| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.9215976407100398, | |
| "eval_steps": 500, | |
| "global_step": 180000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 9.974400065535832e-05, | |
| "loss": 1.4263, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.948800131071665e-05, | |
| "loss": 0.6371, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.923200196607497e-05, | |
| "loss": 0.4233, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.89760026214333e-05, | |
| "loss": 0.352, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.872000327679162e-05, | |
| "loss": 0.3121, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.846400393214995e-05, | |
| "loss": 0.2867, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.820800458750826e-05, | |
| "loss": 0.269, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.795200524286658e-05, | |
| "loss": 0.2543, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.769600589822491e-05, | |
| "loss": 0.2435, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.744000655358323e-05, | |
| "loss": 0.2324, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.718400720894156e-05, | |
| "loss": 0.2249, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.692800786429987e-05, | |
| "loss": 0.2178, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.66720085196582e-05, | |
| "loss": 0.2111, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.641600917501652e-05, | |
| "loss": 0.2062, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.616000983037484e-05, | |
| "loss": 0.201, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.590401048573317e-05, | |
| "loss": 0.1965, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.564801114109148e-05, | |
| "loss": 0.1914, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.539201179644981e-05, | |
| "loss": 0.1883, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.513601245180813e-05, | |
| "loss": 0.1849, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.488001310716645e-05, | |
| "loss": 0.1809, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.462401376252478e-05, | |
| "loss": 0.1774, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.43680144178831e-05, | |
| "loss": 0.176, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.411201507324143e-05, | |
| "loss": 0.1731, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.385601572859974e-05, | |
| "loss": 0.1699, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.360001638395807e-05, | |
| "loss": 0.1689, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.334401703931639e-05, | |
| "loss": 0.1662, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.30880176946747e-05, | |
| "loss": 0.1641, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.283201835003302e-05, | |
| "loss": 0.1626, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.257601900539134e-05, | |
| "loss": 0.1604, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.232001966074967e-05, | |
| "loss": 0.159, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.206402031610799e-05, | |
| "loss": 0.1568, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.180802097146632e-05, | |
| "loss": 0.1544, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.155202162682463e-05, | |
| "loss": 0.1542, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.129602228218296e-05, | |
| "loss": 0.1523, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.104002293754128e-05, | |
| "loss": 0.1512, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.07840235928996e-05, | |
| "loss": 0.1502, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.052802424825793e-05, | |
| "loss": 0.1484, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.027202490361624e-05, | |
| "loss": 0.1465, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.001602555897457e-05, | |
| "loss": 0.1468, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 8.976002621433289e-05, | |
| "loss": 0.1454, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 8.950402686969122e-05, | |
| "loss": 0.1435, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 8.924802752504954e-05, | |
| "loss": 0.1431, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 8.899202818040785e-05, | |
| "loss": 0.1425, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 8.873602883576618e-05, | |
| "loss": 0.1406, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 8.84800294911245e-05, | |
| "loss": 0.14, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 8.822403014648283e-05, | |
| "loss": 0.1393, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 8.796803080184115e-05, | |
| "loss": 0.1386, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 8.771203145719948e-05, | |
| "loss": 0.1378, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 8.74560321125578e-05, | |
| "loss": 0.1372, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 8.720003276791611e-05, | |
| "loss": 0.1361, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 8.694403342327444e-05, | |
| "loss": 0.1343, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 8.668803407863276e-05, | |
| "loss": 0.1345, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 8.643203473399109e-05, | |
| "loss": 0.1335, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 8.61760353893494e-05, | |
| "loss": 0.1321, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 8.592003604470774e-05, | |
| "loss": 0.1322, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 8.566403670006605e-05, | |
| "loss": 0.1329, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 8.540803735542437e-05, | |
| "loss": 0.1311, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 8.51520380107827e-05, | |
| "loss": 0.1303, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 8.489603866614102e-05, | |
| "loss": 0.1291, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 8.464003932149935e-05, | |
| "loss": 0.1279, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 8.438403997685766e-05, | |
| "loss": 0.1284, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 8.4128040632216e-05, | |
| "loss": 0.1274, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 8.387204128757431e-05, | |
| "loss": 0.1273, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 8.361604194293263e-05, | |
| "loss": 0.1271, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 8.336004259829096e-05, | |
| "loss": 0.1271, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 8.310404325364927e-05, | |
| "loss": 0.1246, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 8.28480439090076e-05, | |
| "loss": 0.1253, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 8.259204456436592e-05, | |
| "loss": 0.1245, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 8.233604521972424e-05, | |
| "loss": 0.1251, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 8.208004587508257e-05, | |
| "loss": 0.1233, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 8.182404653044089e-05, | |
| "loss": 0.1233, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 8.156804718579922e-05, | |
| "loss": 0.1227, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 8.131204784115753e-05, | |
| "loss": 0.1223, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 8.105604849651586e-05, | |
| "loss": 0.1215, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 8.080004915187418e-05, | |
| "loss": 0.1211, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 8.05440498072325e-05, | |
| "loss": 0.1201, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.028805046259083e-05, | |
| "loss": 0.121, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.003205111794914e-05, | |
| "loss": 0.1205, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 7.977605177330746e-05, | |
| "loss": 0.1214, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 7.952005242866578e-05, | |
| "loss": 0.1192, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 7.926405308402411e-05, | |
| "loss": 0.1189, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 7.900805373938242e-05, | |
| "loss": 0.1184, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 7.875205439474075e-05, | |
| "loss": 0.1179, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 7.849605505009907e-05, | |
| "loss": 0.1177, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 7.824005570545739e-05, | |
| "loss": 0.1177, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 7.798405636081572e-05, | |
| "loss": 0.1175, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 7.772805701617403e-05, | |
| "loss": 0.1166, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 7.747205767153236e-05, | |
| "loss": 0.1171, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 7.721605832689068e-05, | |
| "loss": 0.1161, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 7.696005898224901e-05, | |
| "loss": 0.1157, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 7.670405963760733e-05, | |
| "loss": 0.116, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 7.644806029296564e-05, | |
| "loss": 0.116, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 7.619206094832398e-05, | |
| "loss": 0.1152, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 7.593606160368229e-05, | |
| "loss": 0.1139, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 7.568006225904062e-05, | |
| "loss": 0.1137, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 7.542406291439894e-05, | |
| "loss": 0.1135, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 7.516806356975727e-05, | |
| "loss": 0.1136, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 7.491206422511559e-05, | |
| "loss": 0.1146, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 7.46560648804739e-05, | |
| "loss": 0.1129, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 7.440006553583223e-05, | |
| "loss": 0.1125, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 7.414406619119055e-05, | |
| "loss": 0.1122, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 7.388806684654888e-05, | |
| "loss": 0.1128, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 7.36320675019072e-05, | |
| "loss": 0.1112, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 7.337606815726553e-05, | |
| "loss": 0.1113, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 7.312006881262384e-05, | |
| "loss": 0.1116, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 7.286406946798216e-05, | |
| "loss": 0.1106, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 7.260807012334049e-05, | |
| "loss": 0.1099, | |
| "step": 53500 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 7.235207077869881e-05, | |
| "loss": 0.111, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 7.209607143405714e-05, | |
| "loss": 0.1097, | |
| "step": 54500 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 7.184007208941545e-05, | |
| "loss": 0.1105, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 7.158407274477378e-05, | |
| "loss": 0.1109, | |
| "step": 55500 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 7.13280734001321e-05, | |
| "loss": 0.1095, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 7.107207405549042e-05, | |
| "loss": 0.1088, | |
| "step": 56500 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 7.081607471084875e-05, | |
| "loss": 0.1089, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 7.056007536620707e-05, | |
| "loss": 0.1083, | |
| "step": 57500 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 7.03040760215654e-05, | |
| "loss": 0.1087, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 7.004807667692371e-05, | |
| "loss": 0.1082, | |
| "step": 58500 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 6.979207733228203e-05, | |
| "loss": 0.1075, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 6.953607798764036e-05, | |
| "loss": 0.1078, | |
| "step": 59500 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 6.928007864299868e-05, | |
| "loss": 0.1075, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 6.9024079298357e-05, | |
| "loss": 0.108, | |
| "step": 60500 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 6.876807995371532e-05, | |
| "loss": 0.1062, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 6.851208060907365e-05, | |
| "loss": 0.1073, | |
| "step": 61500 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 6.825608126443197e-05, | |
| "loss": 0.1071, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 6.800008191979029e-05, | |
| "loss": 0.1064, | |
| "step": 62500 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 6.774408257514862e-05, | |
| "loss": 0.1061, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 6.748808323050693e-05, | |
| "loss": 0.1061, | |
| "step": 63500 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 6.723208388586526e-05, | |
| "loss": 0.1056, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 6.697608454122358e-05, | |
| "loss": 0.1063, | |
| "step": 64500 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 6.672008519658191e-05, | |
| "loss": 0.1061, | |
| "step": 65000 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 6.646408585194023e-05, | |
| "loss": 0.106, | |
| "step": 65500 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 6.620808650729854e-05, | |
| "loss": 0.1054, | |
| "step": 66000 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 6.595208716265686e-05, | |
| "loss": 0.105, | |
| "step": 66500 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 6.569608781801518e-05, | |
| "loss": 0.1044, | |
| "step": 67000 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 6.544008847337351e-05, | |
| "loss": 0.1041, | |
| "step": 67500 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 6.518408912873182e-05, | |
| "loss": 0.1041, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 6.492808978409016e-05, | |
| "loss": 0.1039, | |
| "step": 68500 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 6.467209043944847e-05, | |
| "loss": 0.1037, | |
| "step": 69000 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 6.44160910948068e-05, | |
| "loss": 0.1038, | |
| "step": 69500 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 6.416009175016512e-05, | |
| "loss": 0.1036, | |
| "step": 70000 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 6.390409240552344e-05, | |
| "loss": 0.1035, | |
| "step": 70500 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 6.364809306088177e-05, | |
| "loss": 0.1024, | |
| "step": 71000 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 6.339209371624008e-05, | |
| "loss": 0.1037, | |
| "step": 71500 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 6.313609437159841e-05, | |
| "loss": 0.1021, | |
| "step": 72000 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 6.288009502695673e-05, | |
| "loss": 0.1021, | |
| "step": 72500 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 6.262409568231506e-05, | |
| "loss": 0.1022, | |
| "step": 73000 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 6.236809633767338e-05, | |
| "loss": 0.1025, | |
| "step": 73500 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 6.21120969930317e-05, | |
| "loss": 0.1017, | |
| "step": 74000 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 6.185609764839002e-05, | |
| "loss": 0.1017, | |
| "step": 74500 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 6.160009830374834e-05, | |
| "loss": 0.1019, | |
| "step": 75000 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 6.134409895910667e-05, | |
| "loss": 0.1016, | |
| "step": 75500 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 6.108809961446499e-05, | |
| "loss": 0.101, | |
| "step": 76000 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 6.083210026982331e-05, | |
| "loss": 0.1011, | |
| "step": 76500 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 6.0576100925181634e-05, | |
| "loss": 0.1009, | |
| "step": 77000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 6.032010158053996e-05, | |
| "loss": 0.1013, | |
| "step": 77500 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 6.006410223589828e-05, | |
| "loss": 0.1, | |
| "step": 78000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 5.98081028912566e-05, | |
| "loss": 0.1008, | |
| "step": 78500 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 5.955210354661492e-05, | |
| "loss": 0.0995, | |
| "step": 79000 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 5.9296104201973245e-05, | |
| "loss": 0.1003, | |
| "step": 79500 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 5.904010485733157e-05, | |
| "loss": 0.1001, | |
| "step": 80000 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 5.878410551268989e-05, | |
| "loss": 0.0995, | |
| "step": 80500 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 5.8528106168048216e-05, | |
| "loss": 0.0993, | |
| "step": 81000 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 5.827210682340654e-05, | |
| "loss": 0.0998, | |
| "step": 81500 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 5.8016107478764856e-05, | |
| "loss": 0.099, | |
| "step": 82000 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 5.776010813412318e-05, | |
| "loss": 0.1, | |
| "step": 82500 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 5.75041087894815e-05, | |
| "loss": 0.0994, | |
| "step": 83000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 5.7248109444839826e-05, | |
| "loss": 0.0987, | |
| "step": 83500 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 5.699211010019815e-05, | |
| "loss": 0.0985, | |
| "step": 84000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 5.673611075555647e-05, | |
| "loss": 0.0986, | |
| "step": 84500 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 5.648011141091479e-05, | |
| "loss": 0.0992, | |
| "step": 85000 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 5.6224112066273114e-05, | |
| "loss": 0.0986, | |
| "step": 85500 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 5.596811272163144e-05, | |
| "loss": 0.0981, | |
| "step": 86000 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 5.571211337698976e-05, | |
| "loss": 0.0979, | |
| "step": 86500 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 5.5456114032348084e-05, | |
| "loss": 0.0983, | |
| "step": 87000 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 5.520011468770641e-05, | |
| "loss": 0.0983, | |
| "step": 87500 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 5.494411534306473e-05, | |
| "loss": 0.0978, | |
| "step": 88000 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 5.468811599842305e-05, | |
| "loss": 0.0973, | |
| "step": 88500 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 5.443211665378137e-05, | |
| "loss": 0.0973, | |
| "step": 89000 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 5.4176117309139695e-05, | |
| "loss": 0.0968, | |
| "step": 89500 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 5.392011796449802e-05, | |
| "loss": 0.0971, | |
| "step": 90000 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 5.366411861985634e-05, | |
| "loss": 0.097, | |
| "step": 90500 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5.3408119275214665e-05, | |
| "loss": 0.0971, | |
| "step": 91000 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5.315211993057299e-05, | |
| "loss": 0.0965, | |
| "step": 91500 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5.2896120585931306e-05, | |
| "loss": 0.096, | |
| "step": 92000 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5.264012124128962e-05, | |
| "loss": 0.0965, | |
| "step": 92500 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 5.238412189664794e-05, | |
| "loss": 0.0958, | |
| "step": 93000 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 5.212812255200626e-05, | |
| "loss": 0.0957, | |
| "step": 93500 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 5.1872123207364586e-05, | |
| "loss": 0.0968, | |
| "step": 94000 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 5.161612386272291e-05, | |
| "loss": 0.0958, | |
| "step": 94500 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 5.136012451808123e-05, | |
| "loss": 0.0958, | |
| "step": 95000 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 5.1104125173439556e-05, | |
| "loss": 0.0959, | |
| "step": 95500 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 5.084812582879788e-05, | |
| "loss": 0.0953, | |
| "step": 96000 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 5.05921264841562e-05, | |
| "loss": 0.0945, | |
| "step": 96500 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 5.033612713951452e-05, | |
| "loss": 0.0947, | |
| "step": 97000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 5.0080127794872844e-05, | |
| "loss": 0.0951, | |
| "step": 97500 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.9824128450231174e-05, | |
| "loss": 0.0948, | |
| "step": 98000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.95681291055895e-05, | |
| "loss": 0.095, | |
| "step": 98500 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.9312129760947814e-05, | |
| "loss": 0.0948, | |
| "step": 99000 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.905613041630613e-05, | |
| "loss": 0.0943, | |
| "step": 99500 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.8800131071664454e-05, | |
| "loss": 0.0944, | |
| "step": 100000 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.854413172702278e-05, | |
| "loss": 0.0948, | |
| "step": 100500 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.82881323823811e-05, | |
| "loss": 0.0946, | |
| "step": 101000 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.8032133037739425e-05, | |
| "loss": 0.0942, | |
| "step": 101500 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.777613369309775e-05, | |
| "loss": 0.0943, | |
| "step": 102000 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.752013434845607e-05, | |
| "loss": 0.0933, | |
| "step": 102500 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.726413500381439e-05, | |
| "loss": 0.0935, | |
| "step": 103000 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.700813565917271e-05, | |
| "loss": 0.0931, | |
| "step": 103500 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.6752136314531036e-05, | |
| "loss": 0.094, | |
| "step": 104000 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.649613696988936e-05, | |
| "loss": 0.0935, | |
| "step": 104500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.624013762524768e-05, | |
| "loss": 0.0927, | |
| "step": 105000 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.5984138280606006e-05, | |
| "loss": 0.0935, | |
| "step": 105500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.572813893596433e-05, | |
| "loss": 0.0928, | |
| "step": 106000 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.5472139591322646e-05, | |
| "loss": 0.0923, | |
| "step": 106500 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.521614024668097e-05, | |
| "loss": 0.0925, | |
| "step": 107000 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.4960140902039293e-05, | |
| "loss": 0.092, | |
| "step": 107500 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.470414155739762e-05, | |
| "loss": 0.0925, | |
| "step": 108000 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.444814221275594e-05, | |
| "loss": 0.0922, | |
| "step": 108500 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.4192142868114264e-05, | |
| "loss": 0.0921, | |
| "step": 109000 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.393614352347258e-05, | |
| "loss": 0.0917, | |
| "step": 109500 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.3680144178830904e-05, | |
| "loss": 0.0921, | |
| "step": 110000 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.342414483418923e-05, | |
| "loss": 0.0918, | |
| "step": 110500 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.316814548954755e-05, | |
| "loss": 0.0918, | |
| "step": 111000 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.2912146144905875e-05, | |
| "loss": 0.0915, | |
| "step": 111500 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.265614680026419e-05, | |
| "loss": 0.0913, | |
| "step": 112000 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.2400147455622515e-05, | |
| "loss": 0.0919, | |
| "step": 112500 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.214414811098084e-05, | |
| "loss": 0.0912, | |
| "step": 113000 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.1888148766339155e-05, | |
| "loss": 0.0918, | |
| "step": 113500 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.163214942169748e-05, | |
| "loss": 0.0917, | |
| "step": 114000 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.13761500770558e-05, | |
| "loss": 0.0915, | |
| "step": 114500 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.1120150732414126e-05, | |
| "loss": 0.0908, | |
| "step": 115000 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.086415138777245e-05, | |
| "loss": 0.0906, | |
| "step": 115500 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.060815204313077e-05, | |
| "loss": 0.0914, | |
| "step": 116000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.0352152698489096e-05, | |
| "loss": 0.0904, | |
| "step": 116500 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.009615335384741e-05, | |
| "loss": 0.0905, | |
| "step": 117000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 3.9840154009205736e-05, | |
| "loss": 0.0899, | |
| "step": 117500 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 3.958415466456406e-05, | |
| "loss": 0.0902, | |
| "step": 118000 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 3.932815531992238e-05, | |
| "loss": 0.0902, | |
| "step": 118500 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 3.907215597528071e-05, | |
| "loss": 0.0905, | |
| "step": 119000 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 3.881615663063903e-05, | |
| "loss": 0.0894, | |
| "step": 119500 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 3.856015728599735e-05, | |
| "loss": 0.0898, | |
| "step": 120000 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 3.830415794135567e-05, | |
| "loss": 0.0899, | |
| "step": 120500 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 3.8048158596713994e-05, | |
| "loss": 0.09, | |
| "step": 121000 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 3.779215925207232e-05, | |
| "loss": 0.0896, | |
| "step": 121500 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 3.753615990743064e-05, | |
| "loss": 0.0895, | |
| "step": 122000 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.7280160562788965e-05, | |
| "loss": 0.0895, | |
| "step": 122500 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.702416121814729e-05, | |
| "loss": 0.0897, | |
| "step": 123000 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.6768161873505605e-05, | |
| "loss": 0.0889, | |
| "step": 123500 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.651216252886393e-05, | |
| "loss": 0.0885, | |
| "step": 124000 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 3.625616318422225e-05, | |
| "loss": 0.0888, | |
| "step": 124500 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 3.6000163839580575e-05, | |
| "loss": 0.0891, | |
| "step": 125000 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 3.574416449493889e-05, | |
| "loss": 0.0893, | |
| "step": 125500 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.5488165150297216e-05, | |
| "loss": 0.0888, | |
| "step": 126000 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.523216580565554e-05, | |
| "loss": 0.088, | |
| "step": 126500 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.497616646101386e-05, | |
| "loss": 0.0884, | |
| "step": 127000 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.472016711637218e-05, | |
| "loss": 0.0884, | |
| "step": 127500 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 3.44641677717305e-05, | |
| "loss": 0.0882, | |
| "step": 128000 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 3.4208168427088826e-05, | |
| "loss": 0.0884, | |
| "step": 128500 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 3.395216908244715e-05, | |
| "loss": 0.0883, | |
| "step": 129000 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 3.369616973780547e-05, | |
| "loss": 0.0881, | |
| "step": 129500 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 3.34401703931638e-05, | |
| "loss": 0.0878, | |
| "step": 130000 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 3.3184171048522114e-05, | |
| "loss": 0.0876, | |
| "step": 130500 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 3.292817170388044e-05, | |
| "loss": 0.0873, | |
| "step": 131000 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 3.267217235923876e-05, | |
| "loss": 0.0874, | |
| "step": 131500 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 3.2416173014597084e-05, | |
| "loss": 0.0885, | |
| "step": 132000 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 3.216017366995541e-05, | |
| "loss": 0.0878, | |
| "step": 132500 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 3.190417432531373e-05, | |
| "loss": 0.0874, | |
| "step": 133000 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 3.1648174980672055e-05, | |
| "loss": 0.0869, | |
| "step": 133500 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.139217563603037e-05, | |
| "loss": 0.0876, | |
| "step": 134000 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.1136176291388695e-05, | |
| "loss": 0.0877, | |
| "step": 134500 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.088017694674702e-05, | |
| "loss": 0.0873, | |
| "step": 135000 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.062417760210534e-05, | |
| "loss": 0.087, | |
| "step": 135500 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 3.0368178257463665e-05, | |
| "loss": 0.0864, | |
| "step": 136000 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 3.0112178912821985e-05, | |
| "loss": 0.0866, | |
| "step": 136500 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 2.985617956818031e-05, | |
| "loss": 0.087, | |
| "step": 137000 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 2.9600180223538632e-05, | |
| "loss": 0.0868, | |
| "step": 137500 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 2.9344180878896952e-05, | |
| "loss": 0.0859, | |
| "step": 138000 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 2.9088181534255273e-05, | |
| "loss": 0.0855, | |
| "step": 138500 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 2.8832182189613593e-05, | |
| "loss": 0.0859, | |
| "step": 139000 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 2.8576182844971916e-05, | |
| "loss": 0.0868, | |
| "step": 139500 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 2.832018350033024e-05, | |
| "loss": 0.0862, | |
| "step": 140000 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 2.806418415568856e-05, | |
| "loss": 0.0859, | |
| "step": 140500 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 2.7808184811046883e-05, | |
| "loss": 0.0857, | |
| "step": 141000 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 2.7552185466405207e-05, | |
| "loss": 0.0869, | |
| "step": 141500 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 2.7296186121763527e-05, | |
| "loss": 0.086, | |
| "step": 142000 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 2.704018677712185e-05, | |
| "loss": 0.0855, | |
| "step": 142500 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 2.6784187432480174e-05, | |
| "loss": 0.0855, | |
| "step": 143000 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 2.6528188087838497e-05, | |
| "loss": 0.0849, | |
| "step": 143500 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 2.6272188743196818e-05, | |
| "loss": 0.0849, | |
| "step": 144000 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 2.601618939855514e-05, | |
| "loss": 0.085, | |
| "step": 144500 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 2.5760190053913465e-05, | |
| "loss": 0.0856, | |
| "step": 145000 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 2.5504190709271785e-05, | |
| "loss": 0.0858, | |
| "step": 145500 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 2.5248191364630108e-05, | |
| "loss": 0.0857, | |
| "step": 146000 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 2.499219201998843e-05, | |
| "loss": 0.0848, | |
| "step": 146500 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 2.4736192675346752e-05, | |
| "loss": 0.0848, | |
| "step": 147000 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 2.4480193330705075e-05, | |
| "loss": 0.0848, | |
| "step": 147500 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 2.42241939860634e-05, | |
| "loss": 0.0848, | |
| "step": 148000 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 2.396819464142172e-05, | |
| "loss": 0.085, | |
| "step": 148500 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 2.371219529678004e-05, | |
| "loss": 0.0835, | |
| "step": 149000 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.3456195952138363e-05, | |
| "loss": 0.0852, | |
| "step": 149500 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.3200196607496686e-05, | |
| "loss": 0.0846, | |
| "step": 150000 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.2944197262855006e-05, | |
| "loss": 0.0846, | |
| "step": 150500 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.268819791821333e-05, | |
| "loss": 0.0839, | |
| "step": 151000 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.2432198573571653e-05, | |
| "loss": 0.0846, | |
| "step": 151500 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.2176199228929977e-05, | |
| "loss": 0.0844, | |
| "step": 152000 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.1920199884288297e-05, | |
| "loss": 0.0841, | |
| "step": 152500 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.166420053964662e-05, | |
| "loss": 0.0838, | |
| "step": 153000 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 2.1408201195004944e-05, | |
| "loss": 0.0841, | |
| "step": 153500 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 2.1152201850363264e-05, | |
| "loss": 0.0841, | |
| "step": 154000 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 2.0896202505721587e-05, | |
| "loss": 0.0841, | |
| "step": 154500 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 2.0640203161079908e-05, | |
| "loss": 0.0838, | |
| "step": 155000 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.038420381643823e-05, | |
| "loss": 0.0837, | |
| "step": 155500 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.012820447179655e-05, | |
| "loss": 0.0833, | |
| "step": 156000 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.9872205127154875e-05, | |
| "loss": 0.084, | |
| "step": 156500 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.9616205782513198e-05, | |
| "loss": 0.0835, | |
| "step": 157000 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.9360206437871518e-05, | |
| "loss": 0.0826, | |
| "step": 157500 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.9104207093229842e-05, | |
| "loss": 0.0832, | |
| "step": 158000 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.8848207748588165e-05, | |
| "loss": 0.0833, | |
| "step": 158500 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.859220840394649e-05, | |
| "loss": 0.0833, | |
| "step": 159000 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.833620905930481e-05, | |
| "loss": 0.0832, | |
| "step": 159500 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.8080209714663132e-05, | |
| "loss": 0.0833, | |
| "step": 160000 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.7824210370021456e-05, | |
| "loss": 0.0825, | |
| "step": 160500 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.7568211025379776e-05, | |
| "loss": 0.0825, | |
| "step": 161000 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.7312211680738096e-05, | |
| "loss": 0.0831, | |
| "step": 161500 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.705621233609642e-05, | |
| "loss": 0.0822, | |
| "step": 162000 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.6800212991454743e-05, | |
| "loss": 0.083, | |
| "step": 162500 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.6544213646813063e-05, | |
| "loss": 0.0823, | |
| "step": 163000 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.6288214302171387e-05, | |
| "loss": 0.0827, | |
| "step": 163500 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.603221495752971e-05, | |
| "loss": 0.0823, | |
| "step": 164000 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.577621561288803e-05, | |
| "loss": 0.0819, | |
| "step": 164500 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.5520216268246354e-05, | |
| "loss": 0.0823, | |
| "step": 165000 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.5264216923604677e-05, | |
| "loss": 0.0816, | |
| "step": 165500 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.5008217578962999e-05, | |
| "loss": 0.0817, | |
| "step": 166000 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.4752218234321321e-05, | |
| "loss": 0.0822, | |
| "step": 166500 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.4496218889679644e-05, | |
| "loss": 0.0816, | |
| "step": 167000 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.4240219545037966e-05, | |
| "loss": 0.082, | |
| "step": 167500 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.3984220200396286e-05, | |
| "loss": 0.0821, | |
| "step": 168000 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.3728220855754608e-05, | |
| "loss": 0.0819, | |
| "step": 168500 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.3472221511112932e-05, | |
| "loss": 0.0819, | |
| "step": 169000 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.3216222166471253e-05, | |
| "loss": 0.0817, | |
| "step": 169500 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.2960222821829577e-05, | |
| "loss": 0.0816, | |
| "step": 170000 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.2704223477187899e-05, | |
| "loss": 0.0817, | |
| "step": 170500 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.244822413254622e-05, | |
| "loss": 0.0819, | |
| "step": 171000 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.2192224787904544e-05, | |
| "loss": 0.0818, | |
| "step": 171500 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.1936225443262866e-05, | |
| "loss": 0.082, | |
| "step": 172000 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.168022609862119e-05, | |
| "loss": 0.0812, | |
| "step": 172500 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.142422675397951e-05, | |
| "loss": 0.0809, | |
| "step": 173000 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.1168227409337833e-05, | |
| "loss": 0.081, | |
| "step": 173500 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.0912228064696155e-05, | |
| "loss": 0.0811, | |
| "step": 174000 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.0656228720054477e-05, | |
| "loss": 0.0807, | |
| "step": 174500 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.04002293754128e-05, | |
| "loss": 0.0811, | |
| "step": 175000 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.0144230030771122e-05, | |
| "loss": 0.081, | |
| "step": 175500 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 9.888230686129445e-06, | |
| "loss": 0.0812, | |
| "step": 176000 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 9.632231341487766e-06, | |
| "loss": 0.0804, | |
| "step": 176500 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 9.376231996846087e-06, | |
| "loss": 0.0811, | |
| "step": 177000 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 9.120232652204411e-06, | |
| "loss": 0.0803, | |
| "step": 177500 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 8.864233307562733e-06, | |
| "loss": 0.081, | |
| "step": 178000 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 8.608233962921056e-06, | |
| "loss": 0.0803, | |
| "step": 178500 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 8.352234618279378e-06, | |
| "loss": 0.0811, | |
| "step": 179000 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 8.0962352736377e-06, | |
| "loss": 0.0804, | |
| "step": 179500 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 7.840235928996022e-06, | |
| "loss": 0.0806, | |
| "step": 180000 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 195313, | |
| "num_train_epochs": 1, | |
| "save_steps": 10000, | |
| "total_flos": 4.72531191201792e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |