{ "best_metric": 0.9013693928718567, "best_model_checkpoint": "KELONMYOSA/wav2vec2-xls-r-300m-emotion-ru/checkpoint-8000", "epoch": 0.6666666666666666, "global_step": 8000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 2.0000000000000002e-07, "loss": 1.631, "step": 10 }, { "epoch": 0.0, "learning_rate": 4.0000000000000003e-07, "loss": 1.6296, "step": 20 }, { "epoch": 0.0, "learning_rate": 6.000000000000001e-07, "loss": 1.6358, "step": 30 }, { "epoch": 0.0, "learning_rate": 8.000000000000001e-07, "loss": 1.6236, "step": 40 }, { "epoch": 0.0, "learning_rate": 1.0000000000000002e-06, "loss": 1.6157, "step": 50 }, { "epoch": 0.01, "learning_rate": 1.2000000000000002e-06, "loss": 1.6099, "step": 60 }, { "epoch": 0.01, "learning_rate": 1.4000000000000001e-06, "loss": 1.5963, "step": 70 }, { "epoch": 0.01, "learning_rate": 1.6000000000000001e-06, "loss": 1.5911, "step": 80 }, { "epoch": 0.01, "learning_rate": 1.8000000000000001e-06, "loss": 1.5825, "step": 90 }, { "epoch": 0.01, "learning_rate": 2.0000000000000003e-06, "loss": 1.5604, "step": 100 }, { "epoch": 0.01, "learning_rate": 2.2e-06, "loss": 1.5426, "step": 110 }, { "epoch": 0.01, "learning_rate": 2.4000000000000003e-06, "loss": 1.53, "step": 120 }, { "epoch": 0.01, "learning_rate": 2.6e-06, "loss": 1.5099, "step": 130 }, { "epoch": 0.01, "learning_rate": 2.8000000000000003e-06, "loss": 1.5179, "step": 140 }, { "epoch": 0.01, "learning_rate": 3e-06, "loss": 1.4752, "step": 150 }, { "epoch": 0.01, "learning_rate": 3.2000000000000003e-06, "loss": 1.4143, "step": 160 }, { "epoch": 0.01, "learning_rate": 3.4000000000000005e-06, "loss": 1.4182, "step": 170 }, { "epoch": 0.01, "learning_rate": 3.6000000000000003e-06, "loss": 1.4226, "step": 180 }, { "epoch": 0.02, "learning_rate": 3.8000000000000005e-06, "loss": 1.3295, "step": 190 }, { "epoch": 0.02, "learning_rate": 4.000000000000001e-06, "loss": 1.3089, "step": 200 }, { "epoch": 0.02, "learning_rate": 4.2000000000000004e-06, "loss": 1.2796, "step": 210 }, { "epoch": 0.02, "learning_rate": 4.4e-06, "loss": 1.305, "step": 220 }, { "epoch": 0.02, "learning_rate": 4.600000000000001e-06, "loss": 1.2726, "step": 230 }, { "epoch": 0.02, "learning_rate": 4.800000000000001e-06, "loss": 1.2172, "step": 240 }, { "epoch": 0.02, "learning_rate": 5e-06, "loss": 1.1223, "step": 250 }, { "epoch": 0.02, "learning_rate": 5.2e-06, "loss": 1.0843, "step": 260 }, { "epoch": 0.02, "learning_rate": 5.400000000000001e-06, "loss": 1.2701, "step": 270 }, { "epoch": 0.02, "learning_rate": 5.600000000000001e-06, "loss": 1.0854, "step": 280 }, { "epoch": 0.02, "learning_rate": 5.8e-06, "loss": 1.1641, "step": 290 }, { "epoch": 0.03, "learning_rate": 6e-06, "loss": 1.1195, "step": 300 }, { "epoch": 0.03, "learning_rate": 6.200000000000001e-06, "loss": 1.0115, "step": 310 }, { "epoch": 0.03, "learning_rate": 6.4000000000000006e-06, "loss": 1.1457, "step": 320 }, { "epoch": 0.03, "learning_rate": 6.600000000000001e-06, "loss": 1.0518, "step": 330 }, { "epoch": 0.03, "learning_rate": 6.800000000000001e-06, "loss": 1.0924, "step": 340 }, { "epoch": 0.03, "learning_rate": 7e-06, "loss": 1.0398, "step": 350 }, { "epoch": 0.03, "learning_rate": 7.2000000000000005e-06, "loss": 0.9859, "step": 360 }, { "epoch": 0.03, "learning_rate": 7.4e-06, "loss": 1.0044, "step": 370 }, { "epoch": 0.03, "learning_rate": 7.600000000000001e-06, "loss": 1.1685, "step": 380 }, { "epoch": 0.03, "learning_rate": 7.800000000000002e-06, "loss": 1.1939, "step": 390 }, { "epoch": 0.03, "learning_rate": 8.000000000000001e-06, "loss": 0.9481, "step": 400 }, { "epoch": 0.03, "learning_rate": 8.2e-06, "loss": 1.0322, "step": 410 }, { "epoch": 0.04, "learning_rate": 8.400000000000001e-06, "loss": 1.0411, "step": 420 }, { "epoch": 0.04, "learning_rate": 8.6e-06, "loss": 1.1454, "step": 430 }, { "epoch": 0.04, "learning_rate": 8.8e-06, "loss": 1.0387, "step": 440 }, { "epoch": 0.04, "learning_rate": 9e-06, "loss": 1.079, "step": 450 }, { "epoch": 0.04, "learning_rate": 9.200000000000002e-06, "loss": 1.0594, "step": 460 }, { "epoch": 0.04, "learning_rate": 9.4e-06, "loss": 0.8012, "step": 470 }, { "epoch": 0.04, "learning_rate": 9.600000000000001e-06, "loss": 1.1299, "step": 480 }, { "epoch": 0.04, "learning_rate": 9.800000000000001e-06, "loss": 1.0191, "step": 490 }, { "epoch": 0.04, "learning_rate": 1e-05, "loss": 0.8813, "step": 500 }, { "epoch": 0.04, "learning_rate": 9.991304347826087e-06, "loss": 1.0422, "step": 510 }, { "epoch": 0.04, "learning_rate": 9.982608695652175e-06, "loss": 1.1059, "step": 520 }, { "epoch": 0.04, "learning_rate": 9.973913043478262e-06, "loss": 0.9619, "step": 530 }, { "epoch": 0.04, "learning_rate": 9.965217391304348e-06, "loss": 1.1024, "step": 540 }, { "epoch": 0.05, "learning_rate": 9.956521739130436e-06, "loss": 1.0865, "step": 550 }, { "epoch": 0.05, "learning_rate": 9.947826086956522e-06, "loss": 1.1372, "step": 560 }, { "epoch": 0.05, "learning_rate": 9.93913043478261e-06, "loss": 1.1235, "step": 570 }, { "epoch": 0.05, "learning_rate": 9.930434782608697e-06, "loss": 0.9579, "step": 580 }, { "epoch": 0.05, "learning_rate": 9.921739130434783e-06, "loss": 1.0254, "step": 590 }, { "epoch": 0.05, "learning_rate": 9.913043478260871e-06, "loss": 0.8638, "step": 600 }, { "epoch": 0.05, "learning_rate": 9.904347826086957e-06, "loss": 1.1145, "step": 610 }, { "epoch": 0.05, "learning_rate": 9.895652173913045e-06, "loss": 1.1482, "step": 620 }, { "epoch": 0.05, "learning_rate": 9.886956521739132e-06, "loss": 0.8615, "step": 630 }, { "epoch": 0.05, "learning_rate": 9.878260869565218e-06, "loss": 1.0799, "step": 640 }, { "epoch": 0.05, "learning_rate": 9.869565217391304e-06, "loss": 1.1813, "step": 650 }, { "epoch": 0.06, "learning_rate": 9.860869565217392e-06, "loss": 0.9973, "step": 660 }, { "epoch": 0.06, "learning_rate": 9.852173913043478e-06, "loss": 1.0687, "step": 670 }, { "epoch": 0.06, "learning_rate": 9.843478260869566e-06, "loss": 1.0122, "step": 680 }, { "epoch": 0.06, "learning_rate": 9.834782608695654e-06, "loss": 1.187, "step": 690 }, { "epoch": 0.06, "learning_rate": 9.82608695652174e-06, "loss": 1.0459, "step": 700 }, { "epoch": 0.06, "learning_rate": 9.817391304347826e-06, "loss": 0.9827, "step": 710 }, { "epoch": 0.06, "learning_rate": 9.808695652173913e-06, "loss": 0.986, "step": 720 }, { "epoch": 0.06, "learning_rate": 9.800000000000001e-06, "loss": 1.2352, "step": 730 }, { "epoch": 0.06, "learning_rate": 9.791304347826089e-06, "loss": 0.8748, "step": 740 }, { "epoch": 0.06, "learning_rate": 9.782608695652175e-06, "loss": 1.1145, "step": 750 }, { "epoch": 0.06, "learning_rate": 9.77391304347826e-06, "loss": 1.0772, "step": 760 }, { "epoch": 0.06, "learning_rate": 9.765217391304348e-06, "loss": 0.9784, "step": 770 }, { "epoch": 0.07, "learning_rate": 9.756521739130436e-06, "loss": 1.1624, "step": 780 }, { "epoch": 0.07, "learning_rate": 9.747826086956522e-06, "loss": 1.1276, "step": 790 }, { "epoch": 0.07, "learning_rate": 9.73913043478261e-06, "loss": 1.1284, "step": 800 }, { "epoch": 0.07, "learning_rate": 9.730434782608696e-06, "loss": 1.0081, "step": 810 }, { "epoch": 0.07, "learning_rate": 9.721739130434784e-06, "loss": 1.109, "step": 820 }, { "epoch": 0.07, "learning_rate": 9.713043478260871e-06, "loss": 0.8679, "step": 830 }, { "epoch": 0.07, "learning_rate": 9.704347826086957e-06, "loss": 1.0377, "step": 840 }, { "epoch": 0.07, "learning_rate": 9.695652173913043e-06, "loss": 1.2035, "step": 850 }, { "epoch": 0.07, "learning_rate": 9.686956521739131e-06, "loss": 0.9932, "step": 860 }, { "epoch": 0.07, "learning_rate": 9.678260869565219e-06, "loss": 1.0136, "step": 870 }, { "epoch": 0.07, "learning_rate": 9.669565217391305e-06, "loss": 1.0422, "step": 880 }, { "epoch": 0.07, "learning_rate": 9.660869565217392e-06, "loss": 1.0395, "step": 890 }, { "epoch": 0.07, "learning_rate": 9.652173913043478e-06, "loss": 0.999, "step": 900 }, { "epoch": 0.08, "learning_rate": 9.643478260869566e-06, "loss": 1.1831, "step": 910 }, { "epoch": 0.08, "learning_rate": 9.634782608695654e-06, "loss": 1.2003, "step": 920 }, { "epoch": 0.08, "learning_rate": 9.62608695652174e-06, "loss": 1.263, "step": 930 }, { "epoch": 0.08, "learning_rate": 9.617391304347828e-06, "loss": 1.0303, "step": 940 }, { "epoch": 0.08, "learning_rate": 9.608695652173914e-06, "loss": 1.0558, "step": 950 }, { "epoch": 0.08, "learning_rate": 9.600000000000001e-06, "loss": 1.2095, "step": 960 }, { "epoch": 0.08, "learning_rate": 9.591304347826087e-06, "loss": 1.0295, "step": 970 }, { "epoch": 0.08, "learning_rate": 9.582608695652175e-06, "loss": 1.0398, "step": 980 }, { "epoch": 0.08, "learning_rate": 9.573913043478261e-06, "loss": 1.0955, "step": 990 }, { "epoch": 0.08, "learning_rate": 9.565217391304349e-06, "loss": 1.0268, "step": 1000 }, { "epoch": 0.08, "learning_rate": 9.556521739130435e-06, "loss": 1.1389, "step": 1010 }, { "epoch": 0.09, "learning_rate": 9.547826086956522e-06, "loss": 1.1616, "step": 1020 }, { "epoch": 0.09, "learning_rate": 9.53913043478261e-06, "loss": 0.9908, "step": 1030 }, { "epoch": 0.09, "learning_rate": 9.530434782608696e-06, "loss": 1.1965, "step": 1040 }, { "epoch": 0.09, "learning_rate": 9.521739130434784e-06, "loss": 0.9678, "step": 1050 }, { "epoch": 0.09, "learning_rate": 9.51304347826087e-06, "loss": 0.914, "step": 1060 }, { "epoch": 0.09, "learning_rate": 9.504347826086958e-06, "loss": 0.9368, "step": 1070 }, { "epoch": 0.09, "learning_rate": 9.495652173913045e-06, "loss": 1.0157, "step": 1080 }, { "epoch": 0.09, "learning_rate": 9.486956521739131e-06, "loss": 1.0973, "step": 1090 }, { "epoch": 0.09, "learning_rate": 9.478260869565217e-06, "loss": 1.0243, "step": 1100 }, { "epoch": 0.09, "learning_rate": 9.469565217391305e-06, "loss": 1.2205, "step": 1110 }, { "epoch": 0.09, "learning_rate": 9.460869565217393e-06, "loss": 0.8545, "step": 1120 }, { "epoch": 0.09, "learning_rate": 9.452173913043479e-06, "loss": 0.9946, "step": 1130 }, { "epoch": 0.1, "learning_rate": 9.443478260869566e-06, "loss": 1.0649, "step": 1140 }, { "epoch": 0.1, "learning_rate": 9.434782608695652e-06, "loss": 1.1185, "step": 1150 }, { "epoch": 0.1, "learning_rate": 9.42608695652174e-06, "loss": 1.05, "step": 1160 }, { "epoch": 0.1, "learning_rate": 9.417391304347828e-06, "loss": 1.069, "step": 1170 }, { "epoch": 0.1, "learning_rate": 9.408695652173914e-06, "loss": 0.9964, "step": 1180 }, { "epoch": 0.1, "learning_rate": 9.4e-06, "loss": 0.8281, "step": 1190 }, { "epoch": 0.1, "learning_rate": 9.391304347826087e-06, "loss": 1.0723, "step": 1200 }, { "epoch": 0.1, "learning_rate": 9.382608695652175e-06, "loss": 1.0255, "step": 1210 }, { "epoch": 0.1, "learning_rate": 9.373913043478263e-06, "loss": 0.8457, "step": 1220 }, { "epoch": 0.1, "learning_rate": 9.365217391304349e-06, "loss": 0.9977, "step": 1230 }, { "epoch": 0.1, "learning_rate": 9.356521739130435e-06, "loss": 0.9381, "step": 1240 }, { "epoch": 0.1, "learning_rate": 9.347826086956523e-06, "loss": 0.8583, "step": 1250 }, { "epoch": 0.1, "learning_rate": 9.33913043478261e-06, "loss": 0.663, "step": 1260 }, { "epoch": 0.11, "learning_rate": 9.330434782608696e-06, "loss": 0.9826, "step": 1270 }, { "epoch": 0.11, "learning_rate": 9.321739130434784e-06, "loss": 0.9975, "step": 1280 }, { "epoch": 0.11, "learning_rate": 9.31304347826087e-06, "loss": 0.9189, "step": 1290 }, { "epoch": 0.11, "learning_rate": 9.304347826086956e-06, "loss": 1.1097, "step": 1300 }, { "epoch": 0.11, "learning_rate": 9.295652173913044e-06, "loss": 1.0124, "step": 1310 }, { "epoch": 0.11, "learning_rate": 9.286956521739131e-06, "loss": 0.8796, "step": 1320 }, { "epoch": 0.11, "learning_rate": 9.278260869565217e-06, "loss": 0.956, "step": 1330 }, { "epoch": 0.11, "learning_rate": 9.269565217391305e-06, "loss": 0.8929, "step": 1340 }, { "epoch": 0.11, "learning_rate": 9.260869565217391e-06, "loss": 1.0225, "step": 1350 }, { "epoch": 0.11, "learning_rate": 9.252173913043479e-06, "loss": 0.8236, "step": 1360 }, { "epoch": 0.11, "learning_rate": 9.243478260869567e-06, "loss": 0.7915, "step": 1370 }, { "epoch": 0.12, "learning_rate": 9.234782608695653e-06, "loss": 1.0637, "step": 1380 }, { "epoch": 0.12, "learning_rate": 9.22608695652174e-06, "loss": 0.8341, "step": 1390 }, { "epoch": 0.12, "learning_rate": 9.217391304347826e-06, "loss": 0.8537, "step": 1400 }, { "epoch": 0.12, "learning_rate": 9.208695652173914e-06, "loss": 0.681, "step": 1410 }, { "epoch": 0.12, "learning_rate": 9.200000000000002e-06, "loss": 1.1465, "step": 1420 }, { "epoch": 0.12, "learning_rate": 9.191304347826088e-06, "loss": 0.6344, "step": 1430 }, { "epoch": 0.12, "learning_rate": 9.182608695652174e-06, "loss": 1.002, "step": 1440 }, { "epoch": 0.12, "learning_rate": 9.173913043478261e-06, "loss": 1.0698, "step": 1450 }, { "epoch": 0.12, "learning_rate": 9.165217391304349e-06, "loss": 0.8353, "step": 1460 }, { "epoch": 0.12, "learning_rate": 9.156521739130435e-06, "loss": 0.8719, "step": 1470 }, { "epoch": 0.12, "learning_rate": 9.147826086956523e-06, "loss": 0.6585, "step": 1480 }, { "epoch": 0.12, "learning_rate": 9.139130434782609e-06, "loss": 0.8625, "step": 1490 }, { "epoch": 0.12, "learning_rate": 9.130434782608697e-06, "loss": 0.7229, "step": 1500 }, { "epoch": 0.13, "learning_rate": 9.121739130434784e-06, "loss": 0.904, "step": 1510 }, { "epoch": 0.13, "learning_rate": 9.11304347826087e-06, "loss": 0.8627, "step": 1520 }, { "epoch": 0.13, "learning_rate": 9.104347826086958e-06, "loss": 0.8142, "step": 1530 }, { "epoch": 0.13, "learning_rate": 9.095652173913044e-06, "loss": 0.8203, "step": 1540 }, { "epoch": 0.13, "learning_rate": 9.086956521739132e-06, "loss": 0.8628, "step": 1550 }, { "epoch": 0.13, "learning_rate": 9.07826086956522e-06, "loss": 0.967, "step": 1560 }, { "epoch": 0.13, "learning_rate": 9.069565217391305e-06, "loss": 0.7282, "step": 1570 }, { "epoch": 0.13, "learning_rate": 9.060869565217391e-06, "loss": 0.9502, "step": 1580 }, { "epoch": 0.13, "learning_rate": 9.052173913043479e-06, "loss": 0.7555, "step": 1590 }, { "epoch": 0.13, "learning_rate": 9.043478260869565e-06, "loss": 0.7224, "step": 1600 }, { "epoch": 0.13, "learning_rate": 9.034782608695653e-06, "loss": 0.6932, "step": 1610 }, { "epoch": 0.14, "learning_rate": 9.02608695652174e-06, "loss": 1.0107, "step": 1620 }, { "epoch": 0.14, "learning_rate": 9.017391304347827e-06, "loss": 0.729, "step": 1630 }, { "epoch": 0.14, "learning_rate": 9.008695652173913e-06, "loss": 0.8523, "step": 1640 }, { "epoch": 0.14, "learning_rate": 9e-06, "loss": 0.9994, "step": 1650 }, { "epoch": 0.14, "learning_rate": 8.991304347826088e-06, "loss": 0.9603, "step": 1660 }, { "epoch": 0.14, "learning_rate": 8.982608695652176e-06, "loss": 0.65, "step": 1670 }, { "epoch": 0.14, "learning_rate": 8.973913043478262e-06, "loss": 0.7482, "step": 1680 }, { "epoch": 0.14, "learning_rate": 8.965217391304348e-06, "loss": 0.8648, "step": 1690 }, { "epoch": 0.14, "learning_rate": 8.956521739130435e-06, "loss": 0.5774, "step": 1700 }, { "epoch": 0.14, "learning_rate": 8.947826086956523e-06, "loss": 0.9474, "step": 1710 }, { "epoch": 0.14, "learning_rate": 8.939130434782609e-06, "loss": 1.0509, "step": 1720 }, { "epoch": 0.14, "learning_rate": 8.930434782608697e-06, "loss": 0.7872, "step": 1730 }, { "epoch": 0.14, "learning_rate": 8.921739130434783e-06, "loss": 0.8817, "step": 1740 }, { "epoch": 0.15, "learning_rate": 8.91304347826087e-06, "loss": 0.7162, "step": 1750 }, { "epoch": 0.15, "learning_rate": 8.904347826086958e-06, "loss": 0.5956, "step": 1760 }, { "epoch": 0.15, "learning_rate": 8.895652173913044e-06, "loss": 0.5503, "step": 1770 }, { "epoch": 0.15, "learning_rate": 8.88695652173913e-06, "loss": 0.8549, "step": 1780 }, { "epoch": 0.15, "learning_rate": 8.878260869565218e-06, "loss": 0.8563, "step": 1790 }, { "epoch": 0.15, "learning_rate": 8.869565217391306e-06, "loss": 0.8728, "step": 1800 }, { "epoch": 0.15, "learning_rate": 8.860869565217392e-06, "loss": 0.8529, "step": 1810 }, { "epoch": 0.15, "learning_rate": 8.85217391304348e-06, "loss": 0.6122, "step": 1820 }, { "epoch": 0.15, "learning_rate": 8.843478260869565e-06, "loss": 0.7401, "step": 1830 }, { "epoch": 0.15, "learning_rate": 8.834782608695653e-06, "loss": 0.6384, "step": 1840 }, { "epoch": 0.15, "learning_rate": 8.82608695652174e-06, "loss": 0.7714, "step": 1850 }, { "epoch": 0.15, "learning_rate": 8.817391304347827e-06, "loss": 0.6963, "step": 1860 }, { "epoch": 0.16, "learning_rate": 8.808695652173914e-06, "loss": 0.7344, "step": 1870 }, { "epoch": 0.16, "learning_rate": 8.8e-06, "loss": 0.8102, "step": 1880 }, { "epoch": 0.16, "learning_rate": 8.791304347826088e-06, "loss": 0.641, "step": 1890 }, { "epoch": 0.16, "learning_rate": 8.782608695652174e-06, "loss": 0.772, "step": 1900 }, { "epoch": 0.16, "learning_rate": 8.773913043478262e-06, "loss": 0.6143, "step": 1910 }, { "epoch": 0.16, "learning_rate": 8.765217391304348e-06, "loss": 0.8776, "step": 1920 }, { "epoch": 0.16, "learning_rate": 8.756521739130436e-06, "loss": 0.6775, "step": 1930 }, { "epoch": 0.16, "learning_rate": 8.747826086956522e-06, "loss": 0.7589, "step": 1940 }, { "epoch": 0.16, "learning_rate": 8.73913043478261e-06, "loss": 0.6769, "step": 1950 }, { "epoch": 0.16, "learning_rate": 8.730434782608697e-06, "loss": 0.6676, "step": 1960 }, { "epoch": 0.16, "learning_rate": 8.721739130434783e-06, "loss": 0.7128, "step": 1970 }, { "epoch": 0.17, "learning_rate": 8.71304347826087e-06, "loss": 0.6978, "step": 1980 }, { "epoch": 0.17, "learning_rate": 8.704347826086957e-06, "loss": 0.7745, "step": 1990 }, { "epoch": 0.17, "learning_rate": 8.695652173913044e-06, "loss": 0.7833, "step": 2000 }, { "epoch": 0.17, "learning_rate": 8.686956521739132e-06, "loss": 0.7014, "step": 2010 }, { "epoch": 0.17, "learning_rate": 8.678260869565218e-06, "loss": 0.9515, "step": 2020 }, { "epoch": 0.17, "learning_rate": 8.669565217391304e-06, "loss": 0.8701, "step": 2030 }, { "epoch": 0.17, "learning_rate": 8.660869565217392e-06, "loss": 0.6525, "step": 2040 }, { "epoch": 0.17, "learning_rate": 8.65217391304348e-06, "loss": 0.6805, "step": 2050 }, { "epoch": 0.17, "learning_rate": 8.643478260869566e-06, "loss": 0.7611, "step": 2060 }, { "epoch": 0.17, "learning_rate": 8.634782608695653e-06, "loss": 0.6564, "step": 2070 }, { "epoch": 0.17, "learning_rate": 8.62608695652174e-06, "loss": 0.77, "step": 2080 }, { "epoch": 0.17, "learning_rate": 8.617391304347827e-06, "loss": 0.592, "step": 2090 }, { "epoch": 0.17, "learning_rate": 8.608695652173915e-06, "loss": 0.7016, "step": 2100 }, { "epoch": 0.18, "learning_rate": 8.6e-06, "loss": 0.7554, "step": 2110 }, { "epoch": 0.18, "learning_rate": 8.591304347826087e-06, "loss": 0.7579, "step": 2120 }, { "epoch": 0.18, "learning_rate": 8.582608695652174e-06, "loss": 0.5602, "step": 2130 }, { "epoch": 0.18, "learning_rate": 8.573913043478262e-06, "loss": 0.5452, "step": 2140 }, { "epoch": 0.18, "learning_rate": 8.56521739130435e-06, "loss": 0.5971, "step": 2150 }, { "epoch": 0.18, "learning_rate": 8.556521739130436e-06, "loss": 0.8296, "step": 2160 }, { "epoch": 0.18, "learning_rate": 8.547826086956522e-06, "loss": 0.7813, "step": 2170 }, { "epoch": 0.18, "learning_rate": 8.53913043478261e-06, "loss": 0.6838, "step": 2180 }, { "epoch": 0.18, "learning_rate": 8.530434782608697e-06, "loss": 0.6896, "step": 2190 }, { "epoch": 0.18, "learning_rate": 8.521739130434783e-06, "loss": 0.7821, "step": 2200 }, { "epoch": 0.18, "learning_rate": 8.513043478260871e-06, "loss": 0.6131, "step": 2210 }, { "epoch": 0.18, "learning_rate": 8.504347826086957e-06, "loss": 0.7313, "step": 2220 }, { "epoch": 0.19, "learning_rate": 8.495652173913043e-06, "loss": 0.6606, "step": 2230 }, { "epoch": 0.19, "learning_rate": 8.48695652173913e-06, "loss": 0.5877, "step": 2240 }, { "epoch": 0.19, "learning_rate": 8.478260869565218e-06, "loss": 0.5647, "step": 2250 }, { "epoch": 0.19, "learning_rate": 8.469565217391304e-06, "loss": 0.6218, "step": 2260 }, { "epoch": 0.19, "learning_rate": 8.460869565217392e-06, "loss": 0.6195, "step": 2270 }, { "epoch": 0.19, "learning_rate": 8.452173913043478e-06, "loss": 0.5907, "step": 2280 }, { "epoch": 0.19, "learning_rate": 8.443478260869566e-06, "loss": 0.517, "step": 2290 }, { "epoch": 0.19, "learning_rate": 8.434782608695653e-06, "loss": 0.7105, "step": 2300 }, { "epoch": 0.19, "learning_rate": 8.42608695652174e-06, "loss": 0.5361, "step": 2310 }, { "epoch": 0.19, "learning_rate": 8.417391304347827e-06, "loss": 0.7159, "step": 2320 }, { "epoch": 0.19, "learning_rate": 8.408695652173913e-06, "loss": 0.6338, "step": 2330 }, { "epoch": 0.2, "learning_rate": 8.400000000000001e-06, "loss": 0.6308, "step": 2340 }, { "epoch": 0.2, "learning_rate": 8.391304347826089e-06, "loss": 0.5378, "step": 2350 }, { "epoch": 0.2, "learning_rate": 8.382608695652175e-06, "loss": 0.5812, "step": 2360 }, { "epoch": 0.2, "learning_rate": 8.37391304347826e-06, "loss": 0.8999, "step": 2370 }, { "epoch": 0.2, "learning_rate": 8.365217391304348e-06, "loss": 0.6042, "step": 2380 }, { "epoch": 0.2, "learning_rate": 8.356521739130436e-06, "loss": 0.7057, "step": 2390 }, { "epoch": 0.2, "learning_rate": 8.347826086956522e-06, "loss": 0.6155, "step": 2400 }, { "epoch": 0.2, "learning_rate": 8.33913043478261e-06, "loss": 0.4642, "step": 2410 }, { "epoch": 0.2, "learning_rate": 8.330434782608696e-06, "loss": 0.703, "step": 2420 }, { "epoch": 0.2, "learning_rate": 8.321739130434783e-06, "loss": 0.5134, "step": 2430 }, { "epoch": 0.2, "learning_rate": 8.313043478260871e-06, "loss": 0.5832, "step": 2440 }, { "epoch": 0.2, "learning_rate": 8.304347826086957e-06, "loss": 0.6711, "step": 2450 }, { "epoch": 0.2, "learning_rate": 8.295652173913045e-06, "loss": 0.5107, "step": 2460 }, { "epoch": 0.21, "learning_rate": 8.286956521739131e-06, "loss": 0.5309, "step": 2470 }, { "epoch": 0.21, "learning_rate": 8.278260869565219e-06, "loss": 0.6143, "step": 2480 }, { "epoch": 0.21, "learning_rate": 8.269565217391306e-06, "loss": 0.7147, "step": 2490 }, { "epoch": 0.21, "learning_rate": 8.260869565217392e-06, "loss": 0.6881, "step": 2500 }, { "epoch": 0.21, "learning_rate": 8.252173913043478e-06, "loss": 0.5397, "step": 2510 }, { "epoch": 0.21, "learning_rate": 8.243478260869566e-06, "loss": 0.4793, "step": 2520 }, { "epoch": 0.21, "learning_rate": 8.234782608695652e-06, "loss": 0.5365, "step": 2530 }, { "epoch": 0.21, "learning_rate": 8.22608695652174e-06, "loss": 0.5199, "step": 2540 }, { "epoch": 0.21, "learning_rate": 8.217391304347827e-06, "loss": 0.6197, "step": 2550 }, { "epoch": 0.21, "learning_rate": 8.208695652173913e-06, "loss": 0.575, "step": 2560 }, { "epoch": 0.21, "learning_rate": 8.2e-06, "loss": 0.6319, "step": 2570 }, { "epoch": 0.21, "learning_rate": 8.191304347826087e-06, "loss": 0.7577, "step": 2580 }, { "epoch": 0.22, "learning_rate": 8.182608695652175e-06, "loss": 0.5802, "step": 2590 }, { "epoch": 0.22, "learning_rate": 8.173913043478263e-06, "loss": 0.5303, "step": 2600 }, { "epoch": 0.22, "learning_rate": 8.165217391304349e-06, "loss": 0.4836, "step": 2610 }, { "epoch": 0.22, "learning_rate": 8.156521739130435e-06, "loss": 0.6826, "step": 2620 }, { "epoch": 0.22, "learning_rate": 8.147826086956522e-06, "loss": 0.7211, "step": 2630 }, { "epoch": 0.22, "learning_rate": 8.13913043478261e-06, "loss": 0.5287, "step": 2640 }, { "epoch": 0.22, "learning_rate": 8.130434782608696e-06, "loss": 0.4684, "step": 2650 }, { "epoch": 0.22, "learning_rate": 8.121739130434784e-06, "loss": 0.6053, "step": 2660 }, { "epoch": 0.22, "learning_rate": 8.11304347826087e-06, "loss": 0.4545, "step": 2670 }, { "epoch": 0.22, "learning_rate": 8.104347826086957e-06, "loss": 0.6204, "step": 2680 }, { "epoch": 0.22, "learning_rate": 8.095652173913045e-06, "loss": 0.5319, "step": 2690 }, { "epoch": 0.23, "learning_rate": 8.086956521739131e-06, "loss": 0.6842, "step": 2700 }, { "epoch": 0.23, "learning_rate": 8.078260869565217e-06, "loss": 0.5415, "step": 2710 }, { "epoch": 0.23, "learning_rate": 8.069565217391305e-06, "loss": 0.7711, "step": 2720 }, { "epoch": 0.23, "learning_rate": 8.060869565217392e-06, "loss": 0.7313, "step": 2730 }, { "epoch": 0.23, "learning_rate": 8.052173913043479e-06, "loss": 0.723, "step": 2740 }, { "epoch": 0.23, "learning_rate": 8.043478260869566e-06, "loss": 0.5545, "step": 2750 }, { "epoch": 0.23, "learning_rate": 8.034782608695652e-06, "loss": 0.4794, "step": 2760 }, { "epoch": 0.23, "learning_rate": 8.02608695652174e-06, "loss": 0.5484, "step": 2770 }, { "epoch": 0.23, "learning_rate": 8.017391304347828e-06, "loss": 0.5945, "step": 2780 }, { "epoch": 0.23, "learning_rate": 8.008695652173914e-06, "loss": 0.6069, "step": 2790 }, { "epoch": 0.23, "learning_rate": 8.000000000000001e-06, "loss": 0.3657, "step": 2800 }, { "epoch": 0.23, "learning_rate": 7.991304347826087e-06, "loss": 0.6409, "step": 2810 }, { "epoch": 0.23, "learning_rate": 7.982608695652175e-06, "loss": 0.3748, "step": 2820 }, { "epoch": 0.24, "learning_rate": 7.973913043478261e-06, "loss": 0.5485, "step": 2830 }, { "epoch": 0.24, "learning_rate": 7.965217391304349e-06, "loss": 0.6298, "step": 2840 }, { "epoch": 0.24, "learning_rate": 7.956521739130435e-06, "loss": 0.4057, "step": 2850 }, { "epoch": 0.24, "learning_rate": 7.947826086956522e-06, "loss": 0.4526, "step": 2860 }, { "epoch": 0.24, "learning_rate": 7.939130434782608e-06, "loss": 0.6671, "step": 2870 }, { "epoch": 0.24, "learning_rate": 7.930434782608696e-06, "loss": 0.5103, "step": 2880 }, { "epoch": 0.24, "learning_rate": 7.921739130434784e-06, "loss": 0.3642, "step": 2890 }, { "epoch": 0.24, "learning_rate": 7.91304347826087e-06, "loss": 0.5497, "step": 2900 }, { "epoch": 0.24, "learning_rate": 7.904347826086958e-06, "loss": 0.656, "step": 2910 }, { "epoch": 0.24, "learning_rate": 7.895652173913044e-06, "loss": 0.4231, "step": 2920 }, { "epoch": 0.24, "learning_rate": 7.886956521739131e-06, "loss": 0.5834, "step": 2930 }, { "epoch": 0.24, "learning_rate": 7.878260869565219e-06, "loss": 0.6852, "step": 2940 }, { "epoch": 0.25, "learning_rate": 7.869565217391305e-06, "loss": 0.6258, "step": 2950 }, { "epoch": 0.25, "learning_rate": 7.860869565217391e-06, "loss": 0.4238, "step": 2960 }, { "epoch": 0.25, "learning_rate": 7.852173913043479e-06, "loss": 0.3731, "step": 2970 }, { "epoch": 0.25, "learning_rate": 7.843478260869566e-06, "loss": 0.6287, "step": 2980 }, { "epoch": 0.25, "learning_rate": 7.834782608695652e-06, "loss": 0.4353, "step": 2990 }, { "epoch": 0.25, "learning_rate": 7.82608695652174e-06, "loss": 0.5632, "step": 3000 }, { "epoch": 0.25, "learning_rate": 7.817391304347826e-06, "loss": 0.4435, "step": 3010 }, { "epoch": 0.25, "learning_rate": 7.808695652173914e-06, "loss": 0.4686, "step": 3020 }, { "epoch": 0.25, "learning_rate": 7.800000000000002e-06, "loss": 0.5092, "step": 3030 }, { "epoch": 0.25, "learning_rate": 7.791304347826088e-06, "loss": 0.4861, "step": 3040 }, { "epoch": 0.25, "learning_rate": 7.782608695652174e-06, "loss": 0.342, "step": 3050 }, { "epoch": 0.26, "learning_rate": 7.773913043478261e-06, "loss": 0.5394, "step": 3060 }, { "epoch": 0.26, "learning_rate": 7.765217391304349e-06, "loss": 0.3731, "step": 3070 }, { "epoch": 0.26, "learning_rate": 7.756521739130437e-06, "loss": 0.5001, "step": 3080 }, { "epoch": 0.26, "learning_rate": 7.747826086956523e-06, "loss": 0.5002, "step": 3090 }, { "epoch": 0.26, "learning_rate": 7.739130434782609e-06, "loss": 0.586, "step": 3100 }, { "epoch": 0.26, "learning_rate": 7.730434782608696e-06, "loss": 0.4913, "step": 3110 }, { "epoch": 0.26, "learning_rate": 7.721739130434784e-06, "loss": 0.506, "step": 3120 }, { "epoch": 0.26, "learning_rate": 7.71304347826087e-06, "loss": 0.4639, "step": 3130 }, { "epoch": 0.26, "learning_rate": 7.704347826086958e-06, "loss": 0.4068, "step": 3140 }, { "epoch": 0.26, "learning_rate": 7.695652173913044e-06, "loss": 0.168, "step": 3150 }, { "epoch": 0.26, "learning_rate": 7.68695652173913e-06, "loss": 0.5411, "step": 3160 }, { "epoch": 0.26, "learning_rate": 7.678260869565218e-06, "loss": 0.6719, "step": 3170 }, { "epoch": 0.27, "learning_rate": 7.669565217391305e-06, "loss": 0.3891, "step": 3180 }, { "epoch": 0.27, "learning_rate": 7.660869565217391e-06, "loss": 0.5448, "step": 3190 }, { "epoch": 0.27, "learning_rate": 7.652173913043479e-06, "loss": 0.4928, "step": 3200 }, { "epoch": 0.27, "learning_rate": 7.643478260869565e-06, "loss": 0.7067, "step": 3210 }, { "epoch": 0.27, "learning_rate": 7.634782608695653e-06, "loss": 0.4927, "step": 3220 }, { "epoch": 0.27, "learning_rate": 7.6260869565217395e-06, "loss": 0.5845, "step": 3230 }, { "epoch": 0.27, "learning_rate": 7.617391304347826e-06, "loss": 0.5043, "step": 3240 }, { "epoch": 0.27, "learning_rate": 7.608695652173914e-06, "loss": 0.3509, "step": 3250 }, { "epoch": 0.27, "learning_rate": 7.600000000000001e-06, "loss": 0.5087, "step": 3260 }, { "epoch": 0.27, "learning_rate": 7.591304347826087e-06, "loss": 0.4615, "step": 3270 }, { "epoch": 0.27, "learning_rate": 7.582608695652175e-06, "loss": 0.4291, "step": 3280 }, { "epoch": 0.27, "learning_rate": 7.5739130434782615e-06, "loss": 0.4949, "step": 3290 }, { "epoch": 0.28, "learning_rate": 7.565217391304348e-06, "loss": 0.6364, "step": 3300 }, { "epoch": 0.28, "learning_rate": 7.556521739130436e-06, "loss": 0.4992, "step": 3310 }, { "epoch": 0.28, "learning_rate": 7.547826086956522e-06, "loss": 0.3306, "step": 3320 }, { "epoch": 0.28, "learning_rate": 7.539130434782609e-06, "loss": 0.6495, "step": 3330 }, { "epoch": 0.28, "learning_rate": 7.530434782608697e-06, "loss": 0.6128, "step": 3340 }, { "epoch": 0.28, "learning_rate": 7.5217391304347835e-06, "loss": 0.5655, "step": 3350 }, { "epoch": 0.28, "learning_rate": 7.5130434782608695e-06, "loss": 0.4669, "step": 3360 }, { "epoch": 0.28, "learning_rate": 7.504347826086957e-06, "loss": 0.6109, "step": 3370 }, { "epoch": 0.28, "learning_rate": 7.495652173913044e-06, "loss": 0.6664, "step": 3380 }, { "epoch": 0.28, "learning_rate": 7.486956521739132e-06, "loss": 0.4606, "step": 3390 }, { "epoch": 0.28, "learning_rate": 7.478260869565218e-06, "loss": 0.5342, "step": 3400 }, { "epoch": 0.28, "learning_rate": 7.469565217391305e-06, "loss": 0.528, "step": 3410 }, { "epoch": 0.28, "learning_rate": 7.460869565217392e-06, "loss": 0.5597, "step": 3420 }, { "epoch": 0.29, "learning_rate": 7.452173913043479e-06, "loss": 0.4642, "step": 3430 }, { "epoch": 0.29, "learning_rate": 7.443478260869565e-06, "loss": 0.5932, "step": 3440 }, { "epoch": 0.29, "learning_rate": 7.434782608695653e-06, "loss": 0.5054, "step": 3450 }, { "epoch": 0.29, "learning_rate": 7.42608695652174e-06, "loss": 0.5701, "step": 3460 }, { "epoch": 0.29, "learning_rate": 7.417391304347827e-06, "loss": 0.5426, "step": 3470 }, { "epoch": 0.29, "learning_rate": 7.408695652173914e-06, "loss": 0.5007, "step": 3480 }, { "epoch": 0.29, "learning_rate": 7.4e-06, "loss": 0.4458, "step": 3490 }, { "epoch": 0.29, "learning_rate": 7.391304347826087e-06, "loss": 0.5264, "step": 3500 }, { "epoch": 0.29, "learning_rate": 7.382608695652175e-06, "loss": 0.5666, "step": 3510 }, { "epoch": 0.29, "learning_rate": 7.373913043478262e-06, "loss": 0.4319, "step": 3520 }, { "epoch": 0.29, "learning_rate": 7.365217391304348e-06, "loss": 0.4092, "step": 3530 }, { "epoch": 0.29, "learning_rate": 7.3565217391304354e-06, "loss": 0.5245, "step": 3540 }, { "epoch": 0.3, "learning_rate": 7.347826086956522e-06, "loss": 0.251, "step": 3550 }, { "epoch": 0.3, "learning_rate": 7.33913043478261e-06, "loss": 0.4116, "step": 3560 }, { "epoch": 0.3, "learning_rate": 7.330434782608696e-06, "loss": 0.4938, "step": 3570 }, { "epoch": 0.3, "learning_rate": 7.321739130434783e-06, "loss": 0.4155, "step": 3580 }, { "epoch": 0.3, "learning_rate": 7.3130434782608706e-06, "loss": 0.5594, "step": 3590 }, { "epoch": 0.3, "learning_rate": 7.304347826086957e-06, "loss": 0.6741, "step": 3600 }, { "epoch": 0.3, "learning_rate": 7.295652173913043e-06, "loss": 0.361, "step": 3610 }, { "epoch": 0.3, "learning_rate": 7.286956521739131e-06, "loss": 0.4328, "step": 3620 }, { "epoch": 0.3, "learning_rate": 7.278260869565218e-06, "loss": 0.5439, "step": 3630 }, { "epoch": 0.3, "learning_rate": 7.269565217391305e-06, "loss": 0.6128, "step": 3640 }, { "epoch": 0.3, "learning_rate": 7.2608695652173925e-06, "loss": 0.4837, "step": 3650 }, { "epoch": 0.3, "learning_rate": 7.2521739130434785e-06, "loss": 0.5195, "step": 3660 }, { "epoch": 0.31, "learning_rate": 7.243478260869565e-06, "loss": 0.5842, "step": 3670 }, { "epoch": 0.31, "learning_rate": 7.234782608695653e-06, "loss": 0.499, "step": 3680 }, { "epoch": 0.31, "learning_rate": 7.226086956521739e-06, "loss": 0.5089, "step": 3690 }, { "epoch": 0.31, "learning_rate": 7.217391304347827e-06, "loss": 0.4383, "step": 3700 }, { "epoch": 0.31, "learning_rate": 7.208695652173914e-06, "loss": 0.4841, "step": 3710 }, { "epoch": 0.31, "learning_rate": 7.2000000000000005e-06, "loss": 0.7326, "step": 3720 }, { "epoch": 0.31, "learning_rate": 7.191304347826088e-06, "loss": 0.3815, "step": 3730 }, { "epoch": 0.31, "learning_rate": 7.182608695652174e-06, "loss": 0.6135, "step": 3740 }, { "epoch": 0.31, "learning_rate": 7.173913043478261e-06, "loss": 0.3503, "step": 3750 }, { "epoch": 0.31, "learning_rate": 7.165217391304349e-06, "loss": 0.4625, "step": 3760 }, { "epoch": 0.31, "learning_rate": 7.156521739130436e-06, "loss": 0.51, "step": 3770 }, { "epoch": 0.32, "learning_rate": 7.147826086956522e-06, "loss": 0.6291, "step": 3780 }, { "epoch": 0.32, "learning_rate": 7.139130434782609e-06, "loss": 0.6093, "step": 3790 }, { "epoch": 0.32, "learning_rate": 7.130434782608696e-06, "loss": 0.5644, "step": 3800 }, { "epoch": 0.32, "learning_rate": 7.121739130434783e-06, "loss": 0.3497, "step": 3810 }, { "epoch": 0.32, "learning_rate": 7.113043478260871e-06, "loss": 0.5172, "step": 3820 }, { "epoch": 0.32, "learning_rate": 7.104347826086957e-06, "loss": 0.4977, "step": 3830 }, { "epoch": 0.32, "learning_rate": 7.095652173913044e-06, "loss": 0.3976, "step": 3840 }, { "epoch": 0.32, "learning_rate": 7.086956521739131e-06, "loss": 0.4518, "step": 3850 }, { "epoch": 0.32, "learning_rate": 7.078260869565217e-06, "loss": 0.293, "step": 3860 }, { "epoch": 0.32, "learning_rate": 7.069565217391305e-06, "loss": 0.6805, "step": 3870 }, { "epoch": 0.32, "learning_rate": 7.060869565217392e-06, "loss": 0.4503, "step": 3880 }, { "epoch": 0.32, "learning_rate": 7.052173913043479e-06, "loss": 0.4023, "step": 3890 }, { "epoch": 0.33, "learning_rate": 7.0434782608695665e-06, "loss": 0.5328, "step": 3900 }, { "epoch": 0.33, "learning_rate": 7.0347826086956525e-06, "loss": 0.4585, "step": 3910 }, { "epoch": 0.33, "learning_rate": 7.026086956521739e-06, "loss": 0.3256, "step": 3920 }, { "epoch": 0.33, "learning_rate": 7.017391304347827e-06, "loss": 0.4024, "step": 3930 }, { "epoch": 0.33, "learning_rate": 7.008695652173914e-06, "loss": 0.6103, "step": 3940 }, { "epoch": 0.33, "learning_rate": 7e-06, "loss": 0.3278, "step": 3950 }, { "epoch": 0.33, "learning_rate": 6.991304347826088e-06, "loss": 0.3631, "step": 3960 }, { "epoch": 0.33, "learning_rate": 6.9826086956521745e-06, "loss": 0.3458, "step": 3970 }, { "epoch": 0.33, "learning_rate": 6.973913043478261e-06, "loss": 0.2739, "step": 3980 }, { "epoch": 0.33, "learning_rate": 6.965217391304349e-06, "loss": 0.4365, "step": 3990 }, { "epoch": 0.33, "learning_rate": 6.956521739130435e-06, "loss": 0.4414, "step": 4000 }, { "epoch": 0.33, "eval_accuracy": 0.8835794925689697, "eval_loss": 0.39457714557647705, "eval_runtime": 2532.2842, "eval_samples_per_second": 9.545, "eval_steps_per_second": 1.193, "step": 4000 }, { "epoch": 0.33, "learning_rate": 6.947826086956523e-06, "loss": 0.513, "step": 4010 }, { "epoch": 0.34, "learning_rate": 6.93913043478261e-06, "loss": 0.5764, "step": 4020 }, { "epoch": 0.34, "learning_rate": 6.930434782608696e-06, "loss": 0.5618, "step": 4030 }, { "epoch": 0.34, "learning_rate": 6.921739130434783e-06, "loss": 0.6223, "step": 4040 }, { "epoch": 0.34, "learning_rate": 6.91304347826087e-06, "loss": 0.303, "step": 4050 }, { "epoch": 0.34, "learning_rate": 6.904347826086957e-06, "loss": 0.6319, "step": 4060 }, { "epoch": 0.34, "learning_rate": 6.895652173913045e-06, "loss": 0.4602, "step": 4070 }, { "epoch": 0.34, "learning_rate": 6.886956521739131e-06, "loss": 0.6363, "step": 4080 }, { "epoch": 0.34, "learning_rate": 6.8782608695652176e-06, "loss": 0.3161, "step": 4090 }, { "epoch": 0.34, "learning_rate": 6.869565217391305e-06, "loss": 0.5234, "step": 4100 }, { "epoch": 0.34, "learning_rate": 6.860869565217392e-06, "loss": 0.3137, "step": 4110 }, { "epoch": 0.34, "learning_rate": 6.852173913043478e-06, "loss": 0.4803, "step": 4120 }, { "epoch": 0.34, "learning_rate": 6.843478260869566e-06, "loss": 0.4827, "step": 4130 }, { "epoch": 0.34, "learning_rate": 6.834782608695653e-06, "loss": 0.6113, "step": 4140 }, { "epoch": 0.35, "learning_rate": 6.8260869565217395e-06, "loss": 0.4802, "step": 4150 }, { "epoch": 0.35, "learning_rate": 6.817391304347826e-06, "loss": 0.544, "step": 4160 }, { "epoch": 0.35, "learning_rate": 6.808695652173913e-06, "loss": 0.6566, "step": 4170 }, { "epoch": 0.35, "learning_rate": 6.800000000000001e-06, "loss": 0.2843, "step": 4180 }, { "epoch": 0.35, "learning_rate": 6.791304347826088e-06, "loss": 0.2734, "step": 4190 }, { "epoch": 0.35, "learning_rate": 6.782608695652174e-06, "loss": 0.5542, "step": 4200 }, { "epoch": 0.35, "learning_rate": 6.7739130434782615e-06, "loss": 0.4205, "step": 4210 }, { "epoch": 0.35, "learning_rate": 6.765217391304348e-06, "loss": 0.2538, "step": 4220 }, { "epoch": 0.35, "learning_rate": 6.756521739130435e-06, "loss": 0.5415, "step": 4230 }, { "epoch": 0.35, "learning_rate": 6.747826086956523e-06, "loss": 0.3251, "step": 4240 }, { "epoch": 0.35, "learning_rate": 6.739130434782609e-06, "loss": 0.7859, "step": 4250 }, { "epoch": 0.35, "learning_rate": 6.730434782608696e-06, "loss": 0.6057, "step": 4260 }, { "epoch": 0.36, "learning_rate": 6.7217391304347835e-06, "loss": 0.5861, "step": 4270 }, { "epoch": 0.36, "learning_rate": 6.71304347826087e-06, "loss": 0.3474, "step": 4280 }, { "epoch": 0.36, "learning_rate": 6.704347826086956e-06, "loss": 0.3905, "step": 4290 }, { "epoch": 0.36, "learning_rate": 6.695652173913044e-06, "loss": 0.4625, "step": 4300 }, { "epoch": 0.36, "learning_rate": 6.686956521739131e-06, "loss": 0.4694, "step": 4310 }, { "epoch": 0.36, "learning_rate": 6.678260869565219e-06, "loss": 0.5185, "step": 4320 }, { "epoch": 0.36, "learning_rate": 6.669565217391305e-06, "loss": 0.4148, "step": 4330 }, { "epoch": 0.36, "learning_rate": 6.6608695652173915e-06, "loss": 0.4534, "step": 4340 }, { "epoch": 0.36, "learning_rate": 6.652173913043479e-06, "loss": 0.4899, "step": 4350 }, { "epoch": 0.36, "learning_rate": 6.643478260869566e-06, "loss": 0.4686, "step": 4360 }, { "epoch": 0.36, "learning_rate": 6.634782608695652e-06, "loss": 0.4451, "step": 4370 }, { "epoch": 0.36, "learning_rate": 6.62608695652174e-06, "loss": 0.3401, "step": 4380 }, { "epoch": 0.37, "learning_rate": 6.617391304347827e-06, "loss": 0.4483, "step": 4390 }, { "epoch": 0.37, "learning_rate": 6.6086956521739135e-06, "loss": 0.4115, "step": 4400 }, { "epoch": 0.37, "learning_rate": 6.600000000000001e-06, "loss": 0.4291, "step": 4410 }, { "epoch": 0.37, "learning_rate": 6.591304347826087e-06, "loss": 0.4232, "step": 4420 }, { "epoch": 0.37, "learning_rate": 6.582608695652174e-06, "loss": 0.4715, "step": 4430 }, { "epoch": 0.37, "learning_rate": 6.573913043478262e-06, "loss": 0.7002, "step": 4440 }, { "epoch": 0.37, "learning_rate": 6.565217391304349e-06, "loss": 0.5288, "step": 4450 }, { "epoch": 0.37, "learning_rate": 6.556521739130435e-06, "loss": 0.4238, "step": 4460 }, { "epoch": 0.37, "learning_rate": 6.547826086956522e-06, "loss": 0.5096, "step": 4470 }, { "epoch": 0.37, "learning_rate": 6.539130434782609e-06, "loss": 0.3473, "step": 4480 }, { "epoch": 0.37, "learning_rate": 6.530434782608697e-06, "loss": 0.5329, "step": 4490 }, { "epoch": 0.38, "learning_rate": 6.521739130434783e-06, "loss": 0.3741, "step": 4500 }, { "epoch": 0.38, "learning_rate": 6.51304347826087e-06, "loss": 0.2783, "step": 4510 }, { "epoch": 0.38, "learning_rate": 6.5043478260869574e-06, "loss": 0.3165, "step": 4520 }, { "epoch": 0.38, "learning_rate": 6.495652173913044e-06, "loss": 0.48, "step": 4530 }, { "epoch": 0.38, "learning_rate": 6.48695652173913e-06, "loss": 0.5586, "step": 4540 }, { "epoch": 0.38, "learning_rate": 6.478260869565218e-06, "loss": 0.3564, "step": 4550 }, { "epoch": 0.38, "learning_rate": 6.469565217391305e-06, "loss": 0.5294, "step": 4560 }, { "epoch": 0.38, "learning_rate": 6.460869565217392e-06, "loss": 0.3656, "step": 4570 }, { "epoch": 0.38, "learning_rate": 6.4521739130434794e-06, "loss": 0.4936, "step": 4580 }, { "epoch": 0.38, "learning_rate": 6.4434782608695654e-06, "loss": 0.5711, "step": 4590 }, { "epoch": 0.38, "learning_rate": 6.434782608695652e-06, "loss": 0.3656, "step": 4600 }, { "epoch": 0.38, "learning_rate": 6.42608695652174e-06, "loss": 0.3292, "step": 4610 }, { "epoch": 0.39, "learning_rate": 6.417391304347827e-06, "loss": 0.4431, "step": 4620 }, { "epoch": 0.39, "learning_rate": 6.408695652173914e-06, "loss": 0.3815, "step": 4630 }, { "epoch": 0.39, "learning_rate": 6.4000000000000006e-06, "loss": 0.5099, "step": 4640 }, { "epoch": 0.39, "learning_rate": 6.391304347826087e-06, "loss": 0.5013, "step": 4650 }, { "epoch": 0.39, "learning_rate": 6.382608695652175e-06, "loss": 0.4564, "step": 4660 }, { "epoch": 0.39, "learning_rate": 6.373913043478261e-06, "loss": 0.3184, "step": 4670 }, { "epoch": 0.39, "learning_rate": 6.365217391304348e-06, "loss": 0.4151, "step": 4680 }, { "epoch": 0.39, "learning_rate": 6.356521739130436e-06, "loss": 0.56, "step": 4690 }, { "epoch": 0.39, "learning_rate": 6.3478260869565225e-06, "loss": 0.2881, "step": 4700 }, { "epoch": 0.39, "learning_rate": 6.3391304347826085e-06, "loss": 0.5104, "step": 4710 }, { "epoch": 0.39, "learning_rate": 6.330434782608696e-06, "loss": 0.4712, "step": 4720 }, { "epoch": 0.39, "learning_rate": 6.321739130434783e-06, "loss": 0.3975, "step": 4730 }, { "epoch": 0.4, "learning_rate": 6.31304347826087e-06, "loss": 0.2051, "step": 4740 }, { "epoch": 0.4, "learning_rate": 6.304347826086958e-06, "loss": 0.3613, "step": 4750 }, { "epoch": 0.4, "learning_rate": 6.295652173913044e-06, "loss": 0.4886, "step": 4760 }, { "epoch": 0.4, "learning_rate": 6.2869565217391305e-06, "loss": 0.5602, "step": 4770 }, { "epoch": 0.4, "learning_rate": 6.278260869565218e-06, "loss": 0.3032, "step": 4780 }, { "epoch": 0.4, "learning_rate": 6.269565217391304e-06, "loss": 0.3661, "step": 4790 }, { "epoch": 0.4, "learning_rate": 6.260869565217392e-06, "loss": 0.3678, "step": 4800 }, { "epoch": 0.4, "learning_rate": 6.252173913043479e-06, "loss": 0.4583, "step": 4810 }, { "epoch": 0.4, "learning_rate": 6.243478260869566e-06, "loss": 0.3001, "step": 4820 }, { "epoch": 0.4, "learning_rate": 6.234782608695653e-06, "loss": 0.2632, "step": 4830 }, { "epoch": 0.4, "learning_rate": 6.226086956521739e-06, "loss": 0.6544, "step": 4840 }, { "epoch": 0.4, "learning_rate": 6.217391304347826e-06, "loss": 0.3837, "step": 4850 }, { "epoch": 0.41, "learning_rate": 6.208695652173914e-06, "loss": 0.3821, "step": 4860 }, { "epoch": 0.41, "learning_rate": 6.200000000000001e-06, "loss": 0.5827, "step": 4870 }, { "epoch": 0.41, "learning_rate": 6.191304347826087e-06, "loss": 0.784, "step": 4880 }, { "epoch": 0.41, "learning_rate": 6.1826086956521745e-06, "loss": 0.4045, "step": 4890 }, { "epoch": 0.41, "learning_rate": 6.173913043478261e-06, "loss": 0.5202, "step": 4900 }, { "epoch": 0.41, "learning_rate": 6.165217391304348e-06, "loss": 0.3289, "step": 4910 }, { "epoch": 0.41, "learning_rate": 6.156521739130436e-06, "loss": 0.4096, "step": 4920 }, { "epoch": 0.41, "learning_rate": 6.147826086956522e-06, "loss": 0.4728, "step": 4930 }, { "epoch": 0.41, "learning_rate": 6.13913043478261e-06, "loss": 0.4807, "step": 4940 }, { "epoch": 0.41, "learning_rate": 6.1304347826086965e-06, "loss": 0.5515, "step": 4950 }, { "epoch": 0.41, "learning_rate": 6.1217391304347825e-06, "loss": 0.3108, "step": 4960 }, { "epoch": 0.41, "learning_rate": 6.11304347826087e-06, "loss": 0.4474, "step": 4970 }, { "epoch": 0.41, "learning_rate": 6.104347826086957e-06, "loss": 0.235, "step": 4980 }, { "epoch": 0.42, "learning_rate": 6.095652173913044e-06, "loss": 0.4078, "step": 4990 }, { "epoch": 0.42, "learning_rate": 6.086956521739132e-06, "loss": 0.3467, "step": 5000 }, { "epoch": 0.42, "learning_rate": 6.078260869565218e-06, "loss": 0.5282, "step": 5010 }, { "epoch": 0.42, "learning_rate": 6.0695652173913045e-06, "loss": 0.673, "step": 5020 }, { "epoch": 0.42, "learning_rate": 6.060869565217392e-06, "loss": 0.4286, "step": 5030 }, { "epoch": 0.42, "learning_rate": 6.052173913043479e-06, "loss": 0.3458, "step": 5040 }, { "epoch": 0.42, "learning_rate": 6.043478260869565e-06, "loss": 0.4361, "step": 5050 }, { "epoch": 0.42, "learning_rate": 6.034782608695653e-06, "loss": 0.4609, "step": 5060 }, { "epoch": 0.42, "learning_rate": 6.02608695652174e-06, "loss": 0.3551, "step": 5070 }, { "epoch": 0.42, "learning_rate": 6.0173913043478264e-06, "loss": 0.4154, "step": 5080 }, { "epoch": 0.42, "learning_rate": 6.008695652173913e-06, "loss": 0.348, "step": 5090 }, { "epoch": 0.42, "learning_rate": 6e-06, "loss": 0.5009, "step": 5100 }, { "epoch": 0.43, "learning_rate": 5.991304347826088e-06, "loss": 0.4235, "step": 5110 }, { "epoch": 0.43, "learning_rate": 5.982608695652175e-06, "loss": 0.295, "step": 5120 }, { "epoch": 0.43, "learning_rate": 5.973913043478261e-06, "loss": 0.5528, "step": 5130 }, { "epoch": 0.43, "learning_rate": 5.965217391304348e-06, "loss": 0.4839, "step": 5140 }, { "epoch": 0.43, "learning_rate": 5.956521739130435e-06, "loss": 0.3863, "step": 5150 }, { "epoch": 0.43, "learning_rate": 5.947826086956522e-06, "loss": 0.3885, "step": 5160 }, { "epoch": 0.43, "learning_rate": 5.93913043478261e-06, "loss": 0.4598, "step": 5170 }, { "epoch": 0.43, "learning_rate": 5.930434782608696e-06, "loss": 0.3377, "step": 5180 }, { "epoch": 0.43, "learning_rate": 5.921739130434783e-06, "loss": 0.4796, "step": 5190 }, { "epoch": 0.43, "learning_rate": 5.91304347826087e-06, "loss": 0.3193, "step": 5200 }, { "epoch": 0.43, "learning_rate": 5.904347826086957e-06, "loss": 0.2709, "step": 5210 }, { "epoch": 0.43, "learning_rate": 5.895652173913043e-06, "loss": 0.4047, "step": 5220 }, { "epoch": 0.44, "learning_rate": 5.886956521739131e-06, "loss": 0.2689, "step": 5230 }, { "epoch": 0.44, "learning_rate": 5.878260869565218e-06, "loss": 0.392, "step": 5240 }, { "epoch": 0.44, "learning_rate": 5.8695652173913055e-06, "loss": 0.3118, "step": 5250 }, { "epoch": 0.44, "learning_rate": 5.8608695652173915e-06, "loss": 0.54, "step": 5260 }, { "epoch": 0.44, "learning_rate": 5.852173913043478e-06, "loss": 0.4539, "step": 5270 }, { "epoch": 0.44, "learning_rate": 5.843478260869566e-06, "loss": 0.4643, "step": 5280 }, { "epoch": 0.44, "learning_rate": 5.834782608695653e-06, "loss": 0.4554, "step": 5290 }, { "epoch": 0.44, "learning_rate": 5.826086956521739e-06, "loss": 0.5387, "step": 5300 }, { "epoch": 0.44, "learning_rate": 5.817391304347827e-06, "loss": 0.625, "step": 5310 }, { "epoch": 0.44, "learning_rate": 5.8086956521739135e-06, "loss": 0.4794, "step": 5320 }, { "epoch": 0.44, "learning_rate": 5.8e-06, "loss": 0.5017, "step": 5330 }, { "epoch": 0.45, "learning_rate": 5.791304347826088e-06, "loss": 0.2624, "step": 5340 }, { "epoch": 0.45, "learning_rate": 5.782608695652174e-06, "loss": 0.2418, "step": 5350 }, { "epoch": 0.45, "learning_rate": 5.773913043478261e-06, "loss": 0.5305, "step": 5360 }, { "epoch": 0.45, "learning_rate": 5.765217391304349e-06, "loss": 0.3797, "step": 5370 }, { "epoch": 0.45, "learning_rate": 5.7565217391304355e-06, "loss": 0.338, "step": 5380 }, { "epoch": 0.45, "learning_rate": 5.7478260869565215e-06, "loss": 0.4566, "step": 5390 }, { "epoch": 0.45, "learning_rate": 5.739130434782609e-06, "loss": 0.2623, "step": 5400 }, { "epoch": 0.45, "learning_rate": 5.730434782608696e-06, "loss": 0.419, "step": 5410 }, { "epoch": 0.45, "learning_rate": 5.721739130434784e-06, "loss": 0.4284, "step": 5420 }, { "epoch": 0.45, "learning_rate": 5.71304347826087e-06, "loss": 0.5037, "step": 5430 }, { "epoch": 0.45, "learning_rate": 5.704347826086957e-06, "loss": 0.3832, "step": 5440 }, { "epoch": 0.45, "learning_rate": 5.695652173913044e-06, "loss": 0.4093, "step": 5450 }, { "epoch": 0.46, "learning_rate": 5.686956521739131e-06, "loss": 0.4119, "step": 5460 }, { "epoch": 0.46, "learning_rate": 5.678260869565217e-06, "loss": 0.5424, "step": 5470 }, { "epoch": 0.46, "learning_rate": 5.669565217391305e-06, "loss": 0.3854, "step": 5480 }, { "epoch": 0.46, "learning_rate": 5.660869565217392e-06, "loss": 0.4611, "step": 5490 }, { "epoch": 0.46, "learning_rate": 5.652173913043479e-06, "loss": 0.3213, "step": 5500 }, { "epoch": 0.46, "learning_rate": 5.643478260869566e-06, "loss": 0.2388, "step": 5510 }, { "epoch": 0.46, "learning_rate": 5.634782608695652e-06, "loss": 0.3259, "step": 5520 }, { "epoch": 0.46, "learning_rate": 5.626086956521739e-06, "loss": 0.5464, "step": 5530 }, { "epoch": 0.46, "learning_rate": 5.617391304347827e-06, "loss": 0.4079, "step": 5540 }, { "epoch": 0.46, "learning_rate": 5.608695652173914e-06, "loss": 0.5889, "step": 5550 }, { "epoch": 0.46, "learning_rate": 5.600000000000001e-06, "loss": 0.2886, "step": 5560 }, { "epoch": 0.46, "learning_rate": 5.5913043478260874e-06, "loss": 0.4331, "step": 5570 }, { "epoch": 0.47, "learning_rate": 5.582608695652174e-06, "loss": 0.4719, "step": 5580 }, { "epoch": 0.47, "learning_rate": 5.573913043478262e-06, "loss": 0.4235, "step": 5590 }, { "epoch": 0.47, "learning_rate": 5.565217391304348e-06, "loss": 0.4622, "step": 5600 }, { "epoch": 0.47, "learning_rate": 5.556521739130435e-06, "loss": 0.3789, "step": 5610 }, { "epoch": 0.47, "learning_rate": 5.5478260869565226e-06, "loss": 0.4403, "step": 5620 }, { "epoch": 0.47, "learning_rate": 5.5391304347826094e-06, "loss": 0.3822, "step": 5630 }, { "epoch": 0.47, "learning_rate": 5.5304347826086954e-06, "loss": 0.4849, "step": 5640 }, { "epoch": 0.47, "learning_rate": 5.521739130434783e-06, "loss": 0.3665, "step": 5650 }, { "epoch": 0.47, "learning_rate": 5.51304347826087e-06, "loss": 0.4176, "step": 5660 }, { "epoch": 0.47, "learning_rate": 5.504347826086957e-06, "loss": 0.35, "step": 5670 }, { "epoch": 0.47, "learning_rate": 5.4956521739130446e-06, "loss": 0.607, "step": 5680 }, { "epoch": 0.47, "learning_rate": 5.4869565217391306e-06, "loss": 0.4098, "step": 5690 }, { "epoch": 0.47, "learning_rate": 5.478260869565217e-06, "loss": 0.4243, "step": 5700 }, { "epoch": 0.48, "learning_rate": 5.469565217391305e-06, "loss": 0.3778, "step": 5710 }, { "epoch": 0.48, "learning_rate": 5.460869565217391e-06, "loss": 0.5509, "step": 5720 }, { "epoch": 0.48, "learning_rate": 5.452173913043479e-06, "loss": 0.3405, "step": 5730 }, { "epoch": 0.48, "learning_rate": 5.443478260869566e-06, "loss": 0.2789, "step": 5740 }, { "epoch": 0.48, "learning_rate": 5.4347826086956525e-06, "loss": 0.3013, "step": 5750 }, { "epoch": 0.48, "learning_rate": 5.42608695652174e-06, "loss": 0.5437, "step": 5760 }, { "epoch": 0.48, "learning_rate": 5.417391304347826e-06, "loss": 0.4119, "step": 5770 }, { "epoch": 0.48, "learning_rate": 5.408695652173913e-06, "loss": 0.3299, "step": 5780 }, { "epoch": 0.48, "learning_rate": 5.400000000000001e-06, "loss": 0.469, "step": 5790 }, { "epoch": 0.48, "learning_rate": 5.391304347826088e-06, "loss": 0.3592, "step": 5800 }, { "epoch": 0.48, "learning_rate": 5.382608695652174e-06, "loss": 0.3249, "step": 5810 }, { "epoch": 0.48, "learning_rate": 5.373913043478261e-06, "loss": 0.4799, "step": 5820 }, { "epoch": 0.49, "learning_rate": 5.365217391304348e-06, "loss": 0.4695, "step": 5830 }, { "epoch": 0.49, "learning_rate": 5.356521739130435e-06, "loss": 0.3424, "step": 5840 }, { "epoch": 0.49, "learning_rate": 5.347826086956523e-06, "loss": 0.3642, "step": 5850 }, { "epoch": 0.49, "learning_rate": 5.339130434782609e-06, "loss": 0.3982, "step": 5860 }, { "epoch": 0.49, "learning_rate": 5.3304347826086965e-06, "loss": 0.272, "step": 5870 }, { "epoch": 0.49, "learning_rate": 5.321739130434783e-06, "loss": 0.4771, "step": 5880 }, { "epoch": 0.49, "learning_rate": 5.313043478260869e-06, "loss": 0.3766, "step": 5890 }, { "epoch": 0.49, "learning_rate": 5.304347826086957e-06, "loss": 0.4422, "step": 5900 }, { "epoch": 0.49, "learning_rate": 5.295652173913044e-06, "loss": 0.361, "step": 5910 }, { "epoch": 0.49, "learning_rate": 5.286956521739131e-06, "loss": 0.4423, "step": 5920 }, { "epoch": 0.49, "learning_rate": 5.2782608695652185e-06, "loss": 0.3152, "step": 5930 }, { "epoch": 0.49, "learning_rate": 5.2695652173913045e-06, "loss": 0.3038, "step": 5940 }, { "epoch": 0.5, "learning_rate": 5.260869565217391e-06, "loss": 0.4352, "step": 5950 }, { "epoch": 0.5, "learning_rate": 5.252173913043479e-06, "loss": 0.3773, "step": 5960 }, { "epoch": 0.5, "learning_rate": 5.243478260869566e-06, "loss": 0.4983, "step": 5970 }, { "epoch": 0.5, "learning_rate": 5.234782608695652e-06, "loss": 0.4574, "step": 5980 }, { "epoch": 0.5, "learning_rate": 5.22608695652174e-06, "loss": 0.4429, "step": 5990 }, { "epoch": 0.5, "learning_rate": 5.2173913043478265e-06, "loss": 0.4335, "step": 6000 }, { "epoch": 0.5, "learning_rate": 5.208695652173913e-06, "loss": 0.4459, "step": 6010 }, { "epoch": 0.5, "learning_rate": 5.2e-06, "loss": 0.2153, "step": 6020 }, { "epoch": 0.5, "learning_rate": 5.191304347826087e-06, "loss": 0.3343, "step": 6030 }, { "epoch": 0.5, "learning_rate": 5.182608695652175e-06, "loss": 0.3966, "step": 6040 }, { "epoch": 0.5, "learning_rate": 5.173913043478262e-06, "loss": 0.3575, "step": 6050 }, { "epoch": 0.51, "learning_rate": 5.165217391304348e-06, "loss": 0.4498, "step": 6060 }, { "epoch": 0.51, "learning_rate": 5.156521739130435e-06, "loss": 0.2789, "step": 6070 }, { "epoch": 0.51, "learning_rate": 5.147826086956522e-06, "loss": 0.4404, "step": 6080 }, { "epoch": 0.51, "learning_rate": 5.139130434782609e-06, "loss": 0.6099, "step": 6090 }, { "epoch": 0.51, "learning_rate": 5.130434782608697e-06, "loss": 0.3584, "step": 6100 }, { "epoch": 0.51, "learning_rate": 5.121739130434783e-06, "loss": 0.4152, "step": 6110 }, { "epoch": 0.51, "learning_rate": 5.11304347826087e-06, "loss": 0.3748, "step": 6120 }, { "epoch": 0.51, "learning_rate": 5.104347826086957e-06, "loss": 0.3673, "step": 6130 }, { "epoch": 0.51, "learning_rate": 5.095652173913044e-06, "loss": 0.4327, "step": 6140 }, { "epoch": 0.51, "learning_rate": 5.08695652173913e-06, "loss": 0.3909, "step": 6150 }, { "epoch": 0.51, "learning_rate": 5.078260869565218e-06, "loss": 0.3319, "step": 6160 }, { "epoch": 0.51, "learning_rate": 5.069565217391305e-06, "loss": 0.2865, "step": 6170 }, { "epoch": 0.52, "learning_rate": 5.060869565217392e-06, "loss": 0.2821, "step": 6180 }, { "epoch": 0.52, "learning_rate": 5.0521739130434784e-06, "loss": 0.3279, "step": 6190 }, { "epoch": 0.52, "learning_rate": 5.043478260869565e-06, "loss": 0.4764, "step": 6200 }, { "epoch": 0.52, "learning_rate": 5.034782608695653e-06, "loss": 0.3014, "step": 6210 }, { "epoch": 0.52, "learning_rate": 5.02608695652174e-06, "loss": 0.2445, "step": 6220 }, { "epoch": 0.52, "learning_rate": 5.017391304347826e-06, "loss": 0.31, "step": 6230 }, { "epoch": 0.52, "learning_rate": 5.0086956521739136e-06, "loss": 0.3301, "step": 6240 }, { "epoch": 0.52, "learning_rate": 5e-06, "loss": 0.4598, "step": 6250 }, { "epoch": 0.52, "learning_rate": 4.991304347826087e-06, "loss": 0.2277, "step": 6260 }, { "epoch": 0.52, "learning_rate": 4.982608695652174e-06, "loss": 0.5797, "step": 6270 }, { "epoch": 0.52, "learning_rate": 4.973913043478261e-06, "loss": 0.4112, "step": 6280 }, { "epoch": 0.52, "learning_rate": 4.965217391304349e-06, "loss": 0.3317, "step": 6290 }, { "epoch": 0.53, "learning_rate": 4.9565217391304355e-06, "loss": 0.2217, "step": 6300 }, { "epoch": 0.53, "learning_rate": 4.947826086956522e-06, "loss": 0.5037, "step": 6310 }, { "epoch": 0.53, "learning_rate": 4.939130434782609e-06, "loss": 0.2105, "step": 6320 }, { "epoch": 0.53, "learning_rate": 4.930434782608696e-06, "loss": 0.6034, "step": 6330 }, { "epoch": 0.53, "learning_rate": 4.921739130434783e-06, "loss": 0.3291, "step": 6340 }, { "epoch": 0.53, "learning_rate": 4.91304347826087e-06, "loss": 0.4337, "step": 6350 }, { "epoch": 0.53, "learning_rate": 4.904347826086957e-06, "loss": 0.3334, "step": 6360 }, { "epoch": 0.53, "learning_rate": 4.895652173913044e-06, "loss": 0.469, "step": 6370 }, { "epoch": 0.53, "learning_rate": 4.88695652173913e-06, "loss": 0.444, "step": 6380 }, { "epoch": 0.53, "learning_rate": 4.878260869565218e-06, "loss": 0.4402, "step": 6390 }, { "epoch": 0.53, "learning_rate": 4.869565217391305e-06, "loss": 0.3783, "step": 6400 }, { "epoch": 0.53, "learning_rate": 4.860869565217392e-06, "loss": 0.4571, "step": 6410 }, { "epoch": 0.54, "learning_rate": 4.852173913043479e-06, "loss": 0.3927, "step": 6420 }, { "epoch": 0.54, "learning_rate": 4.8434782608695655e-06, "loss": 0.3934, "step": 6430 }, { "epoch": 0.54, "learning_rate": 4.834782608695652e-06, "loss": 0.298, "step": 6440 }, { "epoch": 0.54, "learning_rate": 4.826086956521739e-06, "loss": 0.2527, "step": 6450 }, { "epoch": 0.54, "learning_rate": 4.817391304347827e-06, "loss": 0.4499, "step": 6460 }, { "epoch": 0.54, "learning_rate": 4.808695652173914e-06, "loss": 0.4651, "step": 6470 }, { "epoch": 0.54, "learning_rate": 4.800000000000001e-06, "loss": 0.2735, "step": 6480 }, { "epoch": 0.54, "learning_rate": 4.7913043478260875e-06, "loss": 0.1421, "step": 6490 }, { "epoch": 0.54, "learning_rate": 4.782608695652174e-06, "loss": 0.5493, "step": 6500 }, { "epoch": 0.54, "learning_rate": 4.773913043478261e-06, "loss": 0.3543, "step": 6510 }, { "epoch": 0.54, "learning_rate": 4.765217391304348e-06, "loss": 0.361, "step": 6520 }, { "epoch": 0.54, "learning_rate": 4.756521739130435e-06, "loss": 0.4458, "step": 6530 }, { "epoch": 0.55, "learning_rate": 4.747826086956523e-06, "loss": 0.5206, "step": 6540 }, { "epoch": 0.55, "learning_rate": 4.739130434782609e-06, "loss": 0.4621, "step": 6550 }, { "epoch": 0.55, "learning_rate": 4.730434782608696e-06, "loss": 0.2962, "step": 6560 }, { "epoch": 0.55, "learning_rate": 4.721739130434783e-06, "loss": 0.53, "step": 6570 }, { "epoch": 0.55, "learning_rate": 4.71304347826087e-06, "loss": 0.4722, "step": 6580 }, { "epoch": 0.55, "learning_rate": 4.704347826086957e-06, "loss": 0.5836, "step": 6590 }, { "epoch": 0.55, "learning_rate": 4.695652173913044e-06, "loss": 0.3245, "step": 6600 }, { "epoch": 0.55, "learning_rate": 4.6869565217391314e-06, "loss": 0.3038, "step": 6610 }, { "epoch": 0.55, "learning_rate": 4.6782608695652175e-06, "loss": 0.492, "step": 6620 }, { "epoch": 0.55, "learning_rate": 4.669565217391305e-06, "loss": 0.3223, "step": 6630 }, { "epoch": 0.55, "learning_rate": 4.660869565217392e-06, "loss": 0.4352, "step": 6640 }, { "epoch": 0.55, "learning_rate": 4.652173913043478e-06, "loss": 0.237, "step": 6650 }, { "epoch": 0.56, "learning_rate": 4.643478260869566e-06, "loss": 0.4075, "step": 6660 }, { "epoch": 0.56, "learning_rate": 4.634782608695653e-06, "loss": 0.3246, "step": 6670 }, { "epoch": 0.56, "learning_rate": 4.6260869565217394e-06, "loss": 0.3202, "step": 6680 }, { "epoch": 0.56, "learning_rate": 4.617391304347826e-06, "loss": 0.2525, "step": 6690 }, { "epoch": 0.56, "learning_rate": 4.608695652173913e-06, "loss": 0.6513, "step": 6700 }, { "epoch": 0.56, "learning_rate": 4.600000000000001e-06, "loss": 0.5217, "step": 6710 }, { "epoch": 0.56, "learning_rate": 4.591304347826087e-06, "loss": 0.5156, "step": 6720 }, { "epoch": 0.56, "learning_rate": 4.5826086956521746e-06, "loss": 0.3318, "step": 6730 }, { "epoch": 0.56, "learning_rate": 4.573913043478261e-06, "loss": 0.2152, "step": 6740 }, { "epoch": 0.56, "learning_rate": 4.565217391304348e-06, "loss": 0.2645, "step": 6750 }, { "epoch": 0.56, "learning_rate": 4.556521739130435e-06, "loss": 0.4943, "step": 6760 }, { "epoch": 0.56, "learning_rate": 4.547826086956522e-06, "loss": 0.4055, "step": 6770 }, { "epoch": 0.56, "learning_rate": 4.53913043478261e-06, "loss": 0.3357, "step": 6780 }, { "epoch": 0.57, "learning_rate": 4.530434782608696e-06, "loss": 0.3506, "step": 6790 }, { "epoch": 0.57, "learning_rate": 4.5217391304347826e-06, "loss": 0.3944, "step": 6800 }, { "epoch": 0.57, "learning_rate": 4.51304347826087e-06, "loss": 0.3678, "step": 6810 }, { "epoch": 0.57, "learning_rate": 4.504347826086956e-06, "loss": 0.287, "step": 6820 }, { "epoch": 0.57, "learning_rate": 4.495652173913044e-06, "loss": 0.4159, "step": 6830 }, { "epoch": 0.57, "learning_rate": 4.486956521739131e-06, "loss": 0.2688, "step": 6840 }, { "epoch": 0.57, "learning_rate": 4.478260869565218e-06, "loss": 0.6472, "step": 6850 }, { "epoch": 0.57, "learning_rate": 4.4695652173913045e-06, "loss": 0.349, "step": 6860 }, { "epoch": 0.57, "learning_rate": 4.460869565217391e-06, "loss": 0.5685, "step": 6870 }, { "epoch": 0.57, "learning_rate": 4.452173913043479e-06, "loss": 0.2058, "step": 6880 }, { "epoch": 0.57, "learning_rate": 4.443478260869565e-06, "loss": 0.4423, "step": 6890 }, { "epoch": 0.57, "learning_rate": 4.434782608695653e-06, "loss": 0.3688, "step": 6900 }, { "epoch": 0.58, "learning_rate": 4.42608695652174e-06, "loss": 0.262, "step": 6910 }, { "epoch": 0.58, "learning_rate": 4.4173913043478265e-06, "loss": 0.5532, "step": 6920 }, { "epoch": 0.58, "learning_rate": 4.408695652173913e-06, "loss": 0.3034, "step": 6930 }, { "epoch": 0.58, "learning_rate": 4.4e-06, "loss": 0.3576, "step": 6940 }, { "epoch": 0.58, "learning_rate": 4.391304347826087e-06, "loss": 0.5227, "step": 6950 }, { "epoch": 0.58, "learning_rate": 4.382608695652174e-06, "loss": 0.351, "step": 6960 }, { "epoch": 0.58, "learning_rate": 4.373913043478261e-06, "loss": 0.2825, "step": 6970 }, { "epoch": 0.58, "learning_rate": 4.3652173913043485e-06, "loss": 0.4252, "step": 6980 }, { "epoch": 0.58, "learning_rate": 4.356521739130435e-06, "loss": 0.3026, "step": 6990 }, { "epoch": 0.58, "learning_rate": 4.347826086956522e-06, "loss": 0.2679, "step": 7000 }, { "epoch": 0.58, "learning_rate": 4.339130434782609e-06, "loss": 0.3627, "step": 7010 }, { "epoch": 0.58, "learning_rate": 4.330434782608696e-06, "loss": 0.3735, "step": 7020 }, { "epoch": 0.59, "learning_rate": 4.321739130434783e-06, "loss": 0.305, "step": 7030 }, { "epoch": 0.59, "learning_rate": 4.31304347826087e-06, "loss": 0.347, "step": 7040 }, { "epoch": 0.59, "learning_rate": 4.304347826086957e-06, "loss": 0.2816, "step": 7050 }, { "epoch": 0.59, "learning_rate": 4.295652173913043e-06, "loss": 0.4648, "step": 7060 }, { "epoch": 0.59, "learning_rate": 4.286956521739131e-06, "loss": 0.3769, "step": 7070 }, { "epoch": 0.59, "learning_rate": 4.278260869565218e-06, "loss": 0.5504, "step": 7080 }, { "epoch": 0.59, "learning_rate": 4.269565217391305e-06, "loss": 0.3963, "step": 7090 }, { "epoch": 0.59, "learning_rate": 4.260869565217392e-06, "loss": 0.3197, "step": 7100 }, { "epoch": 0.59, "learning_rate": 4.2521739130434785e-06, "loss": 0.2514, "step": 7110 }, { "epoch": 0.59, "learning_rate": 4.243478260869565e-06, "loss": 0.3367, "step": 7120 }, { "epoch": 0.59, "learning_rate": 4.234782608695652e-06, "loss": 0.2859, "step": 7130 }, { "epoch": 0.59, "learning_rate": 4.226086956521739e-06, "loss": 0.7163, "step": 7140 }, { "epoch": 0.6, "learning_rate": 4.217391304347827e-06, "loss": 0.3885, "step": 7150 }, { "epoch": 0.6, "learning_rate": 4.208695652173914e-06, "loss": 0.3329, "step": 7160 }, { "epoch": 0.6, "learning_rate": 4.2000000000000004e-06, "loss": 0.3612, "step": 7170 }, { "epoch": 0.6, "learning_rate": 4.191304347826087e-06, "loss": 0.3814, "step": 7180 }, { "epoch": 0.6, "learning_rate": 4.182608695652174e-06, "loss": 0.3134, "step": 7190 }, { "epoch": 0.6, "learning_rate": 4.173913043478261e-06, "loss": 0.391, "step": 7200 }, { "epoch": 0.6, "learning_rate": 4.165217391304348e-06, "loss": 0.4616, "step": 7210 }, { "epoch": 0.6, "learning_rate": 4.1565217391304356e-06, "loss": 0.308, "step": 7220 }, { "epoch": 0.6, "learning_rate": 4.1478260869565224e-06, "loss": 0.2954, "step": 7230 }, { "epoch": 0.6, "learning_rate": 4.139130434782609e-06, "loss": 0.3238, "step": 7240 }, { "epoch": 0.6, "learning_rate": 4.130434782608696e-06, "loss": 0.5962, "step": 7250 }, { "epoch": 0.6, "learning_rate": 4.121739130434783e-06, "loss": 0.4165, "step": 7260 }, { "epoch": 0.61, "learning_rate": 4.11304347826087e-06, "loss": 0.227, "step": 7270 }, { "epoch": 0.61, "learning_rate": 4.104347826086957e-06, "loss": 0.5643, "step": 7280 }, { "epoch": 0.61, "learning_rate": 4.0956521739130436e-06, "loss": 0.3175, "step": 7290 }, { "epoch": 0.61, "learning_rate": 4.086956521739131e-06, "loss": 0.5319, "step": 7300 }, { "epoch": 0.61, "learning_rate": 4.078260869565217e-06, "loss": 0.2988, "step": 7310 }, { "epoch": 0.61, "learning_rate": 4.069565217391305e-06, "loss": 0.2626, "step": 7320 }, { "epoch": 0.61, "learning_rate": 4.060869565217392e-06, "loss": 0.5552, "step": 7330 }, { "epoch": 0.61, "learning_rate": 4.052173913043479e-06, "loss": 0.3646, "step": 7340 }, { "epoch": 0.61, "learning_rate": 4.0434782608695655e-06, "loss": 0.5061, "step": 7350 }, { "epoch": 0.61, "learning_rate": 4.034782608695652e-06, "loss": 0.4709, "step": 7360 }, { "epoch": 0.61, "learning_rate": 4.026086956521739e-06, "loss": 0.402, "step": 7370 }, { "epoch": 0.61, "learning_rate": 4.017391304347826e-06, "loss": 0.3012, "step": 7380 }, { "epoch": 0.62, "learning_rate": 4.008695652173914e-06, "loss": 0.4486, "step": 7390 }, { "epoch": 0.62, "learning_rate": 4.000000000000001e-06, "loss": 0.2445, "step": 7400 }, { "epoch": 0.62, "learning_rate": 3.9913043478260875e-06, "loss": 0.4592, "step": 7410 }, { "epoch": 0.62, "learning_rate": 3.982608695652174e-06, "loss": 0.4136, "step": 7420 }, { "epoch": 0.62, "learning_rate": 3.973913043478261e-06, "loss": 0.1555, "step": 7430 }, { "epoch": 0.62, "learning_rate": 3.965217391304348e-06, "loss": 0.3056, "step": 7440 }, { "epoch": 0.62, "learning_rate": 3.956521739130435e-06, "loss": 0.3233, "step": 7450 }, { "epoch": 0.62, "learning_rate": 3.947826086956522e-06, "loss": 0.2379, "step": 7460 }, { "epoch": 0.62, "learning_rate": 3.9391304347826095e-06, "loss": 0.299, "step": 7470 }, { "epoch": 0.62, "learning_rate": 3.9304347826086955e-06, "loss": 0.3384, "step": 7480 }, { "epoch": 0.62, "learning_rate": 3.921739130434783e-06, "loss": 0.1043, "step": 7490 }, { "epoch": 0.62, "learning_rate": 3.91304347826087e-06, "loss": 0.2503, "step": 7500 }, { "epoch": 0.63, "learning_rate": 3.904347826086957e-06, "loss": 0.3636, "step": 7510 }, { "epoch": 0.63, "learning_rate": 3.895652173913044e-06, "loss": 0.1322, "step": 7520 }, { "epoch": 0.63, "learning_rate": 3.886956521739131e-06, "loss": 0.5761, "step": 7530 }, { "epoch": 0.63, "learning_rate": 3.878260869565218e-06, "loss": 0.3227, "step": 7540 }, { "epoch": 0.63, "learning_rate": 3.869565217391304e-06, "loss": 0.5, "step": 7550 }, { "epoch": 0.63, "learning_rate": 3.860869565217392e-06, "loss": 0.3253, "step": 7560 }, { "epoch": 0.63, "learning_rate": 3.852173913043479e-06, "loss": 0.3708, "step": 7570 }, { "epoch": 0.63, "learning_rate": 3.843478260869565e-06, "loss": 0.5366, "step": 7580 }, { "epoch": 0.63, "learning_rate": 3.834782608695653e-06, "loss": 0.4341, "step": 7590 }, { "epoch": 0.63, "learning_rate": 3.8260869565217395e-06, "loss": 0.2575, "step": 7600 }, { "epoch": 0.63, "learning_rate": 3.817391304347826e-06, "loss": 0.3122, "step": 7610 }, { "epoch": 0.64, "learning_rate": 3.808695652173913e-06, "loss": 0.4443, "step": 7620 }, { "epoch": 0.64, "learning_rate": 3.8000000000000005e-06, "loss": 0.3874, "step": 7630 }, { "epoch": 0.64, "learning_rate": 3.7913043478260873e-06, "loss": 0.3871, "step": 7640 }, { "epoch": 0.64, "learning_rate": 3.782608695652174e-06, "loss": 0.491, "step": 7650 }, { "epoch": 0.64, "learning_rate": 3.773913043478261e-06, "loss": 0.4204, "step": 7660 }, { "epoch": 0.64, "learning_rate": 3.7652173913043483e-06, "loss": 0.3397, "step": 7670 }, { "epoch": 0.64, "learning_rate": 3.7565217391304347e-06, "loss": 0.3577, "step": 7680 }, { "epoch": 0.64, "learning_rate": 3.747826086956522e-06, "loss": 0.3142, "step": 7690 }, { "epoch": 0.64, "learning_rate": 3.739130434782609e-06, "loss": 0.4641, "step": 7700 }, { "epoch": 0.64, "learning_rate": 3.730434782608696e-06, "loss": 0.1786, "step": 7710 }, { "epoch": 0.64, "learning_rate": 3.7217391304347826e-06, "loss": 0.3017, "step": 7720 }, { "epoch": 0.64, "learning_rate": 3.71304347826087e-06, "loss": 0.5301, "step": 7730 }, { "epoch": 0.65, "learning_rate": 3.704347826086957e-06, "loss": 0.3472, "step": 7740 }, { "epoch": 0.65, "learning_rate": 3.6956521739130436e-06, "loss": 0.3056, "step": 7750 }, { "epoch": 0.65, "learning_rate": 3.686956521739131e-06, "loss": 0.2355, "step": 7760 }, { "epoch": 0.65, "learning_rate": 3.6782608695652177e-06, "loss": 0.4231, "step": 7770 }, { "epoch": 0.65, "learning_rate": 3.669565217391305e-06, "loss": 0.2324, "step": 7780 }, { "epoch": 0.65, "learning_rate": 3.6608695652173914e-06, "loss": 0.4111, "step": 7790 }, { "epoch": 0.65, "learning_rate": 3.6521739130434787e-06, "loss": 0.6188, "step": 7800 }, { "epoch": 0.65, "learning_rate": 3.6434782608695656e-06, "loss": 0.4993, "step": 7810 }, { "epoch": 0.65, "learning_rate": 3.6347826086956524e-06, "loss": 0.3065, "step": 7820 }, { "epoch": 0.65, "learning_rate": 3.6260869565217393e-06, "loss": 0.5656, "step": 7830 }, { "epoch": 0.65, "learning_rate": 3.6173913043478265e-06, "loss": 0.4403, "step": 7840 }, { "epoch": 0.65, "learning_rate": 3.6086956521739134e-06, "loss": 0.3038, "step": 7850 }, { "epoch": 0.66, "learning_rate": 3.6000000000000003e-06, "loss": 0.4118, "step": 7860 }, { "epoch": 0.66, "learning_rate": 3.591304347826087e-06, "loss": 0.3002, "step": 7870 }, { "epoch": 0.66, "learning_rate": 3.5826086956521744e-06, "loss": 0.3278, "step": 7880 }, { "epoch": 0.66, "learning_rate": 3.573913043478261e-06, "loss": 0.5189, "step": 7890 }, { "epoch": 0.66, "learning_rate": 3.565217391304348e-06, "loss": 0.4405, "step": 7900 }, { "epoch": 0.66, "learning_rate": 3.5565217391304354e-06, "loss": 0.6204, "step": 7910 }, { "epoch": 0.66, "learning_rate": 3.547826086956522e-06, "loss": 0.3556, "step": 7920 }, { "epoch": 0.66, "learning_rate": 3.5391304347826087e-06, "loss": 0.3796, "step": 7930 }, { "epoch": 0.66, "learning_rate": 3.530434782608696e-06, "loss": 0.1067, "step": 7940 }, { "epoch": 0.66, "learning_rate": 3.5217391304347832e-06, "loss": 0.6211, "step": 7950 }, { "epoch": 0.66, "learning_rate": 3.5130434782608697e-06, "loss": 0.2412, "step": 7960 }, { "epoch": 0.66, "learning_rate": 3.504347826086957e-06, "loss": 0.3648, "step": 7970 }, { "epoch": 0.67, "learning_rate": 3.495652173913044e-06, "loss": 0.2585, "step": 7980 }, { "epoch": 0.67, "learning_rate": 3.4869565217391307e-06, "loss": 0.5519, "step": 7990 }, { "epoch": 0.67, "learning_rate": 3.4782608695652175e-06, "loss": 0.5287, "step": 8000 }, { "epoch": 0.67, "eval_accuracy": 0.9013693928718567, "eval_loss": 0.3496170938014984, "eval_runtime": 2530.8058, "eval_samples_per_second": 9.551, "eval_steps_per_second": 1.194, "step": 8000 } ], "max_steps": 12000, "num_train_epochs": 9223372036854775807, "total_flos": 3.098931272354354e+19, "trial_name": null, "trial_params": null }