning_rate": 4.21344254396531e-05, "loss": 0.2599, "step": 34980 }, { "epoch": 4.21, "learning_rate": 4.214647072994459e-05, "loss": 0.1904, "step": 34990 }, { "epoch": 4.22, "learning_rate": 4.215851602023609e-05, "loss": 0.404, "step": 35000 }, { "epoch": 4.22, "learning_rate": 4.217056131052758e-05, "loss": 0.33, "step": 35010 }, { "epoch": 4.22, "learning_rate": 4.218260660081908e-05, "loss": 0.2585, "step": 35020 }, { "epoch": 4.22, "learning_rate": 4.2194651891110574e-05, "loss": 0.3242, "step": 35030 }, { "epoch": 4.22, "learning_rate": 4.220669718140207e-05, "loss": 0.1587, "step": 35040 }, { "epoch": 4.22, "learning_rate": 4.221874247169357e-05, "loss": 0.3058, "step": 35050 }, { "epoch": 4.22, "learning_rate": 4.2230787761985066e-05, "loss": 0.2632, "step": 35060 }, { "epoch": 4.22, "learning_rate": 4.224283305227656e-05, "loss": 0.315, "step": 35070 }, { "epoch": 4.23, "learning_rate": 4.225487834256806e-05, "loss": 0.1706, "step": 35080 }, { "epoch": 4.23, "learning_rate": 4.226692363285956e-05, "loss": 0.3263, "step": 35090 }, { "epoch": 4.23, "learning_rate": 4.2278968923151054e-05, "loss": 0.4642, "step": 35100 }, { "epoch": 4.23, "learning_rate": 4.229101421344255e-05, "loss": 0.3717, "step": 35110 }, { "epoch": 4.23, "learning_rate": 4.2303059503734044e-05, "loss": 0.1077, "step": 35120 }, { "epoch": 4.23, "learning_rate": 4.231510479402554e-05, "loss": 0.2378, "step": 35130 }, { "epoch": 4.23, "learning_rate": 4.2327150084317034e-05, "loss": 0.3399, "step": 35140 }, { "epoch": 4.23, "learning_rate": 4.233919537460853e-05, "loss": 0.2906, "step": 35150 }, { "epoch": 4.23, "learning_rate": 4.2351240664900025e-05, "loss": 0.3618, "step": 35160 }, { "epoch": 4.24, "learning_rate": 4.236328595519152e-05, "loss": 0.2747, "step": 35170 }, { "epoch": 4.24, "learning_rate": 4.237533124548302e-05, "loss": 0.3716, "step": 35180 }, { "epoch": 4.24, "learning_rate": 4.238737653577452e-05, "loss": 0.1685, "step": 35190 }, { "epoch": 4.24, "learning_rate": 4.239942182606601e-05, "loss": 0.1075, "step": 35200 }, { "epoch": 4.24, "learning_rate": 4.241146711635751e-05, "loss": 0.2023, "step": 35210 }, { "epoch": 4.24, "learning_rate": 4.2423512406649e-05, "loss": 0.1142, "step": 35220 }, { "epoch": 4.24, "learning_rate": 4.24355576969405e-05, "loss": 0.1607, "step": 35230 }, { "epoch": 4.24, "learning_rate": 4.244760298723199e-05, "loss": 0.3939, "step": 35240 }, { "epoch": 4.25, "learning_rate": 4.245964827752349e-05, "loss": 0.2126, "step": 35250 }, { "epoch": 4.25, "learning_rate": 4.247169356781498e-05, "loss": 0.2509, "step": 35260 }, { "epoch": 4.25, "learning_rate": 4.248373885810648e-05, "loss": 0.1977, "step": 35270 }, { "epoch": 4.25, "learning_rate": 4.2495784148397974e-05, "loss": 0.3365, "step": 35280 }, { "epoch": 4.25, "learning_rate": 4.2507829438689476e-05, "loss": 0.0871, "step": 35290 }, { "epoch": 4.25, "learning_rate": 4.251987472898097e-05, "loss": 0.593, "step": 35300 }, { "epoch": 4.25, "learning_rate": 4.2531920019272466e-05, "loss": 0.1882, "step": 35310 }, { "epoch": 4.25, "learning_rate": 4.254396530956396e-05, "loss": 0.268, "step": 35320 }, { "epoch": 4.26, "learning_rate": 4.2556010599855456e-05, "loss": 0.1368, "step": 35330 }, { "epoch": 4.26, "learning_rate": 4.256805589014696e-05, "loss": 0.0972, "step": 35340 }, { "epoch": 4.26, "learning_rate": 4.2580101180438453e-05, "loss": 0.3301, "step": 35350 }, { "epoch": 4.26, "learning_rate": 4.259214647072995e-05, "loss": 0.3782, "step": 35360 }, { "epoch": 4.26, "learning_rate": 4.2604191761021444e-05, "loss": 0.3369, "step": 35370 }, { "epoch": 4.26, "learning_rate": 4.261623705131294e-05, "loss": 0.2774, "step": 35380 }, { "epoch": 4.26, "learning_rate": 4.2628282341604434e-05, "loss": 0.1598, "step": 35390 }, { "epoch": 4.26, "learning_rate": 4.264032763189593e-05, "loss": 0.2128, "step": 35400 }, { "epoch": 4.26, "learning_rate": 4.2652372922187425e-05, "loss": 0.2064, "step": 35410 }, { "epoch": 4.27, "learning_rate": 4.2664418212478927e-05, "loss": 0.1388, "step": 35420 }, { "epoch": 4.27, "learning_rate": 4.267646350277042e-05, "loss": 0.2025, "step": 35430 }, { "epoch": 4.27, "learning_rate": 4.268850879306192e-05, "loss": 0.5525, "step": 35440 }, { "epoch": 4.27, "learning_rate": 4.270055408335341e-05, "loss": 0.3266, "step": 35450 }, { "epoch": 4.27, "learning_rate": 4.271259937364491e-05, "loss": 0.1617, "step": 35460 }, { "epoch": 4.27, "learning_rate": 4.27246446639364e-05, "loss": 0.3875, "step": 35470 }, { "epoch": 4.27, "learning_rate": 4.27366899542279e-05, "loss": 0.4432, "step": 35480 }, { "epoch": 4.27, "learning_rate": 4.274873524451939e-05, "loss": 0.1232, "step": 35490 }, { "epoch": 4.28, "learning_rate": 4.276078053481089e-05, "loss": 0.089, "step": 35500 }, { "epoch": 4.28, "learning_rate": 4.277282582510238e-05, "loss": 0.3685, "step": 35510 }, { "epoch": 4.28, "learning_rate": 4.2784871115393885e-05, "loss": 0.3265, "step": 35520 }, { "epoch": 4.28, "learning_rate": 4.279691640568538e-05, "loss": 0.086, "step": 35530 }, { "epoch": 4.28, "learning_rate": 4.2808961695976876e-05, "loss": 0.2044, "step": 35540 }, { "epoch": 4.28, "learning_rate": 4.282100698626837e-05, "loss": 0.0223, "step": 35550 }, { "epoch": 4.28, "learning_rate": 4.2833052276559866e-05, "loss": 0.3998, "step": 35560 }, { "epoch": 4.28, "learning_rate": 4.284509756685136e-05, "loss": 0.1732, "step": 35570 }, { "epoch": 4.29, "learning_rate": 4.2857142857142856e-05, "loss": 0.2048, "step": 35580 }, { "epoch": 4.29, "learning_rate": 4.286918814743435e-05, "loss": 0.3673, "step": 35590 }, { "epoch": 4.29, "learning_rate": 4.2881233437725853e-05, "loss": 0.3091, "step": 35600 }, { "epoch": 4.29, "learning_rate": 4.289327872801735e-05, "loss": 0.1977, "step": 35610 }, { "epoch": 4.29, "learning_rate": 4.2905324018308844e-05, "loss": 0.099, "step": 35620 }, { "epoch": 4.29, "learning_rate": 4.291736930860034e-05, "loss": 0.39, "step": 35630 }, { "epoch": 4.29, "learning_rate": 4.2929414598891834e-05, "loss": 0.3661, "step": 35640 }, { "epoch": 4.29, "learning_rate": 4.2941459889183336e-05, "loss": 0.5066, "step": 35650 }, { "epoch": 4.3, "learning_rate": 4.295350517947483e-05, "loss": 0.2269, "step": 35660 }, { "epoch": 4.3, "learning_rate": 4.2965550469766327e-05, "loss": 0.2134, "step": 35670 }, { "epoch": 4.3, "learning_rate": 4.297759576005782e-05, "loss": 0.3613, "step": 35680 }, { "epoch": 4.3, "learning_rate": 4.298964105034932e-05, "loss": 0.2355, "step": 35690 }, { "epoch": 4.3, "learning_rate": 4.300168634064081e-05, "loss": 0.3032, "step": 35700 }, { "epoch": 4.3, "learning_rate": 4.301373163093231e-05, "loss": 0.2892, "step": 35710 }, { "epoch": 4.3, "learning_rate": 4.30257769212238e-05, "loss": 0.392, "step": 35720 }, { "epoch": 4.3, "learning_rate": 4.30378222115153e-05, "loss": 0.1878, "step": 35730 }, { "epoch": 4.3, "learning_rate": 4.304986750180679e-05, "loss": 0.2843, "step": 35740 }, { "epoch": 4.31, "learning_rate": 4.306191279209829e-05, "loss": 0.5649, "step": 35750 }, { "epoch": 4.31, "learning_rate": 4.307395808238979e-05, "loss": 0.2119, "step": 35760 }, { "epoch": 4.31, "learning_rate": 4.3086003372681285e-05, "loss": 0.2572, "step": 35770 }, { "epoch": 4.31, "learning_rate": 4.309804866297278e-05, "loss": 0.2904, "step": 35780 }, { "epoch": 4.31, "learning_rate": 4.3110093953264276e-05, "loss": 0.2794, "step": 35790 }, { "epoch": 4.31, "learning_rate": 4.312213924355577e-05, "loss": 0.2807, "step": 35800 }, { "epoch": 4.31, "learning_rate": 4.3134184533847266e-05, "loss": 0.2399, "step": 35810 }, { "epoch": 4.31, "learning_rate": 4.314622982413876e-05, "loss": 0.1969, "step": 35820 }, { "epoch": 4.32, "learning_rate": 4.3158275114430256e-05, "loss": 0.1774, "step": 35830 }, { "epoch": 4.32, "learning_rate": 4.317032040472175e-05, "loss": 0.0564, "step": 35840 }, { "epoch": 4.32, "learning_rate": 4.318236569501325e-05, "loss": 0.2285, "step": 35850 }, { "epoch": 4.32, "learning_rate": 4.319441098530474e-05, "loss": 0.353, "step": 35860 }, { "epoch": 4.32, "learning_rate": 4.3206456275596244e-05, "loss": 0.0899, "step": 35870 }, { "epoch": 4.32, "learning_rate": 4.3218501565887746e-05, "loss": 0.3742, "step": 35880 }, { "epoch": 4.32, "learning_rate": 4.323054685617924e-05, "loss": 0.0273, "step": 35890 }, { "epoch": 4.32, "learning_rate": 4.3242592146470736e-05, "loss": 0.363, "step": 35900 }, { "epoch": 4.33, "learning_rate": 4.325463743676223e-05, "loss": 0.2002, "step": 35910 }, { "epoch": 4.33, "learning_rate": 4.3266682727053726e-05, "loss": 0.1672, "step": 35920 }, { "epoch": 4.33, "learning_rate": 4.327872801734522e-05, "loss": 0.5501, "step": 35930 }, { "epoch": 4.33, "learning_rate": 4.329077330763672e-05, "loss": 0.169, "step": 35940 }, { "epoch": 4.33, "learning_rate": 4.330281859792821e-05, "loss": 0.2726, "step": 35950 }, { "epoch": 4.33, "learning_rate": 4.331486388821971e-05, "loss": 0.2268, "step": 35960 }, { "epoch": 4.33, "learning_rate": 4.33269091785112e-05, "loss": 0.2138, "step": 35970 }, { "epoch": 4.33, "learning_rate": 4.33389544688027e-05, "loss": 0.1802, "step": 35980 }, { "epoch": 4.33, "learning_rate": 4.33509997590942e-05, "loss": 0.2475, "step": 35990 }, { "epoch": 4.34, "learning_rate": 4.3363045049385695e-05, "loss": 0.2541, "step": 36000 }, { "epoch": 4.34, "learning_rate": 4.337509033967719e-05, "loss": 0.1657, "step": 36010 }, { "epoch": 4.34, "learning_rate": 4.3387135629968685e-05, "loss": 0.2851, "step": 36020 }, { "epoch": 4.34, "learning_rate": 4.339918092026018e-05, "loss": 0.2052, "step": 36030 }, { "epoch": 4.34, "learning_rate": 4.3411226210551675e-05, "loss": 0.0756, "step": 36040 }, { "epoch": 4.34, "learning_rate": 4.342327150084317e-05, "loss": 0.1689, "step": 36050 }, { "epoch": 4.34, "learning_rate": 4.3435316791134666e-05, "loss": 0.3548, "step": 36060 }, { "epoch": 4.34, "learning_rate": 4.344736208142616e-05, "loss": 0.3876, "step": 36070 }, { "epoch": 4.35, "learning_rate": 4.3459407371717656e-05, "loss": 0.2404, "step": 36080 }, { "epoch": 4.35, "learning_rate": 4.347145266200915e-05, "loss": 0.2429, "step": 36090 }, { "epoch": 4.35, "learning_rate": 4.348349795230065e-05, "loss": 0.1467, "step": 36100 }, { "epoch": 4.35, "learning_rate": 4.349554324259215e-05, "loss": 0.1978, "step": 36110 }, { "epoch": 4.35, "learning_rate": 4.3507588532883644e-05, "loss": 0.0458, "step": 36120 }, { "epoch": 4.35, "learning_rate": 4.3519633823175146e-05, "loss": 0.2649, "step": 36130 }, { "epoch": 4.35, "learning_rate": 4.353167911346664e-05, "loss": 0.2076, "step": 36140 }, { "epoch": 4.35, "learning_rate": 4.3543724403758136e-05, "loss": 0.2791, "step": 36150 }, { "epoch": 4.36, "learning_rate": 4.355576969404963e-05, "loss": 0.3706, "step": 36160 }, { "epoch": 4.36, "learning_rate": 4.3567814984341126e-05, "loss": 0.075, "step": 36170 }, { "epoch": 4.36, "learning_rate": 4.357986027463262e-05, "loss": 0.103, "step": 36180 }, { "epoch": 4.36, "learning_rate": 4.359190556492412e-05, "loss": 0.2825, "step": 36190 }, { "epoch": 4.36, "learning_rate": 4.360395085521561e-05, "loss": 0.2031, "step": 36200 }, { "epoch": 4.36, "learning_rate": 4.361599614550711e-05, "loss": 0.1804, "step": 36210 }, { "epoch": 4.36, "learning_rate": 4.36280414357986e-05, "loss": 0.0889, "step": 36220 }, { "epoch": 4.36, "learning_rate": 4.3640086726090104e-05, "loss": 0.3436, "step": 36230 }, { "epoch": 4.36, "learning_rate": 4.36521320163816e-05, "loss": 0.238, "step": 36240 }, { "epoch": 4.37, "learning_rate": 4.3664177306673095e-05, "loss": 0.1423, "step": 36250 }, { "epoch": 4.37, "learning_rate": 4.367622259696459e-05, "loss": 0.163, "step": 36260 }, { "epoch": 4.37, "learning_rate": 4.3688267887256085e-05, "loss": 0.2892, "step": 36270 }, { "epoch": 4.37, "learning_rate": 4.370031317754758e-05, "loss": 0.0762, "step": 36280 }, { "epoch": 4.37, "learning_rate": 4.3712358467839075e-05, "loss": 0.0243, "step": 36290 }, { "epoch": 4.37, "learning_rate": 4.372440375813057e-05, "loss": 0.0597, "step": 36300 }, { "epoch": 4.37, "learning_rate": 4.3736449048422066e-05, "loss": 0.1657, "step": 36310 }, { "epoch": 4.37, "learning_rate": 4.374849433871356e-05, "loss": 0.3095, "step": 36320 }, { "epoch": 4.38, "learning_rate": 4.376053962900506e-05, "loss": 0.1885, "step": 36330 }, { "epoch": 4.38, "learning_rate": 4.377258491929656e-05, "loss": 0.4176, "step": 36340 }, { "epoch": 4.38, "learning_rate": 4.378463020958805e-05, "loss": 0.5948, "step": 36350 }, { "epoch": 4.38, "learning_rate": 4.379667549987955e-05, "loss": 0.2361, "step": 36360 }, { "epoch": 4.38, "learning_rate": 4.3808720790171044e-05, "loss": 0.462, "step": 36370 }, { "epoch": 4.38, "learning_rate": 4.382076608046254e-05, "loss": 0.1007, "step": 36380 }, { "epoch": 4.38, "learning_rate": 4.383281137075404e-05, "loss": 0.2693, "step": 36390 }, { "epoch": 4.38, "learning_rate": 4.3844856661045536e-05, "loss": 0.243, "step": 36400 }, { "epoch": 4.39, "learning_rate": 4.385690195133703e-05, "loss": 0.2185, "step": 36410 }, { "epoch": 4.39, "learning_rate": 4.3868947241628526e-05, "loss": 0.1573, "step": 36420 }, { "epoch": 4.39, "learning_rate": 4.388099253192002e-05, "loss": 0.361, "step": 36430 }, { "epoch": 4.39, "learning_rate": 4.389303782221152e-05, "loss": 0.3461, "step": 36440 }, { "epoch": 4.39, "learning_rate": 4.390508311250301e-05, "loss": 0.2247, "step": 36450 }, { "epoch": 4.39, "learning_rate": 4.3917128402794514e-05, "loss": 0.1405, "step": 36460 }, { "epoch": 4.39, "learning_rate": 4.392917369308601e-05, "loss": 0.2074, "step": 36470 }, { "epoch": 4.39, "learning_rate": 4.3941218983377504e-05, "loss": 0.2225, "step": 36480 }, { "epoch": 4.4, "learning_rate": 4.3953264273669e-05, "loss": 0.116, "step": 36490 }, { "epoch": 4.4, "learning_rate": 4.3965309563960495e-05, "loss": 0.4674, "step": 36500 }, { "epoch": 4.4, "learning_rate": 4.397735485425199e-05, "loss": 0.2041, "step": 36510 }, { "epoch": 4.4, "learning_rate": 4.3989400144543485e-05, "loss": 0.2481, "step": 36520 }, { "epoch": 4.4, "learning_rate": 4.400144543483498e-05, "loss": 0.1545, "step": 36530 }, { "epoch": 4.4, "learning_rate": 4.4013490725126475e-05, "loss": 0.217, "step": 36540 }, { "epoch": 4.4, "learning_rate": 4.402553601541797e-05, "loss": 0.2324, "step": 36550 }, { "epoch": 4.4, "learning_rate": 4.4037581305709466e-05, "loss": 0.3543, "step": 36560 }, { "epoch": 4.4, "learning_rate": 4.404962659600097e-05, "loss": 0.429, "step": 36570 }, { "epoch": 4.41, "learning_rate": 4.406167188629246e-05, "loss": 0.3969, "step": 36580 }, { "epoch": 4.41, "learning_rate": 4.407371717658396e-05, "loss": 0.1568, "step": 36590 }, { "epoch": 4.41, "learning_rate": 4.408576246687545e-05, "loss": 0.3524, "step": 36600 }, { "epoch": 4.41, "learning_rate": 4.409780775716695e-05, "loss": 0.2715, "step": 36610 }, { "epoch": 4.41, "learning_rate": 4.4109853047458444e-05, "loss": 0.4011, "step": 36620 }, { "epoch": 4.41, "learning_rate": 4.412189833774994e-05, "loss": 0.3286, "step": 36630 }, { "epoch": 4.41, "learning_rate": 4.4133943628041434e-05, "loss": 0.2508, "step": 36640 }, { "epoch": 4.41, "learning_rate": 4.4145988918332936e-05, "loss": 0.1096, "step": 36650 }, { "epoch": 4.42, "learning_rate": 4.415803420862443e-05, "loss": 0.2304, "step": 36660 }, { "epoch": 4.42, "learning_rate": 4.4170079498915926e-05, "loss": 0.3207, "step": 36670 }, { "epoch": 4.42, "learning_rate": 4.418212478920742e-05, "loss": 0.3985, "step": 36680 }, { "epoch": 4.42, "learning_rate": 4.419417007949892e-05, "loss": 0.1007, "step": 36690 }, { "epoch": 4.42, "learning_rate": 4.420621536979042e-05, "loss": 0.2095, "step": 36700 }, { "epoch": 4.42, "learning_rate": 4.4218260660081914e-05, "loss": 0.2333, "step": 36710 }, { "epoch": 4.42, "learning_rate": 4.423030595037341e-05, "loss": 0.1411, "step": 36720 }, { "epoch": 4.42, "learning_rate": 4.4242351240664904e-05, "loss": 0.1767, "step": 36730 }, { "epoch": 4.43, "learning_rate": 4.42543965309564e-05, "loss": 0.321, "step": 36740 }, { "epoch": 4.43, "learning_rate": 4.4266441821247895e-05, "loss": 0.3621, "step": 36750 }, { "epoch": 4.43, "learning_rate": 4.427848711153939e-05, "loss": 0.2616, "step": 36760 }, { "epoch": 4.43, "learning_rate": 4.4290532401830885e-05, "loss": 0.1627, "step": 36770 }, { "epoch": 4.43, "learning_rate": 4.430257769212238e-05, "loss": 0.2903, "step": 36780 }, { "epoch": 4.43, "learning_rate": 4.4314622982413875e-05, "loss": 0.4234, "step": 36790 }, { "epoch": 4.43, "learning_rate": 4.432666827270538e-05, "loss": 0.2649, "step": 36800 }, { "epoch": 4.43, "learning_rate": 4.433871356299687e-05, "loss": 0.4491, "step": 36810 }, { "epoch": 4.43, "learning_rate": 4.435075885328837e-05, "loss": 0.4532, "step": 36820 }, { "epoch": 4.44, "learning_rate": 4.436280414357986e-05, "loss": 0.1327, "step": 36830 }, { "epoch": 4.44, "learning_rate": 4.437484943387136e-05, "loss": 0.3027, "step": 36840 }, { "epoch": 4.44, "learning_rate": 4.438689472416285e-05, "loss": 0.1204, "step": 36850 }, { "epoch": 4.44, "learning_rate": 4.439894001445435e-05, "loss": 0.1811, "step": 36860 }, { "epoch": 4.44, "learning_rate": 4.4410985304745844e-05, "loss": 0.2029, "step": 36870 }, { "epoch": 4.44, "learning_rate": 4.442303059503734e-05, "loss": 0.3184, "step": 36880 }, { "epoch": 4.44, "learning_rate": 4.4435075885328834e-05, "loss": 0.1609, "step": 36890 }, { "epoch": 4.44, "learning_rate": 4.444712117562033e-05, "loss": 0.2699, "step": 36900 }, { "epoch": 4.45, "learning_rate": 4.445916646591183e-05, "loss": 0.354, "step": 36910 }, { "epoch": 4.45, "learning_rate": 4.4471211756203326e-05, "loss": 0.1589, "step": 36920 }, { "epoch": 4.45, "learning_rate": 4.448325704649483e-05, "loss": 0.3103, "step": 36930 }, { "epoch": 4.45, "learning_rate": 4.4495302336786323e-05, "loss": 0.2306, "step": 36940 }, { "epoch": 4.45, "learning_rate": 4.450734762707782e-05, "loss": 0.1168, "step": 36950 }, { "epoch": 4.45, "learning_rate": 4.4519392917369314e-05, "loss": 0.3482, "step": 36960 }, { "epoch": 4.45, "learning_rate": 4.453143820766081e-05, "loss": 0.3636, "step": 36970 }, { "epoch": 4.45, "learning_rate": 4.4543483497952304e-05, "loss": 0.3062, "step": 36980 }, { "epoch": 4.46, "learning_rate": 4.45555287882438e-05, "loss": 0.2258, "step": 36990 }, { "epoch": 4.46, "learning_rate": 4.4567574078535295e-05, "loss": 0.1115, "step": 37000 }, { "epoch": 4.46, "learning_rate": 4.457961936882679e-05, "loss": 0.3165, "step": 37010 }, { "epoch": 4.46, "learning_rate": 4.4591664659118285e-05, "loss": 0.0567, "step": 37020 }, { "epoch": 4.46, "learning_rate": 4.460370994940978e-05, "loss": 0.2371, "step": 37030 }, { "epoch": 4.46, "learning_rate": 4.461575523970128e-05, "loss": 0.2198, "step": 37040 }, { "epoch": 4.46, "learning_rate": 4.462780052999278e-05, "loss": 0.1162, "step": 37050 }, { "epoch": 4.46, "learning_rate": 4.463984582028427e-05, "loss": 0.4547, "step": 37060 }, { "epoch": 4.46, "learning_rate": 4.465189111057577e-05, "loss": 0.3516, "step": 37070 }, { "epoch": 4.47, "learning_rate": 4.466393640086726e-05, "loss": 0.2908, "step": 37080 }, { "epoch": 4.47, "learning_rate": 4.467598169115876e-05, "loss": 0.3206, "step": 37090 }, { "epoch": 4.47, "learning_rate": 4.468802698145025e-05, "loss": 0.315, "step": 37100 }, { "epoch": 4.47, "learning_rate": 4.470007227174175e-05, "loss": 0.2428, "step": 37110 }, { "epoch": 4.47, "learning_rate": 4.4712117562033244e-05, "loss": 0.33, "step": 37120 }, { "epoch": 4.47, "learning_rate": 4.472416285232474e-05, "loss": 0.2741, "step": 37130 }, { "epoch": 4.47, "learning_rate": 4.4736208142616234e-05, "loss": 0.2572, "step": 37140 }, { "epoch": 4.47, "learning_rate": 4.4748253432907736e-05, "loss": 0.3289, "step": 37150 }, { "epoch": 4.48, "learning_rate": 4.476029872319923e-05, "loss": 0.1985, "step": 37160 }, { "epoch": 4.48, "learning_rate": 4.4772344013490726e-05, "loss": 0.315, "step": 37170 }, { "epoch": 4.48, "learning_rate": 4.478438930378223e-05, "loss": 0.4034, "step": 37180 }, { "epoch": 4.48, "learning_rate": 4.479643459407372e-05, "loss": 0.3571, "step": 37190 }, { "epoch": 4.48, "learning_rate": 4.480847988436522e-05, "loss": 0.0749, "step": 37200 }, { "epoch": 4.48, "learning_rate": 4.4820525174656714e-05, "loss": 0.118, "step": 37210 }, { "epoch": 4.48, "learning_rate": 4.483257046494821e-05, "loss": 0.3522, "step": 37220 }, { "epoch": 4.48, "learning_rate": 4.4844615755239704e-05, "loss": 0.1337, "step": 37230 }, { "epoch": 4.49, "learning_rate": 4.48566610455312e-05, "loss": 0.094, "step": 37240 }, { "epoch": 4.49, "learning_rate": 4.4868706335822694e-05, "loss": 0.1713, "step": 37250 }, { "epoch": 4.49, "learning_rate": 4.488075162611419e-05, "loss": 0.2844, "step": 37260 }, { "epoch": 4.49, "learning_rate": 4.489279691640569e-05, "loss": 0.2115, "step": 37270 }, { "epoch": 4.49, "learning_rate": 4.490484220669719e-05, "loss": 0.1393, "step": 37280 }, { "epoch": 4.49, "learning_rate": 4.491688749698868e-05, "loss": 0.3343, "step": 37290 }, { "epoch": 4.49, "learning_rate": 4.492893278728018e-05, "loss": 0.2221, "step": 37300 }, { "epoch": 4.49, "learning_rate": 4.494097807757167e-05, "loss": 0.1967, "step": 37310 }, { "epoch": 4.5, "learning_rate": 4.495302336786317e-05, "loss": 0.3437, "step": 37320 }, { "epoch": 4.5, "learning_rate": 4.496506865815466e-05, "loss": 0.3612, "step": 37330 }, { "epoch": 4.5, "learning_rate": 4.497711394844616e-05, "loss": 0.0685, "step": 37340 }, { "epoch": 4.5, "learning_rate": 4.498915923873765e-05, "loss": 0.3629, "step": 37350 }, { "epoch": 4.5, "learning_rate": 4.500120452902915e-05, "loss": 0.2345, "step": 37360 }, { "epoch": 4.5, "learning_rate": 4.5013249819320643e-05, "loss": 0.2778, "step": 37370 }, { "epoch": 4.5, "learning_rate": 4.5025295109612145e-05, "loss": 0.2619, "step": 37380 }, { "epoch": 4.5, "learning_rate": 4.503734039990364e-05, "loss": 0.335, "step": 37390 }, { "epoch": 4.5, "learning_rate": 4.5049385690195136e-05, "loss": 0.0732, "step": 37400 }, { "epoch": 4.51, "learning_rate": 4.506143098048663e-05, "loss": 0.3391, "step": 37410 }, { "epoch": 4.51, "learning_rate": 4.5073476270778126e-05, "loss": 0.2749, "step": 37420 }, { "epoch": 4.51, "learning_rate": 4.508552156106962e-05, "loss": 0.2605, "step": 37430 }, { "epoch": 4.51, "learning_rate": 4.509756685136112e-05, "loss": 0.2746, "step": 37440 }, { "epoch": 4.51, "learning_rate": 4.510961214165262e-05, "loss": 0.2604, "step": 37450 }, { "epoch": 4.51, "learning_rate": 4.5121657431944114e-05, "loss": 0.1286, "step": 37460 }, { "epoch": 4.51, "learning_rate": 4.513370272223561e-05, "loss": 0.2665, "step": 37470 }, { "epoch": 4.51, "learning_rate": 4.5145748012527104e-05, "loss": 0.2659, "step": 37480 }, { "epoch": 4.52, "learning_rate": 4.51577933028186e-05, "loss": 0.2172, "step": 37490 }, { "epoch": 4.52, "learning_rate": 4.5169838593110094e-05, "loss": 0.0289, "step": 37500 }, { "epoch": 4.52, "learning_rate": 4.5181883883401596e-05, "loss": 0.3102, "step": 37510 }, { "epoch": 4.52, "learning_rate": 4.519392917369309e-05, "loss": 0.2108, "step": 37520 }, { "epoch": 4.52, "learning_rate": 4.520597446398459e-05, "loss": 0.1933, "step": 37530 }, { "epoch": 4.52, "learning_rate": 4.521801975427608e-05, "loss": 0.0755, "step": 37540 }, { "epoch": 4.52, "learning_rate": 4.523006504456758e-05, "loss": 0.3773, "step": 37550 }, { "epoch": 4.52, "learning_rate": 4.524211033485907e-05, "loss": 0.1239, "step": 37560 }, { "epoch": 4.53, "learning_rate": 4.525415562515057e-05, "loss": 0.1945, "step": 37570 }, { "epoch": 4.53, "learning_rate": 4.526620091544206e-05, "loss": 0.4271, "step": 37580 }, { "epoch": 4.53, "learning_rate": 4.527824620573356e-05, "loss": 0.3711, "step": 37590 }, { "epoch": 4.53, "learning_rate": 4.529029149602505e-05, "loss": 0.3843, "step": 37600 }, { "epoch": 4.53, "learning_rate": 4.530233678631655e-05, "loss": 0.2226, "step": 37610 }, { "epoch": 4.53, "learning_rate": 4.531438207660805e-05, "loss": 0.1228, "step": 37620 }, { "epoch": 4.53, "learning_rate": 4.5326427366899545e-05, "loss": 0.1772, "step": 37630 }, { "epoch": 4.53, "learning_rate": 4.533847265719104e-05, "loss": 0.2186, "step": 37640 }, { "epoch": 4.53, "learning_rate": 4.5350517947482536e-05, "loss": 0.2142, "step": 37650 }, { "epoch": 4.54, "learning_rate": 4.536256323777403e-05, "loss": 0.302, "step": 37660 }, { "epoch": 4.54, "learning_rate": 4.5374608528065526e-05, "loss": 0.4555, "step": 37670 }, { "epoch": 4.54, "learning_rate": 4.538665381835702e-05, "loss": 0.228, "step": 37680 }, { "epoch": 4.54, "learning_rate": 4.5398699108648516e-05, "loss": 0.2125, "step": 37690 }, { "epoch": 4.54, "learning_rate": 4.541074439894001e-05, "loss": 0.216, "step": 37700 }, { "epoch": 4.54, "learning_rate": 4.5422789689231514e-05, "loss": 0.2761, "step": 37710 }, { "epoch": 4.54, "learning_rate": 4.543483497952301e-05, "loss": 0.1678, "step": 37720 }, { "epoch": 4.54, "learning_rate": 4.5446880269814504e-05, "loss": 0.4802, "step": 37730 }, { "epoch": 4.55, "learning_rate": 4.5458925560106006e-05, "loss": 0.2093, "step": 37740 }, { "epoch": 4.55, "learning_rate": 4.54709708503975e-05, "loss": 0.1337, "step": 37750 }, { "epoch": 4.55, "learning_rate": 4.5483016140688996e-05, "loss": 0.2551, "step": 37760 }, { "epoch": 4.55, "learning_rate": 4.549506143098049e-05, "loss": 0.1871, "step": 37770 }, { "epoch": 4.55, "learning_rate": 4.550710672127199e-05, "loss": 0.2901, "step": 37780 }, { "epoch": 4.55, "learning_rate": 4.551915201156348e-05, "loss": 0.3214, "step": 37790 }, { "epoch": 4.55, "learning_rate": 4.553119730185498e-05, "loss": 0.0989, "step": 37800 }, { "epoch": 4.55, "learning_rate": 4.554324259214647e-05, "loss": 0.2702, "step": 37810 }, { "epoch": 4.56, "learning_rate": 4.555528788243797e-05, "loss": 0.3401, "step": 37820 }, { "epoch": 4.56, "learning_rate": 4.556733317272946e-05, "loss": 0.3097, "step": 37830 }, { "epoch": 4.56, "learning_rate": 4.557937846302096e-05, "loss": 0.4212, "step": 37840 }, { "epoch": 4.56, "learning_rate": 4.559142375331246e-05, "loss": 0.1121, "step": 37850 }, { "epoch": 4.56, "learning_rate": 4.5603469043603955e-05, "loss": 0.2799, "step": 37860 }, { "epoch": 4.56, "learning_rate": 4.561551433389545e-05, "loss": 0.1932, "step": 37870 }, { "epoch": 4.56, "learning_rate": 4.5627559624186945e-05, "loss": 0.1169, "step": 37880 }, { "epoch": 4.56, "learning_rate": 4.563960491447844e-05, "loss": 0.4246, "step": 37890 }, { "epoch": 4.56, "learning_rate": 4.5651650204769936e-05, "loss": 0.1919, "step": 37900 }, { "epoch": 4.57, "learning_rate": 4.566369549506143e-05, "loss": 0.3077, "step": 37910 }, { "epoch": 4.57, "learning_rate": 4.5675740785352926e-05, "loss": 0.0646, "step": 37920 }, { "epoch": 4.57, "learning_rate": 4.568778607564442e-05, "loss": 0.1999, "step": 37930 }, { "epoch": 4.57, "learning_rate": 4.5699831365935916e-05, "loss": 0.2756, "step": 37940 }, { "epoch": 4.57, "learning_rate": 4.571187665622741e-05, "loss": 0.2845, "step": 37950 }, { "epoch": 4.57, "learning_rate": 4.5723921946518914e-05, "loss": 0.1914, "step": 37960 }, { "epoch": 4.57, "learning_rate": 4.573596723681041e-05, "loss": 0.116, "step": 37970 }, { "epoch": 4.57, "learning_rate": 4.574801252710191e-05, "loss": 0.323, "step": 37980 }, { "epoch": 4.58, "learning_rate": 4.5760057817393406e-05, "loss": 0.2239, "step": 37990 }, { "epoch": 4.58, "learning_rate": 4.57721031076849e-05, "loss": 0.2846, "step": 38000 }, { "epoch": 4.58, "learning_rate": 4.5784148397976396e-05, "loss": 0.2737, "step": 38010 }, { "epoch": 4.58, "learning_rate": 4.579619368826789e-05, "loss": 0.2889, "step": 38020 }, { "epoch": 4.58, "learning_rate": 4.580823897855939e-05, "loss": 0.072, "step": 38030 }, { "epoch": 4.58, "learning_rate": 4.582028426885088e-05, "loss": 0.3168, "step": 38040 }, { "epoch": 4.58, "learning_rate": 4.583232955914238e-05, "loss": 0.3255, "step": 38050 }, { "epoch": 4.58, "learning_rate": 4.584437484943387e-05, "loss": 0.2995, "step": 38060 }, { "epoch": 4.59, "learning_rate": 4.585642013972537e-05, "loss": 0.4198, "step": 38070 }, { "epoch": 4.59, "learning_rate": 4.586846543001686e-05, "loss": 0.291, "step": 38080 }, { "epoch": 4.59, "learning_rate": 4.5880510720308365e-05, "loss": 0.3397, "step": 38090 }, { "epoch": 4.59, "learning_rate": 4.589255601059986e-05, "loss": 0.1235, "step": 38100 }, { "epoch": 4.59, "learning_rate": 4.5904601300891355e-05, "loss": 0.2612, "step": 38110 }, { "epoch": 4.59, "learning_rate": 4.591664659118285e-05, "loss": 0.4197, "step": 38120 }, { "epoch": 4.59, "learning_rate": 4.5928691881474345e-05, "loss": 0.1137, "step": 38130 }, { "epoch": 4.59, "learning_rate": 4.594073717176584e-05, "loss": 0.3811, "step": 38140 }, { "epoch": 4.6, "learning_rate": 4.5952782462057336e-05, "loss": 0.3386, "step": 38150 }, { "epoch": 4.6, "learning_rate": 4.596482775234883e-05, "loss": 0.3137, "step": 38160 }, { "epoch": 4.6, "learning_rate": 4.5976873042640326e-05, "loss": 0.1106, "step": 38170 }, { "epoch": 4.6, "learning_rate": 4.598891833293182e-05, "loss": 0.2022, "step": 38180 }, { "epoch": 4.6, "learning_rate": 4.600096362322332e-05, "loss": 0.2309, "step": 38190 }, { "epoch": 4.6, "learning_rate": 4.601300891351482e-05, "loss": 0.218, "step": 38200 }, { "epoch": 4.6, "learning_rate": 4.6025054203806314e-05, "loss": 0.294, "step": 38210 }, { "epoch": 4.6, "learning_rate": 4.603709949409781e-05, "loss": 0.2929, "step": 38220 }, { "epoch": 4.6, "learning_rate": 4.604914478438931e-05, "loss": 0.2363, "step": 38230 }, { "epoch": 4.61, "learning_rate": 4.6061190074680806e-05, "loss": 0.2947, "step": 38240 }, { "epoch": 4.61, "learning_rate": 4.60732353649723e-05, "loss": 0.178, "step": 38250 }, { "epoch": 4.61, "learning_rate": 4.6085280655263796e-05, "loss": 0.3071, "step": 38260 }, { "epoch": 4.61, "learning_rate": 4.609732594555529e-05, "loss": 0.1764, "step": 38270 }, { "epoch": 4.61, "learning_rate": 4.610937123584679e-05, "loss": 0.2749, "step": 38280 }, { "epoch": 4.61, "learning_rate": 4.612141652613828e-05, "loss": 0.1701, "step": 38290 }, { "epoch": 4.61, "learning_rate": 4.613346181642978e-05, "loss": 0.2564, "step": 38300 }, { "epoch": 4.61, "learning_rate": 4.614550710672127e-05, "loss": 0.256, "step": 38310 }, { "epoch": 4.62, "learning_rate": 4.6157552397012774e-05, "loss": 0.358, "step": 38320 }, { "epoch": 4.62, "learning_rate": 4.616959768730427e-05, "loss": 0.1548, "step": 38330 }, { "epoch": 4.62, "learning_rate": 4.6181642977595765e-05, "loss": 0.1264, "step": 38340 }, { "epoch": 4.62, "learning_rate": 4.619368826788726e-05, "loss": 0.3961, "step": 38350 }, { "epoch": 4.62, "learning_rate": 4.6205733558178755e-05, "loss": 0.2676, "step": 38360 }, { "epoch": 4.62, "learning_rate": 4.621777884847025e-05, "loss": 0.3904, "step": 38370 }, { "epoch": 4.62, "learning_rate": 4.6229824138761745e-05, "loss": 0.3816, "step": 38380 }, { "epoch": 4.62, "learning_rate": 4.624186942905324e-05, "loss": 0.2795, "step": 38390 }, { "epoch": 4.63, "learning_rate": 4.6253914719344736e-05, "loss": 0.1814, "step": 38400 }, { "epoch": 4.63, "learning_rate": 4.626596000963623e-05, "loss": 0.3808, "step": 38410 }, { "epoch": 4.63, "learning_rate": 4.6278005299927726e-05, "loss": 0.3767, "step": 38420 }, { "epoch": 4.63, "learning_rate": 4.629005059021923e-05, "loss": 0.2624, "step": 38430 }, { "epoch": 4.63, "learning_rate": 4.630209588051072e-05, "loss": 0.206, "step": 38440 }, { "epoch": 4.63, "learning_rate": 4.631414117080222e-05, "loss": 0.1811, "step": 38450 }, { "epoch": 4.63, "learning_rate": 4.6326186461093713e-05, "loss": 0.1415, "step": 38460 }, { "epoch": 4.63, "learning_rate": 4.633823175138521e-05, "loss": 0.1309, "step": 38470 }, { "epoch": 4.63, "learning_rate": 4.6350277041676704e-05, "loss": 0.1874, "step": 38480 }, { "epoch": 4.64, "learning_rate": 4.6362322331968206e-05, "loss": 0.5427, "step": 38490 }, { "epoch": 4.64, "learning_rate": 4.63743676222597e-05, "loss": 0.0691, "step": 38500 }, { "epoch": 4.64, "learning_rate": 4.6386412912551196e-05, "loss": 0.1615, "step": 38510 }, { "epoch": 4.64, "learning_rate": 4.639845820284269e-05, "loss": 0.3116, "step": 38520 }, { "epoch": 4.64, "learning_rate": 4.6410503493134187e-05, "loss": 0.2539, "step": 38530 }, { "epoch": 4.64, "learning_rate": 4.642254878342568e-05, "loss": 0.2328, "step": 38540 }, { "epoch": 4.64, "learning_rate": 4.643459407371718e-05, "loss": 0.1626, "step": 38550 }, { "epoch": 4.64, "learning_rate": 4.644663936400868e-05, "loss": 0.1412, "step": 38560 }, { "epoch": 4.65, "learning_rate": 4.6458684654300174e-05, "loss": 0.1373, "step": 38570 }, { "epoch": 4.65, "learning_rate": 4.647072994459167e-05, "loss": 0.325, "step": 38580 }, { "epoch": 4.65, "learning_rate": 4.6482775234883164e-05, "loss": 0.2564, "step": 38590 }, { "epoch": 4.65, "learning_rate": 4.649482052517466e-05, "loss": 0.1454, "step": 38600 }, { "epoch": 4.65, "learning_rate": 4.6506865815466155e-05, "loss": 0.3848, "step": 38610 }, { "epoch": 4.65, "learning_rate": 4.651891110575765e-05, "loss": 0.0814, "step": 38620 }, { "epoch": 4.65, "learning_rate": 4.6530956396049145e-05, "loss": 0.1807, "step": 38630 }, { "epoch": 4.65, "learning_rate": 4.654300168634064e-05, "loss": 0.159, "step": 38640 }, { "epoch": 4.66, "learning_rate": 4.6555046976632136e-05, "loss": 0.384, "step": 38650 }, { "epoch": 4.66, "learning_rate": 4.656709226692364e-05, "loss": 0.2639, "step": 38660 }, { "epoch": 4.66, "learning_rate": 4.657913755721513e-05, "loss": 0.2527, "step": 38670 }, { "epoch": 4.66, "learning_rate": 4.659118284750663e-05, "loss": 0.1406, "step": 38680 }, { "epoch": 4.66, "learning_rate": 4.660322813779812e-05, "loss": 0.2205, "step": 38690 }, { "epoch": 4.66, "learning_rate": 4.661527342808962e-05, "loss": 0.5083, "step": 38700 }, { "epoch": 4.66, "learning_rate": 4.6627318718381113e-05, "loss": 0.2509, "step": 38710 }, { "epoch": 4.66, "learning_rate": 4.663936400867261e-05, "loss": 0.3641, "step": 38720 }, { "epoch": 4.66, "learning_rate": 4.6651409298964104e-05, "loss": 0.3631, "step": 38730 }, { "epoch": 4.67, "learning_rate": 4.66634545892556e-05, "loss": 0.2473, "step": 38740 }, { "epoch": 4.67, "learning_rate": 4.6675499879547094e-05, "loss": 0.381, "step": 38750 }, { "epoch": 4.67, "learning_rate": 4.6687545169838596e-05, "loss": 0.1776, "step": 38760 }, { "epoch": 4.67, "learning_rate": 4.669959046013009e-05, "loss": 0.3014, "step": 38770 }, { "epoch": 4.67, "learning_rate": 4.6711635750421587e-05, "loss": 0.2411, "step": 38780 }, { "epoch": 4.67, "learning_rate": 4.672368104071309e-05, "loss": 0.2537, "step": 38790 }, { "epoch": 4.67, "learning_rate": 4.6735726331004584e-05, "loss": 0.2217, "step": 38800 }, { "epoch": 4.67, "learning_rate": 4.674777162129608e-05, "loss": 0.1692, "step": 38810 }, { "epoch": 4.68, "learning_rate": 4.6759816911587574e-05, "loss": 0.3222, "step": 38820 }, { "epoch": 4.68, "learning_rate": 4.677186220187907e-05, "loss": 0.2248, "step": 38830 }, { "epoch": 4.68, "learning_rate": 4.6783907492170564e-05, "loss": 0.4162, "step": 38840 }, { "epoch": 4.68, "learning_rate": 4.679595278246206e-05, "loss": 0.1155, "step": 38850 }, { "epoch": 4.68, "learning_rate": 4.6807998072753555e-05, "loss": 0.3284, "step": 38860 }, { "epoch": 4.68, "learning_rate": 4.682004336304505e-05, "loss": 0.4412, "step": 38870 }, { "epoch": 4.68, "learning_rate": 4.6832088653336545e-05, "loss": 0.2969, "step": 38880 }, { "epoch": 4.68, "learning_rate": 4.684413394362804e-05, "loss": 0.3358, "step": 38890 }, { "epoch": 4.69, "learning_rate": 4.685617923391954e-05, "loss": 0.2485, "step": 38900 }, { "epoch": 4.69, "learning_rate": 4.686822452421104e-05, "loss": 0.2555, "step": 38910 }, { "epoch": 4.69, "learning_rate": 4.688026981450253e-05, "loss": 0.1909, "step": 38920 }, { "epoch": 4.69, "learning_rate": 4.689231510479403e-05, "loss": 0.2342, "step": 38930 }, { "epoch": 4.69, "learning_rate": 4.690436039508552e-05, "loss": 0.1791, "step": 38940 }, { "epoch": 4.69, "learning_rate": 4.691640568537702e-05, "loss": 0.2582, "step": 38950 }, { "epoch": 4.69, "learning_rate": 4.6928450975668513e-05, "loss": 0.114, "step": 38960 }, { "epoch": 4.69, "learning_rate": 4.694049626596001e-05, "loss": 0.4115, "step": 38970 }, { "epoch": 4.69, "learning_rate": 4.6952541556251504e-05, "loss": 0.3574, "step": 38980 }, { "epoch": 4.7, "learning_rate": 4.6964586846543e-05, "loss": 0.2493, "step": 38990 }, { "epoch": 4.7, "learning_rate": 4.6976632136834494e-05, "loss": 0.0805, "step": 39000 }, { "epoch": 4.7, "learning_rate": 4.6988677427125996e-05, "loss": 0.3367, "step": 39010 }, { "epoch": 4.7, "learning_rate": 4.700072271741749e-05, "loss": 0.4513, "step": 39020 }, { "epoch": 4.7, "learning_rate": 4.701276800770899e-05, "loss": 0.3846, "step": 39030 }, { "epoch": 4.7, "learning_rate": 4.702481329800049e-05, "loss": 0.1616, "step": 39040 }, { "epoch": 4.7, "learning_rate": 4.7036858588291984e-05, "loss": 0.3118, "step": 39050 }, { "epoch": 4.7, "learning_rate": 4.704890387858348e-05, "loss": 0.1291, "step": 39060 }, { "epoch": 4.71, "learning_rate": 4.7060949168874974e-05, "loss": 0.5088, "step": 39070 }, { "epoch": 4.71, "learning_rate": 4.707299445916647e-05, "loss": 0.2192, "step": 39080 }, { "epoch": 4.71, "learning_rate": 4.7085039749457964e-05, "loss": 0.3907, "step": 39090 }, { "epoch": 4.71, "learning_rate": 4.709708503974946e-05, "loss": 0.1095, "step": 39100 }, { "epoch": 4.71, "learning_rate": 4.7109130330040955e-05, "loss": 0.3246, "step": 39110 }, { "epoch": 4.71, "learning_rate": 4.712117562033245e-05, "loss": 0.2343, "step": 39120 }, { "epoch": 4.71, "learning_rate": 4.713322091062395e-05, "loss": 0.1757, "step": 39130 }, { "epoch": 4.71, "learning_rate": 4.714526620091545e-05, "loss": 0.1963, "step": 39140 }, { "epoch": 4.72, "learning_rate": 4.715731149120694e-05, "loss": 0.3622, "step": 39150 }, { "epoch": 4.72, "learning_rate": 4.716935678149844e-05, "loss": 0.3445, "step": 39160 }, { "epoch": 4.72, "learning_rate": 4.718140207178993e-05, "loss": 0.1758, "step": 39170 }, { "epoch": 4.72, "learning_rate": 4.719344736208143e-05, "loss": 0.3365, "step": 39180 }, { "epoch": 4.72, "learning_rate": 4.720549265237292e-05, "loss": 0.2765, "step": 39190 }, { "epoch": 4.72, "learning_rate": 4.721753794266442e-05, "loss": 0.2606, "step": 39200 }, { "epoch": 4.72, "learning_rate": 4.722958323295591e-05, "loss": 0.1623, "step": 39210 }, { "epoch": 4.72, "learning_rate": 4.724162852324741e-05, "loss": 0.2381, "step": 39220 }, { "epoch": 4.73, "learning_rate": 4.7253673813538904e-05, "loss": 0.1543, "step": 39230 }, { "epoch": 4.73, "learning_rate": 4.7265719103830406e-05, "loss": 0.2891, "step": 39240 }, { "epoch": 4.73, "learning_rate": 4.72777643941219e-05, "loss": 0.3069, "step": 39250 }, { "epoch": 4.73, "learning_rate": 4.7289809684413396e-05, "loss": 0.0911, "step": 39260 }, { "epoch": 4.73, "learning_rate": 4.730185497470489e-05, "loss": 0.3672, "step": 39270 }, { "epoch": 4.73, "learning_rate": 4.731390026499639e-05, "loss": 0.4102, "step": 39280 }, { "epoch": 4.73, "learning_rate": 4.732594555528789e-05, "loss": 0.1742, "step": 39290 }, { "epoch": 4.73, "learning_rate": 4.7337990845579384e-05, "loss": 0.4201, "step": 39300 }, { "epoch": 4.73, "learning_rate": 4.735003613587088e-05, "loss": 0.4159, "step": 39310 }, { "epoch": 4.74, "learning_rate": 4.7362081426162374e-05, "loss": 0.2523, "step": 39320 }, { "epoch": 4.74, "learning_rate": 4.737412671645387e-05, "loss": 0.2951, "step": 39330 }, { "epoch": 4.74, "learning_rate": 4.7386172006745364e-05, "loss": 0.1453, "step": 39340 }, { "epoch": 4.74, "learning_rate": 4.739821729703686e-05, "loss": 0.188, "step": 39350 }, { "epoch": 4.74, "learning_rate": 4.7410262587328355e-05, "loss": 0.2381, "step": 39360 }, { "epoch": 4.74, "learning_rate": 4.742230787761986e-05, "loss": 0.3507, "step": 39370 }, { "epoch": 4.74, "learning_rate": 4.743435316791135e-05, "loss": 0.2198, "step": 39380 }, { "epoch": 4.74, "learning_rate": 4.744639845820285e-05, "loss": 0.2701, "step": 39390 }, { "epoch": 4.75, "learning_rate": 4.745844374849434e-05, "loss": 0.0925, "step": 39400 }, { "epoch": 4.75, "learning_rate": 4.747048903878584e-05, "loss": 0.17, "step": 39410 }, { "epoch": 4.75, "learning_rate": 4.748253432907733e-05, "loss": 0.404, "step": 39420 }, { "epoch": 4.75, "learning_rate": 4.749457961936883e-05, "loss": 0.3197, "step": 39430 }, { "epoch": 4.75, "learning_rate": 4.750662490966032e-05, "loss": 0.4805, "step": 39440 }, { "epoch": 4.75, "learning_rate": 4.751867019995182e-05, "loss": 0.2292, "step": 39450 }, { "epoch": 4.75, "learning_rate": 4.753071549024331e-05, "loss": 0.1112, "step": 39460 }, { "epoch": 4.75, "learning_rate": 4.754276078053481e-05, "loss": 0.1421, "step": 39470 }, { "epoch": 4.76, "learning_rate": 4.755480607082631e-05, "loss": 0.2773, "step": 39480 }, { "epoch": 4.76, "learning_rate": 4.7566851361117806e-05, "loss": 0.238, "step": 39490 }, { "epoch": 4.76, "learning_rate": 4.75788966514093e-05, "loss": 0.4375, "step": 39500 }, { "epoch": 4.76, "learning_rate": 4.7590941941700796e-05, "loss": 0.2647, "step": 39510 }, { "epoch": 4.76, "learning_rate": 4.760298723199229e-05, "loss": 0.3542, "step": 39520 }, { "epoch": 4.76, "learning_rate": 4.7615032522283786e-05, "loss": 0.2566, "step": 39530 }, { "epoch": 4.76, "learning_rate": 4.762707781257529e-05, "loss": 0.0921, "step": 39540 }, { "epoch": 4.76, "learning_rate": 4.7639123102866784e-05, "loss": 0.0273, "step": 39550 }, { "epoch": 4.76, "learning_rate": 4.765116839315828e-05, "loss": 0.1625, "step": 39560 }, { "epoch": 4.77, "learning_rate": 4.7663213683449774e-05, "loss": 0.1007, "step": 39570 }, { "epoch": 4.77, "learning_rate": 4.767525897374127e-05, "loss": 0.2033, "step": 39580 }, { "epoch": 4.77, "learning_rate": 4.7687304264032764e-05, "loss": 0.2559, "step": 39590 }, { "epoch": 4.77, "learning_rate": 4.7699349554324266e-05, "loss": 0.0627, "step": 39600 }, { "epoch": 4.77, "learning_rate": 4.771139484461576e-05, "loss": 0.1555, "step": 39610 }, { "epoch": 4.77, "learning_rate": 4.7723440134907257e-05, "loss": 0.284, "step": 39620 }, { "epoch": 4.77, "learning_rate": 4.773548542519875e-05, "loss": 0.2138, "step": 39630 }, { "epoch": 4.77, "learning_rate": 4.774753071549025e-05, "loss": 0.3323, "step": 39640 }, { "epoch": 4.78, "learning_rate": 4.775957600578174e-05, "loss": 0.2948, "step": 39650 }, { "epoch": 4.78, "learning_rate": 4.777162129607324e-05, "loss": 0.3255, "step": 39660 }, { "epoch": 4.78, "learning_rate": 4.778366658636473e-05, "loss": 0.3457, "step": 39670 }, { "epoch": 4.78, "learning_rate": 4.779571187665623e-05, "loss": 0.1822, "step": 39680 }, { "epoch": 4.78, "learning_rate": 4.780775716694772e-05, "loss": 0.3229, "step": 39690 }, { "epoch": 4.78, "learning_rate": 4.781980245723922e-05, "loss": 0.1927, "step": 39700 }, { "epoch": 4.78, "learning_rate": 4.783184774753072e-05, "loss": 0.1177, "step": 39710 }, { "epoch": 4.78, "learning_rate": 4.7843893037822215e-05, "loss": 0.1644, "step": 39720 }, { "epoch": 4.79, "learning_rate": 4.785593832811371e-05, "loss": 0.3014, "step": 39730 }, { "epoch": 4.79, "learning_rate": 4.7867983618405206e-05, "loss": 0.2359, "step": 39740 }, { "epoch": 4.79, "learning_rate": 4.78800289086967e-05, "loss": 0.3562, "step": 39750 }, { "epoch": 4.79, "learning_rate": 4.7892074198988196e-05, "loss": 0.4367, "step": 39760 }, { "epoch": 4.79, "learning_rate": 4.790411948927969e-05, "loss": 0.4191, "step": 39770 }, { "epoch": 4.79, "learning_rate": 4.7916164779571186e-05, "loss": 0.2267, "step": 39780 }, { "epoch": 4.79, "learning_rate": 4.792821006986268e-05, "loss": 0.1633, "step": 39790 }, { "epoch": 4.79, "learning_rate": 4.794025536015418e-05, "loss": 0.1223, "step": 39800 }, { "epoch": 4.79, "learning_rate": 4.795230065044568e-05, "loss": 0.2787, "step": 39810 }, { "epoch": 4.8, "learning_rate": 4.7964345940737174e-05, "loss": 0.2278, "step": 39820 }, { "epoch": 4.8, "learning_rate": 4.797639123102867e-05, "loss": 0.1735, "step": 39830 }, { "epoch": 4.8, "learning_rate": 4.798843652132017e-05, "loss": 0.1615, "step": 39840 }, { "epoch": 4.8, "learning_rate": 4.8000481811611666e-05, "loss": 0.3231, "step": 39850 }, { "epoch": 4.8, "learning_rate": 4.801252710190316e-05, "loss": 0.2421, "step": 39860 }, { "epoch": 4.8, "learning_rate": 4.8024572392194657e-05, "loss": 0.1981, "step": 39870 }, { "epoch": 4.8, "learning_rate": 4.803661768248615e-05, "loss": 0.2372, "step": 39880 }, { "epoch": 4.8, "learning_rate": 4.804866297277765e-05, "loss": 0.3957, "step": 39890 }, { "epoch": 4.81, "learning_rate": 4.806070826306914e-05, "loss": 0.1762, "step": 39900 }, { "epoch": 4.81, "learning_rate": 4.807275355336064e-05, "loss": 0.2971, "step": 39910 }, { "epoch": 4.81, "learning_rate": 4.808479884365213e-05, "loss": 0.3588, "step": 39920 }, { "epoch": 4.81, "learning_rate": 4.809684413394363e-05, "loss": 0.4163, "step": 39930 }, { "epoch": 4.81, "learning_rate": 4.810888942423513e-05, "loss": 0.2094, "step": 39940 }, { "epoch": 4.81, "learning_rate": 4.8120934714526625e-05, "loss": 0.4248, "step": 39950 }, { "epoch": 4.81, "learning_rate": 4.813298000481812e-05, "loss": 0.3472, "step": 39960 }, { "epoch": 4.81, "learning_rate": 4.8145025295109615e-05, "loss": 0.3238, "step": 39970 }, { "epoch": 4.82, "learning_rate": 4.815707058540111e-05, "loss": 0.1751, "step": 39980 }, { "epoch": 4.82, "learning_rate": 4.8169115875692606e-05, "loss": 0.21, "step": 39990 }, { "epoch": 4.82, "learning_rate": 4.81811611659841e-05, "loss": 0.2921, "step": 40000 }, { "epoch": 4.82, "learning_rate": 4.8193206456275596e-05, "loss": 0.3349, "step": 40010 }, { "epoch": 4.82, "learning_rate": 4.820525174656709e-05, "loss": 0.2892, "step": 40020 }, { "epoch": 4.82, "learning_rate": 4.8217297036858586e-05, "loss": 0.1223, "step": 40030 }, { "epoch": 4.82, "learning_rate": 4.822934232715008e-05, "loss": 0.3061, "step": 40040 }, { "epoch": 4.82, "learning_rate": 4.8241387617441583e-05, "loss": 0.1247, "step": 40050 }, { "epoch": 4.83, "learning_rate": 4.825343290773308e-05, "loss": 0.1978, "step": 40060 }, { "epoch": 4.83, "learning_rate": 4.826547819802458e-05, "loss": 0.3971, "step": 40070 }, { "epoch": 4.83, "learning_rate": 4.8277523488316076e-05, "loss": 0.3181, "step": 40080 }, { "epoch": 4.83, "learning_rate": 4.828956877860757e-05, "loss": 0.2885, "step": 40090 }, { "epoch": 4.83, "learning_rate": 4.8301614068899066e-05, "loss": 0.3062, "step": 40100 }, { "epoch": 4.83, "learning_rate": 4.831365935919056e-05, "loss": 0.1994, "step": 40110 }, { "epoch": 4.83, "learning_rate": 4.8325704649482056e-05, "loss": 0.1717, "step": 40120 }, { "epoch": 4.83, "learning_rate": 4.833774993977355e-05, "loss": 0.3691, "step": 40130 }, { "epoch": 4.83, "learning_rate": 4.834979523006505e-05, "loss": 0.2178, "step": 40140 }, { "epoch": 4.84, "learning_rate": 4.836184052035654e-05, "loss": 0.5002, "step": 40150 }, { "epoch": 4.84, "learning_rate": 4.837388581064804e-05, "loss": 0.1747, "step": 40160 }, { "epoch": 4.84, "learning_rate": 4.838593110093953e-05, "loss": 0.2908, "step": 40170 }, { "epoch": 4.84, "learning_rate": 4.8397976391231034e-05, "loss": 0.0878, "step": 40180 }, { "epoch": 4.84, "learning_rate": 4.841002168152253e-05, "loss": 0.218, "step": 40190 }, { "epoch": 4.84, "learning_rate": 4.8422066971814025e-05, "loss": 0.1294, "step": 40200 }, { "epoch": 4.84, "learning_rate": 4.843411226210552e-05, "loss": 0.2163, "step": 40210 }, { "epoch": 4.84, "learning_rate": 4.8446157552397015e-05, "loss": 0.0881, "step": 40220 }, { "epoch": 4.85, "learning_rate": 4.845820284268851e-05, "loss": 0.2322, "step": 40230 }, { "epoch": 4.85, "learning_rate": 4.8470248132980005e-05, "loss": 0.144, "step": 40240 }, { "epoch": 4.85, "learning_rate": 4.84822934232715e-05, "loss": 0.232, "step": 40250 }, { "epoch": 4.85, "learning_rate": 4.8494338713562996e-05, "loss": 0.323, "step": 40260 }, { "epoch": 4.85, "learning_rate": 4.850638400385449e-05, "loss": 0.2229, "step": 40270 }, { "epoch": 4.85, "learning_rate": 4.8518429294145986e-05, "loss": 0.2213, "step": 40280 }, { "epoch": 4.85, "learning_rate": 4.853047458443749e-05, "loss": 0.1944, "step": 40290 }, { "epoch": 4.85, "learning_rate": 4.854251987472898e-05, "loss": 0.3094, "step": 40300 }, { "epoch": 4.86, "learning_rate": 4.855456516502048e-05, "loss": 0.3185, "step": 40310 }, { "epoch": 4.86, "learning_rate": 4.8566610455311974e-05, "loss": 0.3551, "step": 40320 }, { "epoch": 4.86, "learning_rate": 4.8578655745603476e-05, "loss": 0.2517, "step": 40330 }, { "epoch": 4.86, "learning_rate": 4.859070103589497e-05, "loss": 0.0906, "step": 40340 }, { "epoch": 4.86, "learning_rate": 4.8602746326186466e-05, "loss": 0.1214, "step": 40350 }, { "epoch": 4.86, "learning_rate": 4.861479161647796e-05, "loss": 0.253, "step": 40360 }, { "epoch": 4.86, "learning_rate": 4.8626836906769456e-05, "loss": 0.2106, "step": 40370 }, { "epoch": 4.86, "learning_rate": 4.863888219706095e-05, "loss": 0.3937, "step": 40380 }, { "epoch": 4.86, "learning_rate": 4.865092748735245e-05, "loss": 0.2631, "step": 40390 }, { "epoch": 4.87, "learning_rate": 4.866297277764394e-05, "loss": 0.3241, "step": 40400 }, { "epoch": 4.87, "learning_rate": 4.8675018067935444e-05, "loss": 0.3431, "step": 40410 }, { "epoch": 4.87, "learning_rate": 4.868706335822694e-05, "loss": 0.2578, "step": 40420 }, { "epoch": 4.87, "learning_rate": 4.8699108648518434e-05, "loss": 0.3506, "step": 40430 }, { "epoch": 4.87, "learning_rate": 4.871115393880993e-05, "loss": 0.1138, "step": 40440 }, { "epoch": 4.87, "learning_rate": 4.8723199229101425e-05, "loss": 0.489, "step": 40450 }, { "epoch": 4.87, "learning_rate": 4.873524451939292e-05, "loss": 0.2473, "step": 40460 }, { "epoch": 4.87, "learning_rate": 4.8747289809684415e-05, "loss": 0.1872, "step": 40470 }, { "epoch": 4.88, "learning_rate": 4.875933509997591e-05, "loss": 0.098, "step": 40480 }, { "epoch": 4.88, "learning_rate": 4.8771380390267405e-05, "loss": 0.2768, "step": 40490 }, { "epoch": 4.88, "learning_rate": 4.87834256805589e-05, "loss": 0.4325, "step": 40500 }, { "epoch": 4.88, "learning_rate": 4.8795470970850396e-05, "loss": 0.242, "step": 40510 }, { "epoch": 4.88, "learning_rate": 4.88075162611419e-05, "loss": 0.1788, "step": 40520 }, { "epoch": 4.88, "learning_rate": 4.881956155143339e-05, "loss": 0.1835, "step": 40530 }, { "epoch": 4.88, "learning_rate": 4.883160684172489e-05, "loss": 0.1613, "step": 40540 }, { "epoch": 4.88, "learning_rate": 4.884365213201638e-05, "loss": 0.2692, "step": 40550 }, { "epoch": 4.89, "learning_rate": 4.885569742230788e-05, "loss": 0.2826, "step": 40560 }, { "epoch": 4.89, "learning_rate": 4.8867742712599374e-05, "loss": 0.1606, "step": 40570 }, { "epoch": 4.89, "learning_rate": 4.887978800289087e-05, "loss": 0.2468, "step": 40580 }, { "epoch": 4.89, "learning_rate": 4.8891833293182364e-05, "loss": 0.4099, "step": 40590 }, { "epoch": 4.89, "learning_rate": 4.8903878583473866e-05, "loss": 0.1925, "step": 40600 }, { "epoch": 4.89, "learning_rate": 4.891592387376536e-05, "loss": 0.3035, "step": 40610 }, { "epoch": 4.89, "learning_rate": 4.8927969164056856e-05, "loss": 0.1511, "step": 40620 }, { "epoch": 4.89, "learning_rate": 4.894001445434835e-05, "loss": 0.2644, "step": 40630 }, { "epoch": 4.89, "learning_rate": 4.895205974463985e-05, "loss": 0.2407, "step": 40640 }, { "epoch": 4.9, "learning_rate": 4.896410503493135e-05, "loss": 0.1441, "step": 40650 }, { "epoch": 4.9, "learning_rate": 4.8976150325222844e-05, "loss": 0.3296, "step": 40660 }, { "epoch": 4.9, "learning_rate": 4.898819561551434e-05, "loss": 0.2568, "step": 40670 }, { "epoch": 4.9, "learning_rate": 4.9000240905805834e-05, "loss": 0.4292, "step": 40680 }, { "epoch": 4.9, "learning_rate": 4.901228619609733e-05, "loss": 0.1919, "step": 40690 }, { "epoch": 4.9, "learning_rate": 4.9024331486388825e-05, "loss": 0.3391, "step": 40700 }, { "epoch": 4.9, "learning_rate": 4.903637677668032e-05, "loss": 0.3385, "step": 40710 }, { "epoch": 4.9, "learning_rate": 4.9048422066971815e-05, "loss": 0.2143, "step": 40720 }, { "epoch": 4.91, "learning_rate": 4.906046735726331e-05, "loss": 0.1824, "step": 40730 }, { "epoch": 4.91, "learning_rate": 4.9072512647554805e-05, "loss": 0.3566, "step": 40740 }, { "epoch": 4.91, "learning_rate": 4.90845579378463e-05, "loss": 0.3458, "step": 40750 }, { "epoch": 4.91, "learning_rate": 4.90966032281378e-05, "loss": 0.3154, "step": 40760 }, { "epoch": 4.91, "learning_rate": 4.91086485184293e-05, "loss": 0.3923, "step": 40770 }, { "epoch": 4.91, "learning_rate": 4.912069380872079e-05, "loss": 0.1826, "step": 40780 }, { "epoch": 4.91, "learning_rate": 4.913273909901229e-05, "loss": 0.1384, "step": 40790 }, { "epoch": 4.91, "learning_rate": 4.914478438930378e-05, "loss": 0.2136, "step": 40800 }, { "epoch": 4.92, "learning_rate": 4.915682967959528e-05, "loss": 0.2585, "step": 40810 }, { "epoch": 4.92, "learning_rate": 4.9168874969886774e-05, "loss": 0.2256, "step": 40820 }, { "epoch": 4.92, "learning_rate": 4.918092026017827e-05, "loss": 0.1375, "step": 40830 }, { "epoch": 4.92, "learning_rate": 4.9192965550469764e-05, "loss": 0.0845, "step": 40840 }, { "epoch": 4.92, "learning_rate": 4.920501084076126e-05, "loss": 0.1365, "step": 40850 }, { "epoch": 4.92, "learning_rate": 4.921705613105276e-05, "loss": 0.1562, "step": 40860 }, { "epoch": 4.92, "learning_rate": 4.9229101421344256e-05, "loss": 0.2378, "step": 40870 }, { "epoch": 4.92, "learning_rate": 4.924114671163576e-05, "loss": 0.2446, "step": 40880 }, { "epoch": 4.93, "learning_rate": 4.9253192001927253e-05, "loss": 0.3895, "step": 40890 }, { "epoch": 4.93, "learning_rate": 4.926523729221875e-05, "loss": 0.2501, "step": 40900 }, { "epoch": 4.93, "learning_rate": 4.9277282582510244e-05, "loss": 0.1819, "step": 40910 }, { "epoch": 4.93, "learning_rate": 4.928932787280174e-05, "loss": 0.1762, "step": 40920 }, { "epoch": 4.93, "learning_rate": 4.9301373163093234e-05, "loss": 0.1746, "step": 40930 }, { "epoch": 4.93, "learning_rate": 4.931341845338473e-05, "loss": 0.2707, "step": 40940 }, { "epoch": 4.93, "learning_rate": 4.9325463743676225e-05, "loss": 0.2249, "step": 40950 }, { "epoch": 4.93, "learning_rate": 4.933750903396772e-05, "loss": 0.2353, "step": 40960 }, { "epoch": 4.93, "learning_rate": 4.9349554324259215e-05, "loss": 0.2236, "step": 40970 }, { "epoch": 4.94, "learning_rate": 4.936159961455071e-05, "loss": 0.1565, "step": 40980 }, { "epoch": 4.94, "learning_rate": 4.937364490484221e-05, "loss": 0.0992, "step": 40990 }, { "epoch": 4.94, "learning_rate": 4.938569019513371e-05, "loss": 0.0727, "step": 41000 }, { "epoch": 4.94, "learning_rate": 4.93977354854252e-05, "loss": 0.1844, "step": 41010 }, { "epoch": 4.94, "learning_rate": 4.94097807757167e-05, "loss": 0.3889, "step": 41020 }, { "epoch": 4.94, "learning_rate": 4.942182606600819e-05, "loss": 0.3507, "step": 41030 }, { "epoch": 4.94, "learning_rate": 4.943387135629969e-05, "loss": 0.2408, "step": 41040 }, { "epoch": 4.94, "learning_rate": 4.944591664659118e-05, "loss": 0.2676, "step": 41050 }, { "epoch": 4.95, "learning_rate": 4.945796193688268e-05, "loss": 0.1112, "step": 41060 }, { "epoch": 4.95, "learning_rate": 4.9470007227174174e-05, "loss": 0.2481, "step": 41070 }, { "epoch": 4.95, "learning_rate": 4.948205251746567e-05, "loss": 0.2062, "step": 41080 }, { "epoch": 4.95, "learning_rate": 4.9494097807757164e-05, "loss": 0.2198, "step": 41090 }, { "epoch": 4.95, "learning_rate": 4.9506143098048666e-05, "loss": 0.5001, "step": 41100 }, { "epoch": 4.95, "learning_rate": 4.951818838834016e-05, "loss": 0.4397, "step": 41110 }, { "epoch": 4.95, "learning_rate": 4.953023367863166e-05, "loss": 0.3584, "step": 41120 }, { "epoch": 4.95, "learning_rate": 4.954227896892316e-05, "loss": 0.2333, "step": 41130 }, { "epoch": 4.96, "learning_rate": 4.9554324259214653e-05, "loss": 0.476, "step": 41140 }, { "epoch": 4.96, "learning_rate": 4.956636954950615e-05, "loss": 0.2551, "step": 41150 }, { "epoch": 4.96, "learning_rate": 4.9578414839797644e-05, "loss": 0.2103, "step": 41160 }, { "epoch": 4.96, "learning_rate": 4.959046013008914e-05, "loss": 0.1598, "step": 41170 }, { "epoch": 4.96, "learning_rate": 4.9602505420380634e-05, "loss": 0.2675, "step": 41180 }, { "epoch": 4.96, "learning_rate": 4.961455071067213e-05, "loss": 0.5322, "step": 41190 }, { "epoch": 4.96, "learning_rate": 4.9626596000963625e-05, "loss": 0.159, "step": 41200 }, { "epoch": 4.96, "learning_rate": 4.963864129125512e-05, "loss": 0.1916, "step": 41210 }, { "epoch": 4.96, "learning_rate": 4.9650686581546615e-05, "loss": 0.1815, "step": 41220 }, { "epoch": 4.97, "learning_rate": 4.966273187183812e-05, "loss": 0.2658, "step": 41230 }, { "epoch": 4.97, "learning_rate": 4.967477716212961e-05, "loss": 0.3224, "step": 41240 }, { "epoch": 4.97, "learning_rate": 4.968682245242111e-05, "loss": 0.3146, "step": 41250 }, { "epoch": 4.97, "learning_rate": 4.96988677427126e-05, "loss": 0.1385, "step": 41260 }, { "epoch": 4.97, "learning_rate": 4.97109130330041e-05, "loss": 0.1515, "step": 41270 }, { "epoch": 4.97, "learning_rate": 4.972295832329559e-05, "loss": 0.2369, "step": 41280 }, { "epoch": 4.97, "learning_rate": 4.973500361358709e-05, "loss": 0.2079, "step": 41290 }, { "epoch": 4.97, "learning_rate": 4.974704890387858e-05, "loss": 0.1965, "step": 41300 }, { "epoch": 4.98, "learning_rate": 4.975909419417008e-05, "loss": 0.1529, "step": 41310 }, { "epoch": 4.98, "learning_rate": 4.9771139484461574e-05, "loss": 0.14, "step": 41320 }, { "epoch": 4.98, "learning_rate": 4.9783184774753076e-05, "loss": 0.3655, "step": 41330 }, { "epoch": 4.98, "learning_rate": 4.979523006504457e-05, "loss": 0.314, "step": 41340 }, { "epoch": 4.98, "learning_rate": 4.9807275355336066e-05, "loss": 0.1959, "step": 41350 }, { "epoch": 4.98, "learning_rate": 4.981932064562756e-05, "loss": 0.1931, "step": 41360 }, { "epoch": 4.98, "learning_rate": 4.9831365935919056e-05, "loss": 0.3248, "step": 41370 }, { "epoch": 4.98, "learning_rate": 4.984341122621056e-05, "loss": 0.2557, "step": 41380 }, { "epoch": 4.99, "learning_rate": 4.9855456516502053e-05, "loss": 0.0866, "step": 41390 }, { "epoch": 4.99, "learning_rate": 4.986750180679355e-05, "loss": 0.2265, "step": 41400 }, { "epoch": 4.99, "learning_rate": 4.9879547097085044e-05, "loss": 0.5567, "step": 41410 }, { "epoch": 4.99, "learning_rate": 4.989159238737654e-05, "loss": 0.347, "step": 41420 }, { "epoch": 4.99, "learning_rate": 4.9903637677668034e-05, "loss": 0.1481, "step": 41430 }, { "epoch": 4.99, "learning_rate": 4.991568296795953e-05, "loss": 0.1131, "step": 41440 }, { "epoch": 4.99, "learning_rate": 4.9927728258251025e-05, "loss": 0.2222, "step": 41450 }, { "epoch": 4.99, "learning_rate": 4.9939773548542526e-05, "loss": 0.2748, "step": 41460 }, { "epoch": 4.99, "learning_rate": 4.995181883883402e-05, "loss": 0.4202, "step": 41470 }, { "epoch": 5.0, "learning_rate": 4.996386412912552e-05, "loss": 0.3751, "step": 41480 }, { "epoch": 5.0, "learning_rate": 4.997590941941701e-05, "loss": 0.3678, "step": 41490 }, { "epoch": 5.0, "learning_rate": 4.998795470970851e-05, "loss": 0.1462, "step": 41500 }, { "epoch": 5.0, "learning_rate": 5e-05, "loss": 0.266, "step": 41510 }, { "epoch": 5.0, "eval_accuracy": { "accuracy": 0.9312857573020175 }, "eval_f1": { "f1": 0.9313123811138734 }, "eval_loss": 0.25953179597854614, "eval_precision": { "precision": 0.9327488749607135 }, "eval_recall": { "recall": 0.931043574955592 }, "eval_runtime": 430.4725, "eval_samples_per_second": 38.574, "eval_steps_per_second": 19.288, "step": 41512 } ], "logging_steps": 10, "max_steps": 415100, "num_input_tokens_seen": 0, "num_train_epochs": 50, "save_steps": 500, "total_flos": 2.573539424704798e+19, "train_batch_size": 2, "trial_name": null, "trial_params": null }