{ "best_metric": null, "best_model_checkpoint": null, "epoch": 0.49995365459282987, "global_step": 3034, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 9.96704021094265e-09, "loss": 0.2145, "step": 10 }, { "epoch": 0.0, "learning_rate": 9.9340804218853e-09, "loss": 0.1998, "step": 20 }, { "epoch": 0.0, "learning_rate": 9.90112063282795e-09, "loss": 0.187, "step": 30 }, { "epoch": 0.01, "learning_rate": 9.8681608437706e-09, "loss": 0.1964, "step": 40 }, { "epoch": 0.01, "learning_rate": 9.835201054713249e-09, "loss": 0.1877, "step": 50 }, { "epoch": 0.01, "learning_rate": 9.802241265655901e-09, "loss": 0.1848, "step": 60 }, { "epoch": 0.01, "learning_rate": 9.769281476598551e-09, "loss": 0.2161, "step": 70 }, { "epoch": 0.01, "learning_rate": 9.7363216875412e-09, "loss": 0.2022, "step": 80 }, { "epoch": 0.01, "learning_rate": 9.70336189848385e-09, "loss": 0.1829, "step": 90 }, { "epoch": 0.02, "learning_rate": 9.6704021094265e-09, "loss": 0.1805, "step": 100 }, { "epoch": 0.02, "learning_rate": 9.63744232036915e-09, "loss": 0.1835, "step": 110 }, { "epoch": 0.02, "learning_rate": 9.6044825313118e-09, "loss": 0.1925, "step": 120 }, { "epoch": 0.02, "learning_rate": 9.57152274225445e-09, "loss": 0.1996, "step": 130 }, { "epoch": 0.02, "learning_rate": 9.5385629531971e-09, "loss": 0.1905, "step": 140 }, { "epoch": 0.02, "learning_rate": 9.50560316413975e-09, "loss": 0.2047, "step": 150 }, { "epoch": 0.03, "learning_rate": 9.472643375082399e-09, "loss": 0.1751, "step": 160 }, { "epoch": 0.03, "learning_rate": 9.439683586025049e-09, "loss": 0.1791, "step": 170 }, { "epoch": 0.03, "learning_rate": 9.4067237969677e-09, "loss": 0.1996, "step": 180 }, { "epoch": 0.03, "learning_rate": 9.373764007910349e-09, "loss": 0.2002, "step": 190 }, { "epoch": 0.03, "learning_rate": 9.340804218852999e-09, "loss": 0.1816, "step": 200 }, { "epoch": 0.03, "learning_rate": 9.30784442979565e-09, "loss": 0.1834, "step": 210 }, { "epoch": 0.04, "learning_rate": 9.2748846407383e-09, "loss": 0.1751, "step": 220 }, { "epoch": 0.04, "learning_rate": 9.24192485168095e-09, "loss": 0.1758, "step": 230 }, { "epoch": 0.04, "learning_rate": 9.2089650626236e-09, "loss": 0.1822, "step": 240 }, { "epoch": 0.04, "learning_rate": 9.17600527356625e-09, "loss": 0.1543, "step": 250 }, { "epoch": 0.04, "learning_rate": 9.1430454845089e-09, "loss": 0.1768, "step": 260 }, { "epoch": 0.04, "learning_rate": 9.110085695451548e-09, "loss": 0.1779, "step": 270 }, { "epoch": 0.05, "learning_rate": 9.077125906394198e-09, "loss": 0.1783, "step": 280 }, { "epoch": 0.05, "learning_rate": 9.04416611733685e-09, "loss": 0.1911, "step": 290 }, { "epoch": 0.05, "learning_rate": 9.011206328279499e-09, "loss": 0.175, "step": 300 }, { "epoch": 0.05, "learning_rate": 8.978246539222149e-09, "loss": 0.1991, "step": 310 }, { "epoch": 0.05, "learning_rate": 8.945286750164799e-09, "loss": 0.1881, "step": 320 }, { "epoch": 0.05, "learning_rate": 8.912326961107449e-09, "loss": 0.1655, "step": 330 }, { "epoch": 0.06, "learning_rate": 8.879367172050099e-09, "loss": 0.1694, "step": 340 }, { "epoch": 0.06, "learning_rate": 8.846407382992749e-09, "loss": 0.2049, "step": 350 }, { "epoch": 0.06, "learning_rate": 8.8134475939354e-09, "loss": 0.187, "step": 360 }, { "epoch": 0.06, "learning_rate": 8.78048780487805e-09, "loss": 0.1891, "step": 370 }, { "epoch": 0.06, "learning_rate": 8.7475280158207e-09, "loss": 0.178, "step": 380 }, { "epoch": 0.06, "learning_rate": 8.714568226763348e-09, "loss": 0.19, "step": 390 }, { "epoch": 0.07, "learning_rate": 8.681608437706e-09, "loss": 0.1625, "step": 400 }, { "epoch": 0.07, "learning_rate": 8.64864864864865e-09, "loss": 0.1688, "step": 410 }, { "epoch": 0.07, "learning_rate": 8.615688859591298e-09, "loss": 0.1737, "step": 420 }, { "epoch": 0.07, "learning_rate": 8.582729070533948e-09, "loss": 0.1822, "step": 430 }, { "epoch": 0.07, "learning_rate": 8.549769281476598e-09, "loss": 0.1653, "step": 440 }, { "epoch": 0.07, "learning_rate": 8.516809492419249e-09, "loss": 0.1622, "step": 450 }, { "epoch": 0.08, "learning_rate": 8.483849703361899e-09, "loss": 0.1734, "step": 460 }, { "epoch": 0.08, "learning_rate": 8.450889914304549e-09, "loss": 0.1668, "step": 470 }, { "epoch": 0.08, "learning_rate": 8.417930125247199e-09, "loss": 0.1554, "step": 480 }, { "epoch": 0.08, "learning_rate": 8.384970336189849e-09, "loss": 0.1717, "step": 490 }, { "epoch": 0.08, "learning_rate": 8.352010547132498e-09, "loss": 0.1747, "step": 500 }, { "epoch": 0.08, "learning_rate": 8.319050758075148e-09, "loss": 0.1676, "step": 510 }, { "epoch": 0.09, "learning_rate": 8.2860909690178e-09, "loss": 0.1545, "step": 520 }, { "epoch": 0.09, "learning_rate": 8.253131179960448e-09, "loss": 0.184, "step": 530 }, { "epoch": 0.09, "learning_rate": 8.220171390903098e-09, "loss": 0.1584, "step": 540 }, { "epoch": 0.09, "learning_rate": 8.187211601845748e-09, "loss": 0.1576, "step": 550 }, { "epoch": 0.09, "learning_rate": 8.154251812788398e-09, "loss": 0.1828, "step": 560 }, { "epoch": 0.09, "learning_rate": 8.121292023731048e-09, "loss": 0.1786, "step": 570 }, { "epoch": 0.1, "learning_rate": 8.088332234673698e-09, "loss": 0.1876, "step": 580 }, { "epoch": 0.1, "learning_rate": 8.055372445616349e-09, "loss": 0.1549, "step": 590 }, { "epoch": 0.1, "learning_rate": 8.022412656558999e-09, "loss": 0.168, "step": 600 }, { "epoch": 0.1, "learning_rate": 7.989452867501649e-09, "loss": 0.1685, "step": 610 }, { "epoch": 0.1, "learning_rate": 7.956493078444297e-09, "loss": 0.1663, "step": 620 }, { "epoch": 0.1, "learning_rate": 7.923533289386949e-09, "loss": 0.1661, "step": 630 }, { "epoch": 0.11, "learning_rate": 7.890573500329597e-09, "loss": 0.1688, "step": 640 }, { "epoch": 0.11, "learning_rate": 7.857613711272248e-09, "loss": 0.1897, "step": 650 }, { "epoch": 0.11, "learning_rate": 7.824653922214898e-09, "loss": 0.1831, "step": 660 }, { "epoch": 0.11, "learning_rate": 7.791694133157548e-09, "loss": 0.1788, "step": 670 }, { "epoch": 0.11, "learning_rate": 7.758734344100198e-09, "loss": 0.1649, "step": 680 }, { "epoch": 0.11, "learning_rate": 7.725774555042848e-09, "loss": 0.182, "step": 690 }, { "epoch": 0.12, "learning_rate": 7.692814765985498e-09, "loss": 0.1725, "step": 700 }, { "epoch": 0.12, "learning_rate": 7.659854976928148e-09, "loss": 0.1548, "step": 710 }, { "epoch": 0.12, "learning_rate": 7.626895187870798e-09, "loss": 0.1675, "step": 720 }, { "epoch": 0.12, "learning_rate": 7.593935398813447e-09, "loss": 0.177, "step": 730 }, { "epoch": 0.12, "learning_rate": 7.560975609756099e-09, "loss": 0.1571, "step": 740 }, { "epoch": 0.12, "learning_rate": 7.528015820698749e-09, "loss": 0.1749, "step": 750 }, { "epoch": 0.13, "learning_rate": 7.495056031641397e-09, "loss": 0.1468, "step": 760 }, { "epoch": 0.13, "learning_rate": 7.462096242584047e-09, "loss": 0.1482, "step": 770 }, { "epoch": 0.13, "learning_rate": 7.429136453526698e-09, "loss": 0.1721, "step": 780 }, { "epoch": 0.13, "learning_rate": 7.3961766644693474e-09, "loss": 0.1708, "step": 790 }, { "epoch": 0.13, "learning_rate": 7.3632168754119975e-09, "loss": 0.1439, "step": 800 }, { "epoch": 0.13, "learning_rate": 7.330257086354647e-09, "loss": 0.1504, "step": 810 }, { "epoch": 0.14, "learning_rate": 7.297297297297297e-09, "loss": 0.1649, "step": 820 }, { "epoch": 0.14, "learning_rate": 7.264337508239948e-09, "loss": 0.1481, "step": 830 }, { "epoch": 0.14, "learning_rate": 7.231377719182597e-09, "loss": 0.1709, "step": 840 }, { "epoch": 0.14, "learning_rate": 7.198417930125247e-09, "loss": 0.1576, "step": 850 }, { "epoch": 0.14, "learning_rate": 7.165458141067897e-09, "loss": 0.1763, "step": 860 }, { "epoch": 0.14, "learning_rate": 7.132498352010547e-09, "loss": 0.1451, "step": 870 }, { "epoch": 0.15, "learning_rate": 7.099538562953198e-09, "loss": 0.1728, "step": 880 }, { "epoch": 0.15, "learning_rate": 7.066578773895848e-09, "loss": 0.1577, "step": 890 }, { "epoch": 0.15, "learning_rate": 7.033618984838497e-09, "loss": 0.1722, "step": 900 }, { "epoch": 0.15, "learning_rate": 7.000659195781147e-09, "loss": 0.1565, "step": 910 }, { "epoch": 0.15, "learning_rate": 6.967699406723797e-09, "loss": 0.1495, "step": 920 }, { "epoch": 0.15, "learning_rate": 6.9347396176664465e-09, "loss": 0.1609, "step": 930 }, { "epoch": 0.15, "learning_rate": 6.9017798286090974e-09, "loss": 0.1717, "step": 940 }, { "epoch": 0.16, "learning_rate": 6.8688200395517476e-09, "loss": 0.1598, "step": 950 }, { "epoch": 0.16, "learning_rate": 6.835860250494397e-09, "loss": 0.167, "step": 960 }, { "epoch": 0.16, "learning_rate": 6.802900461437047e-09, "loss": 0.1698, "step": 970 }, { "epoch": 0.16, "learning_rate": 6.769940672379697e-09, "loss": 0.1733, "step": 980 }, { "epoch": 0.16, "learning_rate": 6.736980883322346e-09, "loss": 0.15, "step": 990 }, { "epoch": 0.16, "learning_rate": 6.704021094264997e-09, "loss": 0.1804, "step": 1000 }, { "epoch": 0.17, "learning_rate": 6.6710613052076466e-09, "loss": 0.1467, "step": 1010 }, { "epoch": 0.17, "learning_rate": 6.638101516150297e-09, "loss": 0.1544, "step": 1020 }, { "epoch": 0.17, "learning_rate": 6.605141727092947e-09, "loss": 0.1589, "step": 1030 }, { "epoch": 0.17, "learning_rate": 6.572181938035596e-09, "loss": 0.1485, "step": 1040 }, { "epoch": 0.17, "learning_rate": 6.539222148978247e-09, "loss": 0.1459, "step": 1050 }, { "epoch": 0.17, "learning_rate": 6.506262359920897e-09, "loss": 0.1392, "step": 1060 }, { "epoch": 0.18, "learning_rate": 6.473302570863546e-09, "loss": 0.1639, "step": 1070 }, { "epoch": 0.18, "learning_rate": 6.4403427818061965e-09, "loss": 0.1405, "step": 1080 }, { "epoch": 0.18, "learning_rate": 6.407382992748847e-09, "loss": 0.1555, "step": 1090 }, { "epoch": 0.18, "learning_rate": 6.374423203691496e-09, "loss": 0.1668, "step": 1100 }, { "epoch": 0.18, "learning_rate": 6.341463414634147e-09, "loss": 0.1672, "step": 1110 }, { "epoch": 0.18, "learning_rate": 6.308503625576797e-09, "loss": 0.1519, "step": 1120 }, { "epoch": 0.19, "learning_rate": 6.275543836519446e-09, "loss": 0.1485, "step": 1130 }, { "epoch": 0.19, "learning_rate": 6.242584047462096e-09, "loss": 0.1629, "step": 1140 }, { "epoch": 0.19, "learning_rate": 6.2096242584047465e-09, "loss": 0.1544, "step": 1150 }, { "epoch": 0.19, "learning_rate": 6.176664469347396e-09, "loss": 0.1519, "step": 1160 }, { "epoch": 0.19, "learning_rate": 6.143704680290047e-09, "loss": 0.1346, "step": 1170 }, { "epoch": 0.19, "learning_rate": 6.110744891232696e-09, "loss": 0.1575, "step": 1180 }, { "epoch": 0.2, "learning_rate": 6.077785102175346e-09, "loss": 0.152, "step": 1190 }, { "epoch": 0.2, "learning_rate": 6.044825313117996e-09, "loss": 0.1611, "step": 1200 }, { "epoch": 0.2, "learning_rate": 6.0118655240606455e-09, "loss": 0.1406, "step": 1210 }, { "epoch": 0.2, "learning_rate": 5.9789057350032964e-09, "loss": 0.1401, "step": 1220 }, { "epoch": 0.2, "learning_rate": 5.9459459459459465e-09, "loss": 0.16, "step": 1230 }, { "epoch": 0.2, "learning_rate": 5.912986156888596e-09, "loss": 0.1718, "step": 1240 }, { "epoch": 0.21, "learning_rate": 5.880026367831246e-09, "loss": 0.1598, "step": 1250 }, { "epoch": 0.21, "learning_rate": 5.847066578773896e-09, "loss": 0.1698, "step": 1260 }, { "epoch": 0.21, "learning_rate": 5.814106789716545e-09, "loss": 0.1502, "step": 1270 }, { "epoch": 0.21, "learning_rate": 5.781147000659196e-09, "loss": 0.1615, "step": 1280 }, { "epoch": 0.21, "learning_rate": 5.748187211601846e-09, "loss": 0.1466, "step": 1290 }, { "epoch": 0.21, "learning_rate": 5.715227422544496e-09, "loss": 0.1484, "step": 1300 }, { "epoch": 0.22, "learning_rate": 5.682267633487146e-09, "loss": 0.1417, "step": 1310 }, { "epoch": 0.22, "learning_rate": 5.649307844429796e-09, "loss": 0.1349, "step": 1320 }, { "epoch": 0.22, "learning_rate": 5.616348055372445e-09, "loss": 0.1461, "step": 1330 }, { "epoch": 0.22, "learning_rate": 5.583388266315096e-09, "loss": 0.1605, "step": 1340 }, { "epoch": 0.22, "learning_rate": 5.550428477257745e-09, "loss": 0.1484, "step": 1350 }, { "epoch": 0.22, "learning_rate": 5.5174686882003955e-09, "loss": 0.1438, "step": 1360 }, { "epoch": 0.23, "learning_rate": 5.484508899143046e-09, "loss": 0.156, "step": 1370 }, { "epoch": 0.23, "learning_rate": 5.451549110085695e-09, "loss": 0.1711, "step": 1380 }, { "epoch": 0.23, "learning_rate": 5.418589321028346e-09, "loss": 0.1451, "step": 1390 }, { "epoch": 0.23, "learning_rate": 5.385629531970996e-09, "loss": 0.1531, "step": 1400 }, { "epoch": 0.23, "learning_rate": 5.352669742913645e-09, "loss": 0.1406, "step": 1410 }, { "epoch": 0.23, "learning_rate": 5.319709953856295e-09, "loss": 0.157, "step": 1420 }, { "epoch": 0.24, "learning_rate": 5.2867501647989455e-09, "loss": 0.1404, "step": 1430 }, { "epoch": 0.24, "learning_rate": 5.253790375741595e-09, "loss": 0.1532, "step": 1440 }, { "epoch": 0.24, "learning_rate": 5.220830586684246e-09, "loss": 0.1473, "step": 1450 }, { "epoch": 0.24, "learning_rate": 5.187870797626896e-09, "loss": 0.1426, "step": 1460 }, { "epoch": 0.24, "learning_rate": 5.154911008569545e-09, "loss": 0.1416, "step": 1470 }, { "epoch": 0.24, "learning_rate": 5.121951219512195e-09, "loss": 0.1494, "step": 1480 }, { "epoch": 0.25, "learning_rate": 5.088991430454845e-09, "loss": 0.124, "step": 1490 }, { "epoch": 0.25, "learning_rate": 5.0560316413974946e-09, "loss": 0.1365, "step": 1500 }, { "epoch": 0.25, "learning_rate": 5.0230718523401455e-09, "loss": 0.1422, "step": 1510 }, { "epoch": 0.25, "learning_rate": 4.990112063282795e-09, "loss": 0.1501, "step": 1520 }, { "epoch": 0.25, "learning_rate": 4.957152274225445e-09, "loss": 0.1349, "step": 1530 }, { "epoch": 0.25, "learning_rate": 4.924192485168095e-09, "loss": 0.1496, "step": 1540 }, { "epoch": 0.26, "learning_rate": 4.891232696110745e-09, "loss": 0.1318, "step": 1550 }, { "epoch": 0.26, "learning_rate": 4.858272907053395e-09, "loss": 0.1505, "step": 1560 }, { "epoch": 0.26, "learning_rate": 4.8253131179960445e-09, "loss": 0.1486, "step": 1570 }, { "epoch": 0.26, "learning_rate": 4.792353328938695e-09, "loss": 0.1615, "step": 1580 }, { "epoch": 0.26, "learning_rate": 4.759393539881345e-09, "loss": 0.1598, "step": 1590 }, { "epoch": 0.26, "learning_rate": 4.726433750823995e-09, "loss": 0.1324, "step": 1600 }, { "epoch": 0.27, "learning_rate": 4.693473961766645e-09, "loss": 0.16, "step": 1610 }, { "epoch": 0.27, "learning_rate": 4.660514172709295e-09, "loss": 0.1581, "step": 1620 }, { "epoch": 0.27, "learning_rate": 4.627554383651944e-09, "loss": 0.1486, "step": 1630 }, { "epoch": 0.27, "learning_rate": 4.594594594594595e-09, "loss": 0.1421, "step": 1640 }, { "epoch": 0.27, "learning_rate": 4.561634805537245e-09, "loss": 0.1541, "step": 1650 }, { "epoch": 0.27, "learning_rate": 4.528675016479895e-09, "loss": 0.1565, "step": 1660 }, { "epoch": 0.28, "learning_rate": 4.495715227422544e-09, "loss": 0.1449, "step": 1670 }, { "epoch": 0.28, "learning_rate": 4.462755438365195e-09, "loss": 0.1492, "step": 1680 }, { "epoch": 0.28, "learning_rate": 4.429795649307844e-09, "loss": 0.1346, "step": 1690 }, { "epoch": 0.28, "learning_rate": 4.396835860250494e-09, "loss": 0.1465, "step": 1700 }, { "epoch": 0.28, "learning_rate": 4.3638760711931444e-09, "loss": 0.1534, "step": 1710 }, { "epoch": 0.28, "learning_rate": 4.3309162821357946e-09, "loss": 0.1428, "step": 1720 }, { "epoch": 0.29, "learning_rate": 4.297956493078445e-09, "loss": 0.1272, "step": 1730 }, { "epoch": 0.29, "learning_rate": 4.264996704021095e-09, "loss": 0.1241, "step": 1740 }, { "epoch": 0.29, "learning_rate": 4.232036914963744e-09, "loss": 0.1452, "step": 1750 }, { "epoch": 0.29, "learning_rate": 4.199077125906394e-09, "loss": 0.1237, "step": 1760 }, { "epoch": 0.29, "learning_rate": 4.166117336849044e-09, "loss": 0.1447, "step": 1770 }, { "epoch": 0.29, "learning_rate": 4.133157547791694e-09, "loss": 0.1487, "step": 1780 }, { "epoch": 0.29, "learning_rate": 4.1001977587343445e-09, "loss": 0.1523, "step": 1790 }, { "epoch": 0.3, "learning_rate": 4.067237969676994e-09, "loss": 0.1531, "step": 1800 }, { "epoch": 0.3, "learning_rate": 4.034278180619645e-09, "loss": 0.1424, "step": 1810 }, { "epoch": 0.3, "learning_rate": 4.001318391562294e-09, "loss": 0.135, "step": 1820 }, { "epoch": 0.3, "learning_rate": 3.968358602504944e-09, "loss": 0.1462, "step": 1830 }, { "epoch": 0.3, "learning_rate": 3.935398813447593e-09, "loss": 0.133, "step": 1840 }, { "epoch": 0.3, "learning_rate": 3.902439024390244e-09, "loss": 0.1463, "step": 1850 }, { "epoch": 0.31, "learning_rate": 3.869479235332894e-09, "loss": 0.1258, "step": 1860 }, { "epoch": 0.31, "learning_rate": 3.836519446275544e-09, "loss": 0.1304, "step": 1870 }, { "epoch": 0.31, "learning_rate": 3.803559657218194e-09, "loss": 0.1437, "step": 1880 }, { "epoch": 0.31, "learning_rate": 3.770599868160844e-09, "loss": 0.1507, "step": 1890 }, { "epoch": 0.31, "learning_rate": 3.737640079103494e-09, "loss": 0.1332, "step": 1900 }, { "epoch": 0.31, "learning_rate": 3.704680290046144e-09, "loss": 0.1415, "step": 1910 }, { "epoch": 0.32, "learning_rate": 3.671720500988794e-09, "loss": 0.1547, "step": 1920 }, { "epoch": 0.32, "learning_rate": 3.6387607119314436e-09, "loss": 0.1488, "step": 1930 }, { "epoch": 0.32, "learning_rate": 3.6058009228740933e-09, "loss": 0.1277, "step": 1940 }, { "epoch": 0.32, "learning_rate": 3.572841133816744e-09, "loss": 0.1393, "step": 1950 }, { "epoch": 0.32, "learning_rate": 3.5398813447593935e-09, "loss": 0.1424, "step": 1960 }, { "epoch": 0.32, "learning_rate": 3.5069215557020436e-09, "loss": 0.144, "step": 1970 }, { "epoch": 0.33, "learning_rate": 3.4739617666446937e-09, "loss": 0.1457, "step": 1980 }, { "epoch": 0.33, "learning_rate": 3.4410019775873434e-09, "loss": 0.1581, "step": 1990 }, { "epoch": 0.33, "learning_rate": 3.4080421885299935e-09, "loss": 0.1484, "step": 2000 }, { "epoch": 0.33, "learning_rate": 3.3750823994726432e-09, "loss": 0.1222, "step": 2010 }, { "epoch": 0.33, "learning_rate": 3.3421226104152933e-09, "loss": 0.1391, "step": 2020 }, { "epoch": 0.33, "learning_rate": 3.3091628213579435e-09, "loss": 0.1452, "step": 2030 }, { "epoch": 0.34, "learning_rate": 3.276203032300593e-09, "loss": 0.1467, "step": 2040 }, { "epoch": 0.34, "learning_rate": 3.2432432432432437e-09, "loss": 0.161, "step": 2050 }, { "epoch": 0.34, "learning_rate": 3.2102834541858934e-09, "loss": 0.1392, "step": 2060 }, { "epoch": 0.34, "learning_rate": 3.177323665128543e-09, "loss": 0.1366, "step": 2070 }, { "epoch": 0.34, "learning_rate": 3.1443638760711936e-09, "loss": 0.1261, "step": 2080 }, { "epoch": 0.34, "learning_rate": 3.1114040870138433e-09, "loss": 0.1454, "step": 2090 }, { "epoch": 0.35, "learning_rate": 3.078444297956493e-09, "loss": 0.1475, "step": 2100 }, { "epoch": 0.35, "learning_rate": 3.0454845088991427e-09, "loss": 0.1494, "step": 2110 }, { "epoch": 0.35, "learning_rate": 3.0125247198417932e-09, "loss": 0.1405, "step": 2120 }, { "epoch": 0.35, "learning_rate": 2.979564930784443e-09, "loss": 0.1602, "step": 2130 }, { "epoch": 0.35, "learning_rate": 2.9466051417270926e-09, "loss": 0.1468, "step": 2140 }, { "epoch": 0.35, "learning_rate": 2.913645352669743e-09, "loss": 0.1406, "step": 2150 }, { "epoch": 0.36, "learning_rate": 2.880685563612393e-09, "loss": 0.1243, "step": 2160 }, { "epoch": 0.36, "learning_rate": 2.847725774555043e-09, "loss": 0.147, "step": 2170 }, { "epoch": 0.36, "learning_rate": 2.814765985497693e-09, "loss": 0.1318, "step": 2180 }, { "epoch": 0.36, "learning_rate": 2.7818061964403428e-09, "loss": 0.135, "step": 2190 }, { "epoch": 0.36, "learning_rate": 2.748846407382993e-09, "loss": 0.1441, "step": 2200 }, { "epoch": 0.36, "learning_rate": 2.7158866183256426e-09, "loss": 0.1152, "step": 2210 }, { "epoch": 0.37, "learning_rate": 2.682926829268293e-09, "loss": 0.141, "step": 2220 }, { "epoch": 0.37, "learning_rate": 2.6499670402109428e-09, "loss": 0.143, "step": 2230 }, { "epoch": 0.37, "learning_rate": 2.6170072511535925e-09, "loss": 0.1359, "step": 2240 }, { "epoch": 0.37, "learning_rate": 2.584047462096243e-09, "loss": 0.141, "step": 2250 }, { "epoch": 0.37, "learning_rate": 2.5510876730388927e-09, "loss": 0.1341, "step": 2260 }, { "epoch": 0.37, "learning_rate": 2.5181278839815424e-09, "loss": 0.1377, "step": 2270 }, { "epoch": 0.38, "learning_rate": 2.4851680949241925e-09, "loss": 0.1453, "step": 2280 }, { "epoch": 0.38, "learning_rate": 2.4522083058668426e-09, "loss": 0.1482, "step": 2290 }, { "epoch": 0.38, "learning_rate": 2.4192485168094923e-09, "loss": 0.1359, "step": 2300 }, { "epoch": 0.38, "learning_rate": 2.3862887277521424e-09, "loss": 0.1612, "step": 2310 }, { "epoch": 0.38, "learning_rate": 2.353328938694792e-09, "loss": 0.1447, "step": 2320 }, { "epoch": 0.38, "learning_rate": 2.3203691496374422e-09, "loss": 0.1505, "step": 2330 }, { "epoch": 0.39, "learning_rate": 2.2874093605800924e-09, "loss": 0.1442, "step": 2340 }, { "epoch": 0.39, "learning_rate": 2.2544495715227425e-09, "loss": 0.1506, "step": 2350 }, { "epoch": 0.39, "learning_rate": 2.221489782465392e-09, "loss": 0.1218, "step": 2360 }, { "epoch": 0.39, "learning_rate": 2.1885299934080423e-09, "loss": 0.1285, "step": 2370 }, { "epoch": 0.39, "learning_rate": 2.1555702043506924e-09, "loss": 0.1538, "step": 2380 }, { "epoch": 0.39, "learning_rate": 2.122610415293342e-09, "loss": 0.1398, "step": 2390 }, { "epoch": 0.4, "learning_rate": 2.089650626235992e-09, "loss": 0.1287, "step": 2400 }, { "epoch": 0.4, "learning_rate": 2.0566908371786423e-09, "loss": 0.15, "step": 2410 }, { "epoch": 0.4, "learning_rate": 2.023731048121292e-09, "loss": 0.143, "step": 2420 }, { "epoch": 0.4, "learning_rate": 1.990771259063942e-09, "loss": 0.1306, "step": 2430 }, { "epoch": 0.4, "learning_rate": 1.957811470006592e-09, "loss": 0.1458, "step": 2440 }, { "epoch": 0.4, "learning_rate": 1.924851680949242e-09, "loss": 0.1264, "step": 2450 }, { "epoch": 0.41, "learning_rate": 1.891891891891892e-09, "loss": 0.1267, "step": 2460 }, { "epoch": 0.41, "learning_rate": 1.8589321028345417e-09, "loss": 0.1587, "step": 2470 }, { "epoch": 0.41, "learning_rate": 1.8259723137771919e-09, "loss": 0.1395, "step": 2480 }, { "epoch": 0.41, "learning_rate": 1.7930125247198418e-09, "loss": 0.1441, "step": 2490 }, { "epoch": 0.41, "learning_rate": 1.7600527356624919e-09, "loss": 0.1295, "step": 2500 }, { "epoch": 0.41, "learning_rate": 1.727092946605142e-09, "loss": 0.148, "step": 2510 }, { "epoch": 0.42, "learning_rate": 1.6941331575477917e-09, "loss": 0.1483, "step": 2520 }, { "epoch": 0.42, "learning_rate": 1.6611733684904418e-09, "loss": 0.1288, "step": 2530 }, { "epoch": 0.42, "learning_rate": 1.6282135794330915e-09, "loss": 0.1616, "step": 2540 }, { "epoch": 0.42, "learning_rate": 1.5952537903757416e-09, "loss": 0.1698, "step": 2550 }, { "epoch": 0.42, "learning_rate": 1.5622940013183917e-09, "loss": 0.1292, "step": 2560 }, { "epoch": 0.42, "learning_rate": 1.5293342122610416e-09, "loss": 0.1345, "step": 2570 }, { "epoch": 0.43, "learning_rate": 1.4963744232036915e-09, "loss": 0.1221, "step": 2580 }, { "epoch": 0.43, "learning_rate": 1.4634146341463414e-09, "loss": 0.1431, "step": 2590 }, { "epoch": 0.43, "learning_rate": 1.4304548450889915e-09, "loss": 0.1204, "step": 2600 }, { "epoch": 0.43, "learning_rate": 1.3974950560316416e-09, "loss": 0.114, "step": 2610 }, { "epoch": 0.43, "learning_rate": 1.3645352669742913e-09, "loss": 0.1154, "step": 2620 }, { "epoch": 0.43, "learning_rate": 1.3315754779169415e-09, "loss": 0.1349, "step": 2630 }, { "epoch": 0.44, "learning_rate": 1.2986156888595912e-09, "loss": 0.1392, "step": 2640 }, { "epoch": 0.44, "learning_rate": 1.2656558998022413e-09, "loss": 0.148, "step": 2650 }, { "epoch": 0.44, "learning_rate": 1.2326961107448912e-09, "loss": 0.1355, "step": 2660 }, { "epoch": 0.44, "learning_rate": 1.1997363216875413e-09, "loss": 0.1326, "step": 2670 }, { "epoch": 0.44, "learning_rate": 1.1667765326301912e-09, "loss": 0.1444, "step": 2680 }, { "epoch": 0.44, "learning_rate": 1.133816743572841e-09, "loss": 0.1281, "step": 2690 }, { "epoch": 0.44, "learning_rate": 1.1008569545154912e-09, "loss": 0.1109, "step": 2700 }, { "epoch": 0.45, "learning_rate": 1.0678971654581411e-09, "loss": 0.1293, "step": 2710 }, { "epoch": 0.45, "learning_rate": 1.034937376400791e-09, "loss": 0.1348, "step": 2720 }, { "epoch": 0.45, "learning_rate": 1.001977587343441e-09, "loss": 0.1232, "step": 2730 }, { "epoch": 0.45, "learning_rate": 9.69017798286091e-10, "loss": 0.1364, "step": 2740 }, { "epoch": 0.45, "learning_rate": 9.36058009228741e-10, "loss": 0.1253, "step": 2750 }, { "epoch": 0.45, "learning_rate": 9.030982201713909e-10, "loss": 0.1285, "step": 2760 }, { "epoch": 0.46, "learning_rate": 8.701384311140409e-10, "loss": 0.1404, "step": 2770 }, { "epoch": 0.46, "learning_rate": 8.371786420566909e-10, "loss": 0.151, "step": 2780 }, { "epoch": 0.46, "learning_rate": 8.042188529993408e-10, "loss": 0.1253, "step": 2790 }, { "epoch": 0.46, "learning_rate": 7.712590639419908e-10, "loss": 0.1201, "step": 2800 }, { "epoch": 0.46, "learning_rate": 7.382992748846408e-10, "loss": 0.1506, "step": 2810 }, { "epoch": 0.46, "learning_rate": 7.053394858272908e-10, "loss": 0.1536, "step": 2820 }, { "epoch": 0.47, "learning_rate": 6.723796967699407e-10, "loss": 0.1494, "step": 2830 }, { "epoch": 0.47, "learning_rate": 6.394199077125906e-10, "loss": 0.136, "step": 2840 }, { "epoch": 0.47, "learning_rate": 6.064601186552406e-10, "loss": 0.1384, "step": 2850 }, { "epoch": 0.47, "learning_rate": 5.735003295978906e-10, "loss": 0.1418, "step": 2860 }, { "epoch": 0.47, "learning_rate": 5.405405405405406e-10, "loss": 0.1436, "step": 2870 }, { "epoch": 0.47, "learning_rate": 5.075807514831905e-10, "loss": 0.1458, "step": 2880 }, { "epoch": 0.48, "learning_rate": 4.746209624258404e-10, "loss": 0.1354, "step": 2890 }, { "epoch": 0.48, "learning_rate": 4.416611733684905e-10, "loss": 0.1269, "step": 2900 }, { "epoch": 0.48, "learning_rate": 4.0870138431114044e-10, "loss": 0.1444, "step": 2910 }, { "epoch": 0.48, "learning_rate": 3.7574159525379034e-10, "loss": 0.1406, "step": 2920 }, { "epoch": 0.48, "learning_rate": 3.4278180619644035e-10, "loss": 0.1439, "step": 2930 }, { "epoch": 0.48, "learning_rate": 3.098220171390903e-10, "loss": 0.1371, "step": 2940 }, { "epoch": 0.49, "learning_rate": 2.7686222808174026e-10, "loss": 0.1566, "step": 2950 }, { "epoch": 0.49, "learning_rate": 2.4390243902439027e-10, "loss": 0.1367, "step": 2960 }, { "epoch": 0.49, "learning_rate": 2.109426499670402e-10, "loss": 0.1201, "step": 2970 }, { "epoch": 0.49, "learning_rate": 1.7798286090969018e-10, "loss": 0.1355, "step": 2980 }, { "epoch": 0.49, "learning_rate": 1.4502307185234014e-10, "loss": 0.1304, "step": 2990 }, { "epoch": 0.49, "learning_rate": 1.1206328279499011e-10, "loss": 0.1319, "step": 3000 }, { "epoch": 0.5, "learning_rate": 7.910349373764008e-11, "loss": 0.1249, "step": 3010 }, { "epoch": 0.5, "learning_rate": 4.614370468029005e-11, "loss": 0.1378, "step": 3020 }, { "epoch": 0.5, "learning_rate": 1.3183915622940013e-11, "loss": 0.1469, "step": 3030 }, { "epoch": 0.5, "step": 3034, "total_flos": 1.0736841266823168e+16, "train_loss": 0.15334907445219212, "train_runtime": 5619.9104, "train_samples_per_second": 17.277, "train_steps_per_second": 0.54 } ], "max_steps": 3034, "num_train_epochs": 1, "total_flos": 1.0736841266823168e+16, "trial_name": null, "trial_params": null }