{ "best_metric": null, "best_model_checkpoint": null, "epoch": 1.0, "global_step": 366682, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 9.986364206587725e-06, "loss": 0.37, "step": 500 }, { "epoch": 0.0, "learning_rate": 9.97272841317545e-06, "loss": 0.212, "step": 1000 }, { "epoch": 0.0, "learning_rate": 9.959092619763175e-06, "loss": 0.2018, "step": 1500 }, { "epoch": 0.01, "learning_rate": 9.9454568263509e-06, "loss": 0.1932, "step": 2000 }, { "epoch": 0.01, "learning_rate": 9.931821032938624e-06, "loss": 0.182, "step": 2500 }, { "epoch": 0.01, "learning_rate": 9.918185239526348e-06, "loss": 0.1845, "step": 3000 }, { "epoch": 0.01, "learning_rate": 9.904549446114073e-06, "loss": 0.1735, "step": 3500 }, { "epoch": 0.01, "learning_rate": 9.890913652701796e-06, "loss": 0.1763, "step": 4000 }, { "epoch": 0.01, "learning_rate": 9.87727785928952e-06, "loss": 0.1745, "step": 4500 }, { "epoch": 0.01, "learning_rate": 9.863642065877245e-06, "loss": 0.1688, "step": 5000 }, { "epoch": 0.01, "learning_rate": 9.85000627246497e-06, "loss": 0.1699, "step": 5500 }, { "epoch": 0.02, "learning_rate": 9.836370479052694e-06, "loss": 0.173, "step": 6000 }, { "epoch": 0.02, "learning_rate": 9.822734685640419e-06, "loss": 0.1632, "step": 6500 }, { "epoch": 0.02, "learning_rate": 9.809098892228144e-06, "loss": 0.1675, "step": 7000 }, { "epoch": 0.02, "learning_rate": 9.795463098815868e-06, "loss": 0.1608, "step": 7500 }, { "epoch": 0.02, "learning_rate": 9.781827305403593e-06, "loss": 0.1609, "step": 8000 }, { "epoch": 0.02, "learning_rate": 9.768191511991318e-06, "loss": 0.1587, "step": 8500 }, { "epoch": 0.02, "learning_rate": 9.754555718579042e-06, "loss": 0.1552, "step": 9000 }, { "epoch": 0.03, "learning_rate": 9.740919925166767e-06, "loss": 0.1606, "step": 9500 }, { "epoch": 0.03, "learning_rate": 9.727284131754491e-06, "loss": 0.1578, "step": 10000 }, { "epoch": 0.03, "learning_rate": 9.713648338342216e-06, "loss": 0.154, "step": 10500 }, { "epoch": 0.03, "learning_rate": 9.70001254492994e-06, "loss": 0.1556, "step": 11000 }, { "epoch": 0.03, "learning_rate": 9.686376751517664e-06, "loss": 0.1517, "step": 11500 }, { "epoch": 0.03, "learning_rate": 9.672740958105388e-06, "loss": 0.1591, "step": 12000 }, { "epoch": 0.03, "learning_rate": 9.659105164693113e-06, "loss": 0.1489, "step": 12500 }, { "epoch": 0.04, "learning_rate": 9.645469371280837e-06, "loss": 0.15, "step": 13000 }, { "epoch": 0.04, "learning_rate": 9.631833577868562e-06, "loss": 0.1517, "step": 13500 }, { "epoch": 0.04, "learning_rate": 9.618197784456287e-06, "loss": 0.1516, "step": 14000 }, { "epoch": 0.04, "learning_rate": 9.604561991044011e-06, "loss": 0.1433, "step": 14500 }, { "epoch": 0.04, "learning_rate": 9.590926197631736e-06, "loss": 0.1494, "step": 15000 }, { "epoch": 0.04, "learning_rate": 9.57729040421946e-06, "loss": 0.1531, "step": 15500 }, { "epoch": 0.04, "learning_rate": 9.563654610807185e-06, "loss": 0.151, "step": 16000 }, { "epoch": 0.04, "learning_rate": 9.55001881739491e-06, "loss": 0.154, "step": 16500 }, { "epoch": 0.05, "learning_rate": 9.536383023982634e-06, "loss": 0.1461, "step": 17000 }, { "epoch": 0.05, "learning_rate": 9.522747230570359e-06, "loss": 0.1504, "step": 17500 }, { "epoch": 0.05, "learning_rate": 9.509111437158083e-06, "loss": 0.1468, "step": 18000 }, { "epoch": 0.05, "learning_rate": 9.495475643745808e-06, "loss": 0.1477, "step": 18500 }, { "epoch": 0.05, "learning_rate": 9.481839850333531e-06, "loss": 0.1421, "step": 19000 }, { "epoch": 0.05, "learning_rate": 9.468204056921256e-06, "loss": 0.1435, "step": 19500 }, { "epoch": 0.05, "learning_rate": 9.45456826350898e-06, "loss": 0.1435, "step": 20000 }, { "epoch": 0.06, "learning_rate": 9.440932470096705e-06, "loss": 0.1447, "step": 20500 }, { "epoch": 0.06, "learning_rate": 9.42729667668443e-06, "loss": 0.1407, "step": 21000 }, { "epoch": 0.06, "learning_rate": 9.413660883272154e-06, "loss": 0.141, "step": 21500 }, { "epoch": 0.06, "learning_rate": 9.400025089859879e-06, "loss": 0.1452, "step": 22000 }, { "epoch": 0.06, "learning_rate": 9.386389296447603e-06, "loss": 0.1399, "step": 22500 }, { "epoch": 0.06, "learning_rate": 9.372753503035328e-06, "loss": 0.1445, "step": 23000 }, { "epoch": 0.06, "learning_rate": 9.359117709623054e-06, "loss": 0.1426, "step": 23500 }, { "epoch": 0.07, "learning_rate": 9.345481916210777e-06, "loss": 0.1432, "step": 24000 }, { "epoch": 0.07, "learning_rate": 9.331846122798502e-06, "loss": 0.1427, "step": 24500 }, { "epoch": 0.07, "learning_rate": 9.318210329386226e-06, "loss": 0.1429, "step": 25000 }, { "epoch": 0.07, "learning_rate": 9.304574535973951e-06, "loss": 0.1417, "step": 25500 }, { "epoch": 0.07, "learning_rate": 9.290938742561676e-06, "loss": 0.1429, "step": 26000 }, { "epoch": 0.07, "learning_rate": 9.2773029491494e-06, "loss": 0.1359, "step": 26500 }, { "epoch": 0.07, "learning_rate": 9.263667155737125e-06, "loss": 0.1362, "step": 27000 }, { "epoch": 0.07, "learning_rate": 9.25003136232485e-06, "loss": 0.1392, "step": 27500 }, { "epoch": 0.08, "learning_rate": 9.236395568912574e-06, "loss": 0.1433, "step": 28000 }, { "epoch": 0.08, "learning_rate": 9.222759775500299e-06, "loss": 0.139, "step": 28500 }, { "epoch": 0.08, "learning_rate": 9.209123982088023e-06, "loss": 0.1368, "step": 29000 }, { "epoch": 0.08, "learning_rate": 9.195488188675748e-06, "loss": 0.1443, "step": 29500 }, { "epoch": 0.08, "learning_rate": 9.181852395263472e-06, "loss": 0.1434, "step": 30000 }, { "epoch": 0.08, "learning_rate": 9.168216601851197e-06, "loss": 0.1418, "step": 30500 }, { "epoch": 0.08, "learning_rate": 9.154580808438922e-06, "loss": 0.1406, "step": 31000 }, { "epoch": 0.09, "learning_rate": 9.140945015026645e-06, "loss": 0.1371, "step": 31500 }, { "epoch": 0.09, "learning_rate": 9.12730922161437e-06, "loss": 0.1397, "step": 32000 }, { "epoch": 0.09, "learning_rate": 9.113673428202094e-06, "loss": 0.1382, "step": 32500 }, { "epoch": 0.09, "learning_rate": 9.100037634789818e-06, "loss": 0.1397, "step": 33000 }, { "epoch": 0.09, "learning_rate": 9.086401841377543e-06, "loss": 0.1358, "step": 33500 }, { "epoch": 0.09, "learning_rate": 9.072766047965268e-06, "loss": 0.1387, "step": 34000 }, { "epoch": 0.09, "learning_rate": 9.059130254552992e-06, "loss": 0.1358, "step": 34500 }, { "epoch": 0.1, "learning_rate": 9.045494461140717e-06, "loss": 0.1373, "step": 35000 }, { "epoch": 0.1, "learning_rate": 9.031858667728442e-06, "loss": 0.1354, "step": 35500 }, { "epoch": 0.1, "learning_rate": 9.018222874316166e-06, "loss": 0.1375, "step": 36000 }, { "epoch": 0.1, "learning_rate": 9.00458708090389e-06, "loss": 0.1366, "step": 36500 }, { "epoch": 0.1, "learning_rate": 8.990951287491615e-06, "loss": 0.1361, "step": 37000 }, { "epoch": 0.1, "learning_rate": 8.97731549407934e-06, "loss": 0.1393, "step": 37500 }, { "epoch": 0.1, "learning_rate": 8.963679700667065e-06, "loss": 0.1361, "step": 38000 }, { "epoch": 0.1, "learning_rate": 8.95004390725479e-06, "loss": 0.1291, "step": 38500 }, { "epoch": 0.11, "learning_rate": 8.936408113842512e-06, "loss": 0.1373, "step": 39000 }, { "epoch": 0.11, "learning_rate": 8.922772320430237e-06, "loss": 0.1337, "step": 39500 }, { "epoch": 0.11, "learning_rate": 8.909136527017961e-06, "loss": 0.14, "step": 40000 }, { "epoch": 0.11, "learning_rate": 8.895500733605686e-06, "loss": 0.1321, "step": 40500 }, { "epoch": 0.11, "learning_rate": 8.88186494019341e-06, "loss": 0.1342, "step": 41000 }, { "epoch": 0.11, "learning_rate": 8.868229146781135e-06, "loss": 0.1341, "step": 41500 }, { "epoch": 0.11, "learning_rate": 8.85459335336886e-06, "loss": 0.1336, "step": 42000 }, { "epoch": 0.12, "learning_rate": 8.840957559956584e-06, "loss": 0.133, "step": 42500 }, { "epoch": 0.12, "learning_rate": 8.827321766544309e-06, "loss": 0.1355, "step": 43000 }, { "epoch": 0.12, "learning_rate": 8.813685973132034e-06, "loss": 0.1329, "step": 43500 }, { "epoch": 0.12, "learning_rate": 8.800050179719758e-06, "loss": 0.1279, "step": 44000 }, { "epoch": 0.12, "learning_rate": 8.786414386307483e-06, "loss": 0.1341, "step": 44500 }, { "epoch": 0.12, "learning_rate": 8.772778592895207e-06, "loss": 0.1397, "step": 45000 }, { "epoch": 0.12, "learning_rate": 8.759142799482932e-06, "loss": 0.1313, "step": 45500 }, { "epoch": 0.13, "learning_rate": 8.745507006070657e-06, "loss": 0.1299, "step": 46000 }, { "epoch": 0.13, "learning_rate": 8.73187121265838e-06, "loss": 0.1322, "step": 46500 }, { "epoch": 0.13, "learning_rate": 8.718235419246104e-06, "loss": 0.1342, "step": 47000 }, { "epoch": 0.13, "learning_rate": 8.704599625833829e-06, "loss": 0.1324, "step": 47500 }, { "epoch": 0.13, "learning_rate": 8.690963832421553e-06, "loss": 0.1326, "step": 48000 }, { "epoch": 0.13, "learning_rate": 8.677328039009278e-06, "loss": 0.1284, "step": 48500 }, { "epoch": 0.13, "learning_rate": 8.663692245597003e-06, "loss": 0.1358, "step": 49000 }, { "epoch": 0.13, "learning_rate": 8.650056452184727e-06, "loss": 0.1322, "step": 49500 }, { "epoch": 0.14, "learning_rate": 8.636420658772452e-06, "loss": 0.1287, "step": 50000 }, { "epoch": 0.14, "learning_rate": 8.622784865360177e-06, "loss": 0.1323, "step": 50500 }, { "epoch": 0.14, "learning_rate": 8.609149071947901e-06, "loss": 0.1298, "step": 51000 }, { "epoch": 0.14, "learning_rate": 8.595513278535626e-06, "loss": 0.1323, "step": 51500 }, { "epoch": 0.14, "learning_rate": 8.58187748512335e-06, "loss": 0.1301, "step": 52000 }, { "epoch": 0.14, "learning_rate": 8.568241691711075e-06, "loss": 0.13, "step": 52500 }, { "epoch": 0.14, "learning_rate": 8.5546058982988e-06, "loss": 0.1298, "step": 53000 }, { "epoch": 0.15, "learning_rate": 8.540970104886524e-06, "loss": 0.1285, "step": 53500 }, { "epoch": 0.15, "learning_rate": 8.527334311474247e-06, "loss": 0.1353, "step": 54000 }, { "epoch": 0.15, "learning_rate": 8.513698518061972e-06, "loss": 0.1301, "step": 54500 }, { "epoch": 0.15, "learning_rate": 8.500062724649696e-06, "loss": 0.1255, "step": 55000 }, { "epoch": 0.15, "learning_rate": 8.486426931237421e-06, "loss": 0.1295, "step": 55500 }, { "epoch": 0.15, "learning_rate": 8.472791137825146e-06, "loss": 0.1277, "step": 56000 }, { "epoch": 0.15, "learning_rate": 8.45915534441287e-06, "loss": 0.134, "step": 56500 }, { "epoch": 0.16, "learning_rate": 8.445519551000595e-06, "loss": 0.1309, "step": 57000 }, { "epoch": 0.16, "learning_rate": 8.43188375758832e-06, "loss": 0.1357, "step": 57500 }, { "epoch": 0.16, "learning_rate": 8.418247964176044e-06, "loss": 0.1296, "step": 58000 }, { "epoch": 0.16, "learning_rate": 8.404612170763769e-06, "loss": 0.1253, "step": 58500 }, { "epoch": 0.16, "learning_rate": 8.390976377351493e-06, "loss": 0.1284, "step": 59000 }, { "epoch": 0.16, "learning_rate": 8.377340583939218e-06, "loss": 0.1307, "step": 59500 }, { "epoch": 0.16, "learning_rate": 8.363704790526942e-06, "loss": 0.1279, "step": 60000 }, { "epoch": 0.16, "learning_rate": 8.350068997114667e-06, "loss": 0.1315, "step": 60500 }, { "epoch": 0.17, "learning_rate": 8.336433203702392e-06, "loss": 0.1279, "step": 61000 }, { "epoch": 0.17, "learning_rate": 8.322797410290116e-06, "loss": 0.1305, "step": 61500 }, { "epoch": 0.17, "learning_rate": 8.30916161687784e-06, "loss": 0.1258, "step": 62000 }, { "epoch": 0.17, "learning_rate": 8.295525823465564e-06, "loss": 0.129, "step": 62500 }, { "epoch": 0.17, "learning_rate": 8.281890030053288e-06, "loss": 0.1277, "step": 63000 }, { "epoch": 0.17, "learning_rate": 8.268254236641013e-06, "loss": 0.1283, "step": 63500 }, { "epoch": 0.17, "learning_rate": 8.254618443228738e-06, "loss": 0.1286, "step": 64000 }, { "epoch": 0.18, "learning_rate": 8.240982649816462e-06, "loss": 0.1302, "step": 64500 }, { "epoch": 0.18, "learning_rate": 8.227346856404187e-06, "loss": 0.1262, "step": 65000 }, { "epoch": 0.18, "learning_rate": 8.213711062991911e-06, "loss": 0.1289, "step": 65500 }, { "epoch": 0.18, "learning_rate": 8.200075269579636e-06, "loss": 0.129, "step": 66000 }, { "epoch": 0.18, "learning_rate": 8.18643947616736e-06, "loss": 0.1289, "step": 66500 }, { "epoch": 0.18, "learning_rate": 8.172803682755085e-06, "loss": 0.1299, "step": 67000 }, { "epoch": 0.18, "learning_rate": 8.15916788934281e-06, "loss": 0.1289, "step": 67500 }, { "epoch": 0.19, "learning_rate": 8.145532095930535e-06, "loss": 0.126, "step": 68000 }, { "epoch": 0.19, "learning_rate": 8.13189630251826e-06, "loss": 0.1276, "step": 68500 }, { "epoch": 0.19, "learning_rate": 8.118260509105984e-06, "loss": 0.1245, "step": 69000 }, { "epoch": 0.19, "learning_rate": 8.104624715693708e-06, "loss": 0.1263, "step": 69500 }, { "epoch": 0.19, "learning_rate": 8.090988922281433e-06, "loss": 0.1293, "step": 70000 }, { "epoch": 0.19, "learning_rate": 8.077353128869158e-06, "loss": 0.1255, "step": 70500 }, { "epoch": 0.19, "learning_rate": 8.063717335456882e-06, "loss": 0.1277, "step": 71000 }, { "epoch": 0.19, "learning_rate": 8.050081542044607e-06, "loss": 0.1248, "step": 71500 }, { "epoch": 0.2, "learning_rate": 8.036445748632331e-06, "loss": 0.1199, "step": 72000 }, { "epoch": 0.2, "learning_rate": 8.022809955220056e-06, "loss": 0.1265, "step": 72500 }, { "epoch": 0.2, "learning_rate": 8.00917416180778e-06, "loss": 0.1284, "step": 73000 }, { "epoch": 0.2, "learning_rate": 7.995538368395505e-06, "loss": 0.1292, "step": 73500 }, { "epoch": 0.2, "learning_rate": 7.98190257498323e-06, "loss": 0.1266, "step": 74000 }, { "epoch": 0.2, "learning_rate": 7.968266781570953e-06, "loss": 0.1269, "step": 74500 }, { "epoch": 0.2, "learning_rate": 7.954630988158677e-06, "loss": 0.1238, "step": 75000 }, { "epoch": 0.21, "learning_rate": 7.940995194746402e-06, "loss": 0.1269, "step": 75500 }, { "epoch": 0.21, "learning_rate": 7.927359401334127e-06, "loss": 0.129, "step": 76000 }, { "epoch": 0.21, "learning_rate": 7.913723607921851e-06, "loss": 0.1286, "step": 76500 }, { "epoch": 0.21, "learning_rate": 7.900087814509576e-06, "loss": 0.1225, "step": 77000 }, { "epoch": 0.21, "learning_rate": 7.8864520210973e-06, "loss": 0.1227, "step": 77500 }, { "epoch": 0.21, "learning_rate": 7.872816227685025e-06, "loss": 0.1231, "step": 78000 }, { "epoch": 0.21, "learning_rate": 7.85918043427275e-06, "loss": 0.1261, "step": 78500 }, { "epoch": 0.22, "learning_rate": 7.845544640860474e-06, "loss": 0.1245, "step": 79000 }, { "epoch": 0.22, "learning_rate": 7.831908847448199e-06, "loss": 0.1214, "step": 79500 }, { "epoch": 0.22, "learning_rate": 7.818273054035924e-06, "loss": 0.127, "step": 80000 }, { "epoch": 0.22, "learning_rate": 7.804637260623648e-06, "loss": 0.1292, "step": 80500 }, { "epoch": 0.22, "learning_rate": 7.791001467211373e-06, "loss": 0.1265, "step": 81000 }, { "epoch": 0.22, "learning_rate": 7.777365673799097e-06, "loss": 0.1228, "step": 81500 }, { "epoch": 0.22, "learning_rate": 7.76372988038682e-06, "loss": 0.1248, "step": 82000 }, { "epoch": 0.22, "learning_rate": 7.750094086974545e-06, "loss": 0.119, "step": 82500 }, { "epoch": 0.23, "learning_rate": 7.73645829356227e-06, "loss": 0.1191, "step": 83000 }, { "epoch": 0.23, "learning_rate": 7.722822500149994e-06, "loss": 0.1274, "step": 83500 }, { "epoch": 0.23, "learning_rate": 7.709186706737719e-06, "loss": 0.1232, "step": 84000 }, { "epoch": 0.23, "learning_rate": 7.695550913325443e-06, "loss": 0.1246, "step": 84500 }, { "epoch": 0.23, "learning_rate": 7.681915119913168e-06, "loss": 0.119, "step": 85000 }, { "epoch": 0.23, "learning_rate": 7.668279326500893e-06, "loss": 0.122, "step": 85500 }, { "epoch": 0.23, "learning_rate": 7.654643533088617e-06, "loss": 0.126, "step": 86000 }, { "epoch": 0.24, "learning_rate": 7.641007739676342e-06, "loss": 0.1212, "step": 86500 }, { "epoch": 0.24, "learning_rate": 7.6273719462640664e-06, "loss": 0.121, "step": 87000 }, { "epoch": 0.24, "learning_rate": 7.61373615285179e-06, "loss": 0.1264, "step": 87500 }, { "epoch": 0.24, "learning_rate": 7.600100359439515e-06, "loss": 0.1244, "step": 88000 }, { "epoch": 0.24, "learning_rate": 7.5864645660272394e-06, "loss": 0.1265, "step": 88500 }, { "epoch": 0.24, "learning_rate": 7.572828772614964e-06, "loss": 0.124, "step": 89000 }, { "epoch": 0.24, "learning_rate": 7.559192979202689e-06, "loss": 0.1234, "step": 89500 }, { "epoch": 0.25, "learning_rate": 7.545557185790413e-06, "loss": 0.1206, "step": 90000 }, { "epoch": 0.25, "learning_rate": 7.531921392378138e-06, "loss": 0.1202, "step": 90500 }, { "epoch": 0.25, "learning_rate": 7.5182855989658625e-06, "loss": 0.1272, "step": 91000 }, { "epoch": 0.25, "learning_rate": 7.504649805553586e-06, "loss": 0.1234, "step": 91500 }, { "epoch": 0.25, "learning_rate": 7.491014012141311e-06, "loss": 0.1221, "step": 92000 }, { "epoch": 0.25, "learning_rate": 7.4773782187290355e-06, "loss": 0.1256, "step": 92500 }, { "epoch": 0.25, "learning_rate": 7.46374242531676e-06, "loss": 0.1258, "step": 93000 }, { "epoch": 0.25, "learning_rate": 7.450106631904485e-06, "loss": 0.1222, "step": 93500 }, { "epoch": 0.26, "learning_rate": 7.436470838492209e-06, "loss": 0.1234, "step": 94000 }, { "epoch": 0.26, "learning_rate": 7.422835045079934e-06, "loss": 0.1216, "step": 94500 }, { "epoch": 0.26, "learning_rate": 7.409199251667658e-06, "loss": 0.1192, "step": 95000 }, { "epoch": 0.26, "learning_rate": 7.395563458255382e-06, "loss": 0.1215, "step": 95500 }, { "epoch": 0.26, "learning_rate": 7.381927664843107e-06, "loss": 0.1219, "step": 96000 }, { "epoch": 0.26, "learning_rate": 7.3682918714308315e-06, "loss": 0.1192, "step": 96500 }, { "epoch": 0.26, "learning_rate": 7.354656078018556e-06, "loss": 0.1256, "step": 97000 }, { "epoch": 0.27, "learning_rate": 7.341020284606281e-06, "loss": 0.1268, "step": 97500 }, { "epoch": 0.27, "learning_rate": 7.327384491194005e-06, "loss": 0.1227, "step": 98000 }, { "epoch": 0.27, "learning_rate": 7.31374869778173e-06, "loss": 0.1169, "step": 98500 }, { "epoch": 0.27, "learning_rate": 7.300112904369454e-06, "loss": 0.1201, "step": 99000 }, { "epoch": 0.27, "learning_rate": 7.286477110957178e-06, "loss": 0.1221, "step": 99500 }, { "epoch": 0.27, "learning_rate": 7.272841317544903e-06, "loss": 0.1238, "step": 100000 }, { "epoch": 0.27, "learning_rate": 7.259205524132628e-06, "loss": 0.1213, "step": 100500 }, { "epoch": 0.28, "learning_rate": 7.245569730720352e-06, "loss": 0.1178, "step": 101000 }, { "epoch": 0.28, "learning_rate": 7.231933937308077e-06, "loss": 0.1184, "step": 101500 }, { "epoch": 0.28, "learning_rate": 7.2182981438958014e-06, "loss": 0.1191, "step": 102000 }, { "epoch": 0.28, "learning_rate": 7.204662350483525e-06, "loss": 0.1216, "step": 102500 }, { "epoch": 0.28, "learning_rate": 7.19102655707125e-06, "loss": 0.1225, "step": 103000 }, { "epoch": 0.28, "learning_rate": 7.177390763658974e-06, "loss": 0.1211, "step": 103500 }, { "epoch": 0.28, "learning_rate": 7.163754970246699e-06, "loss": 0.1186, "step": 104000 }, { "epoch": 0.28, "learning_rate": 7.150119176834424e-06, "loss": 0.1213, "step": 104500 }, { "epoch": 0.29, "learning_rate": 7.136483383422148e-06, "loss": 0.1225, "step": 105000 }, { "epoch": 0.29, "learning_rate": 7.122847590009873e-06, "loss": 0.1192, "step": 105500 }, { "epoch": 0.29, "learning_rate": 7.1092117965975975e-06, "loss": 0.1222, "step": 106000 }, { "epoch": 0.29, "learning_rate": 7.095576003185321e-06, "loss": 0.124, "step": 106500 }, { "epoch": 0.29, "learning_rate": 7.081940209773046e-06, "loss": 0.1181, "step": 107000 }, { "epoch": 0.29, "learning_rate": 7.0683044163607705e-06, "loss": 0.1169, "step": 107500 }, { "epoch": 0.29, "learning_rate": 7.054668622948495e-06, "loss": 0.1186, "step": 108000 }, { "epoch": 0.3, "learning_rate": 7.04103282953622e-06, "loss": 0.1219, "step": 108500 }, { "epoch": 0.3, "learning_rate": 7.027397036123944e-06, "loss": 0.1202, "step": 109000 }, { "epoch": 0.3, "learning_rate": 7.013761242711669e-06, "loss": 0.1191, "step": 109500 }, { "epoch": 0.3, "learning_rate": 7.000125449299393e-06, "loss": 0.1214, "step": 110000 }, { "epoch": 0.3, "learning_rate": 6.986489655887117e-06, "loss": 0.1212, "step": 110500 }, { "epoch": 0.3, "learning_rate": 6.972853862474842e-06, "loss": 0.1229, "step": 111000 }, { "epoch": 0.3, "learning_rate": 6.9592180690625665e-06, "loss": 0.1217, "step": 111500 }, { "epoch": 0.31, "learning_rate": 6.945582275650291e-06, "loss": 0.1196, "step": 112000 }, { "epoch": 0.31, "learning_rate": 6.931946482238016e-06, "loss": 0.1169, "step": 112500 }, { "epoch": 0.31, "learning_rate": 6.91831068882574e-06, "loss": 0.1168, "step": 113000 }, { "epoch": 0.31, "learning_rate": 6.904674895413465e-06, "loss": 0.1209, "step": 113500 }, { "epoch": 0.31, "learning_rate": 6.891039102001189e-06, "loss": 0.1217, "step": 114000 }, { "epoch": 0.31, "learning_rate": 6.877403308588913e-06, "loss": 0.1183, "step": 114500 }, { "epoch": 0.31, "learning_rate": 6.863767515176638e-06, "loss": 0.1214, "step": 115000 }, { "epoch": 0.31, "learning_rate": 6.8501317217643634e-06, "loss": 0.125, "step": 115500 }, { "epoch": 0.32, "learning_rate": 6.836495928352088e-06, "loss": 0.1162, "step": 116000 }, { "epoch": 0.32, "learning_rate": 6.822860134939813e-06, "loss": 0.1155, "step": 116500 }, { "epoch": 0.32, "learning_rate": 6.809224341527537e-06, "loss": 0.1218, "step": 117000 }, { "epoch": 0.32, "learning_rate": 6.795588548115262e-06, "loss": 0.1173, "step": 117500 }, { "epoch": 0.32, "learning_rate": 6.7819527547029865e-06, "loss": 0.1197, "step": 118000 }, { "epoch": 0.32, "learning_rate": 6.768316961290711e-06, "loss": 0.1201, "step": 118500 }, { "epoch": 0.32, "learning_rate": 6.754681167878435e-06, "loss": 0.1169, "step": 119000 }, { "epoch": 0.33, "learning_rate": 6.7410453744661595e-06, "loss": 0.1208, "step": 119500 }, { "epoch": 0.33, "learning_rate": 6.727409581053884e-06, "loss": 0.1167, "step": 120000 }, { "epoch": 0.33, "learning_rate": 6.713773787641609e-06, "loss": 0.1218, "step": 120500 }, { "epoch": 0.33, "learning_rate": 6.700137994229333e-06, "loss": 0.1219, "step": 121000 }, { "epoch": 0.33, "learning_rate": 6.686502200817058e-06, "loss": 0.1171, "step": 121500 }, { "epoch": 0.33, "learning_rate": 6.6728664074047825e-06, "loss": 0.1195, "step": 122000 }, { "epoch": 0.33, "learning_rate": 6.659230613992506e-06, "loss": 0.1154, "step": 122500 }, { "epoch": 0.34, "learning_rate": 6.645594820580231e-06, "loss": 0.1135, "step": 123000 }, { "epoch": 0.34, "learning_rate": 6.6319590271679555e-06, "loss": 0.1227, "step": 123500 }, { "epoch": 0.34, "learning_rate": 6.61832323375568e-06, "loss": 0.1203, "step": 124000 }, { "epoch": 0.34, "learning_rate": 6.604687440343405e-06, "loss": 0.1181, "step": 124500 }, { "epoch": 0.34, "learning_rate": 6.591051646931129e-06, "loss": 0.1238, "step": 125000 }, { "epoch": 0.34, "learning_rate": 6.577415853518854e-06, "loss": 0.1185, "step": 125500 }, { "epoch": 0.34, "learning_rate": 6.563780060106579e-06, "loss": 0.1191, "step": 126000 }, { "epoch": 0.34, "learning_rate": 6.550144266694302e-06, "loss": 0.1163, "step": 126500 }, { "epoch": 0.35, "learning_rate": 6.536508473282027e-06, "loss": 0.1187, "step": 127000 }, { "epoch": 0.35, "learning_rate": 6.522872679869752e-06, "loss": 0.12, "step": 127500 }, { "epoch": 0.35, "learning_rate": 6.509236886457476e-06, "loss": 0.1134, "step": 128000 }, { "epoch": 0.35, "learning_rate": 6.495601093045201e-06, "loss": 0.1199, "step": 128500 }, { "epoch": 0.35, "learning_rate": 6.481965299632925e-06, "loss": 0.1169, "step": 129000 }, { "epoch": 0.35, "learning_rate": 6.46832950622065e-06, "loss": 0.119, "step": 129500 }, { "epoch": 0.35, "learning_rate": 6.454693712808374e-06, "loss": 0.1159, "step": 130000 }, { "epoch": 0.36, "learning_rate": 6.441057919396098e-06, "loss": 0.1203, "step": 130500 }, { "epoch": 0.36, "learning_rate": 6.427422125983823e-06, "loss": 0.1191, "step": 131000 }, { "epoch": 0.36, "learning_rate": 6.413786332571548e-06, "loss": 0.1177, "step": 131500 }, { "epoch": 0.36, "learning_rate": 6.400150539159272e-06, "loss": 0.1168, "step": 132000 }, { "epoch": 0.36, "learning_rate": 6.386514745746997e-06, "loss": 0.117, "step": 132500 }, { "epoch": 0.36, "learning_rate": 6.3728789523347215e-06, "loss": 0.1177, "step": 133000 }, { "epoch": 0.36, "learning_rate": 6.359243158922446e-06, "loss": 0.1147, "step": 133500 }, { "epoch": 0.37, "learning_rate": 6.34560736551017e-06, "loss": 0.1151, "step": 134000 }, { "epoch": 0.37, "learning_rate": 6.3319715720978945e-06, "loss": 0.1171, "step": 134500 }, { "epoch": 0.37, "learning_rate": 6.318335778685619e-06, "loss": 0.1126, "step": 135000 }, { "epoch": 0.37, "learning_rate": 6.304699985273344e-06, "loss": 0.1181, "step": 135500 }, { "epoch": 0.37, "learning_rate": 6.291064191861068e-06, "loss": 0.1147, "step": 136000 }, { "epoch": 0.37, "learning_rate": 6.277428398448793e-06, "loss": 0.1149, "step": 136500 }, { "epoch": 0.37, "learning_rate": 6.2637926050365175e-06, "loss": 0.1175, "step": 137000 }, { "epoch": 0.37, "learning_rate": 6.250156811624241e-06, "loss": 0.1157, "step": 137500 }, { "epoch": 0.38, "learning_rate": 6.236521018211966e-06, "loss": 0.1167, "step": 138000 }, { "epoch": 0.38, "learning_rate": 6.2228852247996905e-06, "loss": 0.1139, "step": 138500 }, { "epoch": 0.38, "learning_rate": 6.209249431387415e-06, "loss": 0.1161, "step": 139000 }, { "epoch": 0.38, "learning_rate": 6.19561363797514e-06, "loss": 0.12, "step": 139500 }, { "epoch": 0.38, "learning_rate": 6.181977844562864e-06, "loss": 0.1197, "step": 140000 }, { "epoch": 0.38, "learning_rate": 6.168342051150589e-06, "loss": 0.1187, "step": 140500 }, { "epoch": 0.38, "learning_rate": 6.154706257738314e-06, "loss": 0.1167, "step": 141000 }, { "epoch": 0.39, "learning_rate": 6.141070464326037e-06, "loss": 0.1147, "step": 141500 }, { "epoch": 0.39, "learning_rate": 6.127434670913762e-06, "loss": 0.1124, "step": 142000 }, { "epoch": 0.39, "learning_rate": 6.1137988775014866e-06, "loss": 0.1149, "step": 142500 }, { "epoch": 0.39, "learning_rate": 6.100163084089211e-06, "loss": 0.1151, "step": 143000 }, { "epoch": 0.39, "learning_rate": 6.086527290676936e-06, "loss": 0.1164, "step": 143500 }, { "epoch": 0.39, "learning_rate": 6.07289149726466e-06, "loss": 0.1139, "step": 144000 }, { "epoch": 0.39, "learning_rate": 6.059255703852385e-06, "loss": 0.1169, "step": 144500 }, { "epoch": 0.4, "learning_rate": 6.04561991044011e-06, "loss": 0.12, "step": 145000 }, { "epoch": 0.4, "learning_rate": 6.031984117027833e-06, "loss": 0.1216, "step": 145500 }, { "epoch": 0.4, "learning_rate": 6.018348323615558e-06, "loss": 0.1154, "step": 146000 }, { "epoch": 0.4, "learning_rate": 6.004712530203283e-06, "loss": 0.1193, "step": 146500 }, { "epoch": 0.4, "learning_rate": 5.991076736791007e-06, "loss": 0.1118, "step": 147000 }, { "epoch": 0.4, "learning_rate": 5.977440943378732e-06, "loss": 0.1165, "step": 147500 }, { "epoch": 0.4, "learning_rate": 5.9638051499664565e-06, "loss": 0.1113, "step": 148000 }, { "epoch": 0.4, "learning_rate": 5.950169356554181e-06, "loss": 0.114, "step": 148500 }, { "epoch": 0.41, "learning_rate": 5.936533563141905e-06, "loss": 0.1138, "step": 149000 }, { "epoch": 0.41, "learning_rate": 5.9228977697296294e-06, "loss": 0.111, "step": 149500 }, { "epoch": 0.41, "learning_rate": 5.909261976317354e-06, "loss": 0.1186, "step": 150000 }, { "epoch": 0.41, "learning_rate": 5.895626182905079e-06, "loss": 0.1148, "step": 150500 }, { "epoch": 0.41, "learning_rate": 5.881990389492803e-06, "loss": 0.112, "step": 151000 }, { "epoch": 0.41, "learning_rate": 5.868354596080528e-06, "loss": 0.1163, "step": 151500 }, { "epoch": 0.41, "learning_rate": 5.8547188026682525e-06, "loss": 0.1189, "step": 152000 }, { "epoch": 0.42, "learning_rate": 5.841083009255977e-06, "loss": 0.1099, "step": 152500 }, { "epoch": 0.42, "learning_rate": 5.827447215843701e-06, "loss": 0.1168, "step": 153000 }, { "epoch": 0.42, "learning_rate": 5.8138114224314255e-06, "loss": 0.1131, "step": 153500 }, { "epoch": 0.42, "learning_rate": 5.80017562901915e-06, "loss": 0.1177, "step": 154000 }, { "epoch": 0.42, "learning_rate": 5.786539835606875e-06, "loss": 0.1163, "step": 154500 }, { "epoch": 0.42, "learning_rate": 5.772904042194599e-06, "loss": 0.1166, "step": 155000 }, { "epoch": 0.42, "learning_rate": 5.759268248782324e-06, "loss": 0.1145, "step": 155500 }, { "epoch": 0.43, "learning_rate": 5.7456324553700486e-06, "loss": 0.1158, "step": 156000 }, { "epoch": 0.43, "learning_rate": 5.731996661957772e-06, "loss": 0.117, "step": 156500 }, { "epoch": 0.43, "learning_rate": 5.718360868545497e-06, "loss": 0.1211, "step": 157000 }, { "epoch": 0.43, "learning_rate": 5.7047250751332216e-06, "loss": 0.1183, "step": 157500 }, { "epoch": 0.43, "learning_rate": 5.691089281720946e-06, "loss": 0.1141, "step": 158000 }, { "epoch": 0.43, "learning_rate": 5.677453488308671e-06, "loss": 0.1152, "step": 158500 }, { "epoch": 0.43, "learning_rate": 5.663817694896395e-06, "loss": 0.1166, "step": 159000 }, { "epoch": 0.43, "learning_rate": 5.65018190148412e-06, "loss": 0.1132, "step": 159500 }, { "epoch": 0.44, "learning_rate": 5.636546108071845e-06, "loss": 0.1184, "step": 160000 }, { "epoch": 0.44, "learning_rate": 5.622910314659568e-06, "loss": 0.1114, "step": 160500 }, { "epoch": 0.44, "learning_rate": 5.609274521247293e-06, "loss": 0.1143, "step": 161000 }, { "epoch": 0.44, "learning_rate": 5.595638727835018e-06, "loss": 0.1147, "step": 161500 }, { "epoch": 0.44, "learning_rate": 5.582002934422743e-06, "loss": 0.1124, "step": 162000 }, { "epoch": 0.44, "learning_rate": 5.568367141010468e-06, "loss": 0.1136, "step": 162500 }, { "epoch": 0.44, "learning_rate": 5.554731347598192e-06, "loss": 0.1152, "step": 163000 }, { "epoch": 0.45, "learning_rate": 5.541095554185917e-06, "loss": 0.109, "step": 163500 }, { "epoch": 0.45, "learning_rate": 5.5274597607736415e-06, "loss": 0.1117, "step": 164000 }, { "epoch": 0.45, "learning_rate": 5.513823967361366e-06, "loss": 0.1184, "step": 164500 }, { "epoch": 0.45, "learning_rate": 5.500188173949091e-06, "loss": 0.1143, "step": 165000 }, { "epoch": 0.45, "learning_rate": 5.4865523805368145e-06, "loss": 0.1143, "step": 165500 }, { "epoch": 0.45, "learning_rate": 5.472916587124539e-06, "loss": 0.1131, "step": 166000 }, { "epoch": 0.45, "learning_rate": 5.459280793712264e-06, "loss": 0.1085, "step": 166500 }, { "epoch": 0.46, "learning_rate": 5.445645000299988e-06, "loss": 0.1152, "step": 167000 }, { "epoch": 0.46, "learning_rate": 5.432009206887713e-06, "loss": 0.1188, "step": 167500 }, { "epoch": 0.46, "learning_rate": 5.4183734134754376e-06, "loss": 0.1159, "step": 168000 }, { "epoch": 0.46, "learning_rate": 5.404737620063162e-06, "loss": 0.1105, "step": 168500 }, { "epoch": 0.46, "learning_rate": 5.391101826650886e-06, "loss": 0.1109, "step": 169000 }, { "epoch": 0.46, "learning_rate": 5.3774660332386106e-06, "loss": 0.1164, "step": 169500 }, { "epoch": 0.46, "learning_rate": 5.363830239826335e-06, "loss": 0.1108, "step": 170000 }, { "epoch": 0.46, "learning_rate": 5.35019444641406e-06, "loss": 0.1149, "step": 170500 }, { "epoch": 0.47, "learning_rate": 5.336558653001784e-06, "loss": 0.1142, "step": 171000 }, { "epoch": 0.47, "learning_rate": 5.322922859589509e-06, "loss": 0.1128, "step": 171500 }, { "epoch": 0.47, "learning_rate": 5.309287066177234e-06, "loss": 0.1137, "step": 172000 }, { "epoch": 0.47, "learning_rate": 5.295651272764958e-06, "loss": 0.115, "step": 172500 }, { "epoch": 0.47, "learning_rate": 5.282015479352682e-06, "loss": 0.1141, "step": 173000 }, { "epoch": 0.47, "learning_rate": 5.268379685940407e-06, "loss": 0.1144, "step": 173500 }, { "epoch": 0.47, "learning_rate": 5.254743892528131e-06, "loss": 0.1138, "step": 174000 }, { "epoch": 0.48, "learning_rate": 5.241108099115856e-06, "loss": 0.1137, "step": 174500 }, { "epoch": 0.48, "learning_rate": 5.2274723057035804e-06, "loss": 0.1167, "step": 175000 }, { "epoch": 0.48, "learning_rate": 5.213836512291305e-06, "loss": 0.1114, "step": 175500 }, { "epoch": 0.48, "learning_rate": 5.20020071887903e-06, "loss": 0.1128, "step": 176000 }, { "epoch": 0.48, "learning_rate": 5.1865649254667534e-06, "loss": 0.115, "step": 176500 }, { "epoch": 0.48, "learning_rate": 5.172929132054478e-06, "loss": 0.1059, "step": 177000 }, { "epoch": 0.48, "learning_rate": 5.159293338642203e-06, "loss": 0.1149, "step": 177500 }, { "epoch": 0.49, "learning_rate": 5.145657545229927e-06, "loss": 0.1147, "step": 178000 }, { "epoch": 0.49, "learning_rate": 5.132021751817652e-06, "loss": 0.112, "step": 178500 }, { "epoch": 0.49, "learning_rate": 5.1183859584053765e-06, "loss": 0.1142, "step": 179000 }, { "epoch": 0.49, "learning_rate": 5.104750164993101e-06, "loss": 0.1118, "step": 179500 }, { "epoch": 0.49, "learning_rate": 5.091114371580826e-06, "loss": 0.1113, "step": 180000 }, { "epoch": 0.49, "learning_rate": 5.0774785781685495e-06, "loss": 0.1082, "step": 180500 }, { "epoch": 0.49, "learning_rate": 5.063842784756274e-06, "loss": 0.1143, "step": 181000 }, { "epoch": 0.49, "learning_rate": 5.050206991343999e-06, "loss": 0.1098, "step": 181500 }, { "epoch": 0.5, "learning_rate": 5.036571197931723e-06, "loss": 0.1074, "step": 182000 }, { "epoch": 0.5, "learning_rate": 5.022935404519448e-06, "loss": 0.1127, "step": 182500 }, { "epoch": 0.5, "learning_rate": 5.0092996111071726e-06, "loss": 0.1113, "step": 183000 }, { "epoch": 0.5, "learning_rate": 4.995663817694897e-06, "loss": 0.1175, "step": 183500 }, { "epoch": 0.5, "learning_rate": 4.982028024282621e-06, "loss": 0.1131, "step": 184000 }, { "epoch": 0.5, "learning_rate": 4.9683922308703455e-06, "loss": 0.1084, "step": 184500 }, { "epoch": 0.5, "learning_rate": 4.95475643745807e-06, "loss": 0.1159, "step": 185000 }, { "epoch": 0.51, "learning_rate": 4.941120644045795e-06, "loss": 0.1103, "step": 185500 }, { "epoch": 0.51, "learning_rate": 4.927484850633519e-06, "loss": 0.1133, "step": 186000 }, { "epoch": 0.51, "learning_rate": 4.913849057221244e-06, "loss": 0.1102, "step": 186500 }, { "epoch": 0.51, "learning_rate": 4.900213263808969e-06, "loss": 0.1139, "step": 187000 }, { "epoch": 0.51, "learning_rate": 4.886577470396693e-06, "loss": 0.1171, "step": 187500 }, { "epoch": 0.51, "learning_rate": 4.872941676984417e-06, "loss": 0.117, "step": 188000 }, { "epoch": 0.51, "learning_rate": 4.859305883572142e-06, "loss": 0.115, "step": 188500 }, { "epoch": 0.52, "learning_rate": 4.845670090159866e-06, "loss": 0.1086, "step": 189000 }, { "epoch": 0.52, "learning_rate": 4.832034296747591e-06, "loss": 0.1114, "step": 189500 }, { "epoch": 0.52, "learning_rate": 4.8183985033353154e-06, "loss": 0.1145, "step": 190000 }, { "epoch": 0.52, "learning_rate": 4.80476270992304e-06, "loss": 0.115, "step": 190500 }, { "epoch": 0.52, "learning_rate": 4.791126916510765e-06, "loss": 0.1138, "step": 191000 }, { "epoch": 0.52, "learning_rate": 4.777491123098488e-06, "loss": 0.1103, "step": 191500 }, { "epoch": 0.52, "learning_rate": 4.763855329686213e-06, "loss": 0.1144, "step": 192000 }, { "epoch": 0.52, "learning_rate": 4.750219536273938e-06, "loss": 0.1123, "step": 192500 }, { "epoch": 0.53, "learning_rate": 4.736583742861662e-06, "loss": 0.1115, "step": 193000 }, { "epoch": 0.53, "learning_rate": 4.722947949449387e-06, "loss": 0.1087, "step": 193500 }, { "epoch": 0.53, "learning_rate": 4.7093121560371115e-06, "loss": 0.1153, "step": 194000 }, { "epoch": 0.53, "learning_rate": 4.695676362624836e-06, "loss": 0.1139, "step": 194500 }, { "epoch": 0.53, "learning_rate": 4.682040569212561e-06, "loss": 0.112, "step": 195000 }, { "epoch": 0.53, "learning_rate": 4.6684047758002845e-06, "loss": 0.1087, "step": 195500 }, { "epoch": 0.53, "learning_rate": 4.654768982388009e-06, "loss": 0.1183, "step": 196000 }, { "epoch": 0.54, "learning_rate": 4.6411331889757345e-06, "loss": 0.1069, "step": 196500 }, { "epoch": 0.54, "learning_rate": 4.627497395563459e-06, "loss": 0.1124, "step": 197000 }, { "epoch": 0.54, "learning_rate": 4.613861602151184e-06, "loss": 0.1111, "step": 197500 }, { "epoch": 0.54, "learning_rate": 4.6002258087389075e-06, "loss": 0.1097, "step": 198000 }, { "epoch": 0.54, "learning_rate": 4.586590015326632e-06, "loss": 0.1089, "step": 198500 }, { "epoch": 0.54, "learning_rate": 4.572954221914357e-06, "loss": 0.1123, "step": 199000 }, { "epoch": 0.54, "learning_rate": 4.559318428502081e-06, "loss": 0.1127, "step": 199500 }, { "epoch": 0.55, "learning_rate": 4.545682635089806e-06, "loss": 0.1091, "step": 200000 }, { "epoch": 0.55, "learning_rate": 4.532046841677531e-06, "loss": 0.1092, "step": 200500 }, { "epoch": 0.55, "learning_rate": 4.518411048265255e-06, "loss": 0.1113, "step": 201000 }, { "epoch": 0.55, "learning_rate": 4.504775254852979e-06, "loss": 0.1128, "step": 201500 }, { "epoch": 0.55, "learning_rate": 4.491139461440704e-06, "loss": 0.112, "step": 202000 }, { "epoch": 0.55, "learning_rate": 4.477503668028428e-06, "loss": 0.1135, "step": 202500 }, { "epoch": 0.55, "learning_rate": 4.463867874616153e-06, "loss": 0.1082, "step": 203000 }, { "epoch": 0.55, "learning_rate": 4.4502320812038774e-06, "loss": 0.1102, "step": 203500 }, { "epoch": 0.56, "learning_rate": 4.436596287791602e-06, "loss": 0.1125, "step": 204000 }, { "epoch": 0.56, "learning_rate": 4.422960494379327e-06, "loss": 0.1096, "step": 204500 }, { "epoch": 0.56, "learning_rate": 4.409324700967051e-06, "loss": 0.1146, "step": 205000 }, { "epoch": 0.56, "learning_rate": 4.395688907554775e-06, "loss": 0.1128, "step": 205500 }, { "epoch": 0.56, "learning_rate": 4.3820531141425e-06, "loss": 0.1134, "step": 206000 }, { "epoch": 0.56, "learning_rate": 4.368417320730224e-06, "loss": 0.1116, "step": 206500 }, { "epoch": 0.56, "learning_rate": 4.354781527317949e-06, "loss": 0.1095, "step": 207000 }, { "epoch": 0.57, "learning_rate": 4.3411457339056735e-06, "loss": 0.1097, "step": 207500 }, { "epoch": 0.57, "learning_rate": 4.327509940493398e-06, "loss": 0.1159, "step": 208000 }, { "epoch": 0.57, "learning_rate": 4.313874147081123e-06, "loss": 0.1123, "step": 208500 }, { "epoch": 0.57, "learning_rate": 4.300238353668847e-06, "loss": 0.1157, "step": 209000 }, { "epoch": 0.57, "learning_rate": 4.286602560256571e-06, "loss": 0.1122, "step": 209500 }, { "epoch": 0.57, "learning_rate": 4.272966766844296e-06, "loss": 0.1108, "step": 210000 }, { "epoch": 0.57, "learning_rate": 4.25933097343202e-06, "loss": 0.1121, "step": 210500 }, { "epoch": 0.58, "learning_rate": 4.245695180019745e-06, "loss": 0.1156, "step": 211000 }, { "epoch": 0.58, "learning_rate": 4.2320593866074695e-06, "loss": 0.1123, "step": 211500 }, { "epoch": 0.58, "learning_rate": 4.218423593195194e-06, "loss": 0.1094, "step": 212000 }, { "epoch": 0.58, "learning_rate": 4.204787799782919e-06, "loss": 0.1099, "step": 212500 }, { "epoch": 0.58, "learning_rate": 4.1911520063706425e-06, "loss": 0.1056, "step": 213000 }, { "epoch": 0.58, "learning_rate": 4.177516212958367e-06, "loss": 0.1084, "step": 213500 }, { "epoch": 0.58, "learning_rate": 4.163880419546092e-06, "loss": 0.1147, "step": 214000 }, { "epoch": 0.58, "learning_rate": 4.150244626133816e-06, "loss": 0.1058, "step": 214500 }, { "epoch": 0.59, "learning_rate": 4.136608832721541e-06, "loss": 0.111, "step": 215000 }, { "epoch": 0.59, "learning_rate": 4.122973039309266e-06, "loss": 0.1123, "step": 215500 }, { "epoch": 0.59, "learning_rate": 4.10933724589699e-06, "loss": 0.1097, "step": 216000 }, { "epoch": 0.59, "learning_rate": 4.095701452484715e-06, "loss": 0.1114, "step": 216500 }, { "epoch": 0.59, "learning_rate": 4.0820656590724386e-06, "loss": 0.1099, "step": 217000 }, { "epoch": 0.59, "learning_rate": 4.068429865660163e-06, "loss": 0.106, "step": 217500 }, { "epoch": 0.59, "learning_rate": 4.054794072247888e-06, "loss": 0.1078, "step": 218000 }, { "epoch": 0.6, "learning_rate": 4.041158278835612e-06, "loss": 0.1118, "step": 218500 }, { "epoch": 0.6, "learning_rate": 4.027522485423337e-06, "loss": 0.1129, "step": 219000 }, { "epoch": 0.6, "learning_rate": 4.013886692011062e-06, "loss": 0.1096, "step": 219500 }, { "epoch": 0.6, "learning_rate": 4.000250898598786e-06, "loss": 0.111, "step": 220000 }, { "epoch": 0.6, "learning_rate": 3.986615105186511e-06, "loss": 0.1097, "step": 220500 }, { "epoch": 0.6, "learning_rate": 3.9729793117742355e-06, "loss": 0.1106, "step": 221000 }, { "epoch": 0.6, "learning_rate": 3.95934351836196e-06, "loss": 0.1125, "step": 221500 }, { "epoch": 0.61, "learning_rate": 3.945707724949685e-06, "loss": 0.1155, "step": 222000 }, { "epoch": 0.61, "learning_rate": 3.932071931537409e-06, "loss": 0.1104, "step": 222500 }, { "epoch": 0.61, "learning_rate": 3.918436138125133e-06, "loss": 0.1115, "step": 223000 }, { "epoch": 0.61, "learning_rate": 3.904800344712858e-06, "loss": 0.1102, "step": 223500 }, { "epoch": 0.61, "learning_rate": 3.891164551300582e-06, "loss": 0.1087, "step": 224000 }, { "epoch": 0.61, "learning_rate": 3.877528757888307e-06, "loss": 0.1115, "step": 224500 }, { "epoch": 0.61, "learning_rate": 3.8638929644760315e-06, "loss": 0.1156, "step": 225000 }, { "epoch": 0.61, "learning_rate": 3.850257171063756e-06, "loss": 0.1073, "step": 225500 }, { "epoch": 0.62, "learning_rate": 3.836621377651481e-06, "loss": 0.1059, "step": 226000 }, { "epoch": 0.62, "learning_rate": 3.822985584239205e-06, "loss": 0.112, "step": 226500 }, { "epoch": 0.62, "learning_rate": 3.8093497908269296e-06, "loss": 0.1076, "step": 227000 }, { "epoch": 0.62, "learning_rate": 3.795713997414654e-06, "loss": 0.11, "step": 227500 }, { "epoch": 0.62, "learning_rate": 3.7820782040023784e-06, "loss": 0.109, "step": 228000 }, { "epoch": 0.62, "learning_rate": 3.768442410590103e-06, "loss": 0.1112, "step": 228500 }, { "epoch": 0.62, "learning_rate": 3.7548066171778276e-06, "loss": 0.1096, "step": 229000 }, { "epoch": 0.63, "learning_rate": 3.7411708237655518e-06, "loss": 0.109, "step": 229500 }, { "epoch": 0.63, "learning_rate": 3.7275350303532764e-06, "loss": 0.112, "step": 230000 }, { "epoch": 0.63, "learning_rate": 3.713899236941001e-06, "loss": 0.115, "step": 230500 }, { "epoch": 0.63, "learning_rate": 3.7002634435287256e-06, "loss": 0.1085, "step": 231000 }, { "epoch": 0.63, "learning_rate": 3.68662765011645e-06, "loss": 0.1086, "step": 231500 }, { "epoch": 0.63, "learning_rate": 3.6729918567041744e-06, "loss": 0.1095, "step": 232000 }, { "epoch": 0.63, "learning_rate": 3.659356063291899e-06, "loss": 0.1066, "step": 232500 }, { "epoch": 0.64, "learning_rate": 3.6457202698796236e-06, "loss": 0.11, "step": 233000 }, { "epoch": 0.64, "learning_rate": 3.632084476467348e-06, "loss": 0.1061, "step": 233500 }, { "epoch": 0.64, "learning_rate": 3.6184486830550724e-06, "loss": 0.1117, "step": 234000 }, { "epoch": 0.64, "learning_rate": 3.604812889642797e-06, "loss": 0.1088, "step": 234500 }, { "epoch": 0.64, "learning_rate": 3.5911770962305217e-06, "loss": 0.1095, "step": 235000 }, { "epoch": 0.64, "learning_rate": 3.577541302818246e-06, "loss": 0.1101, "step": 235500 }, { "epoch": 0.64, "learning_rate": 3.5639055094059705e-06, "loss": 0.1105, "step": 236000 }, { "epoch": 0.64, "learning_rate": 3.550269715993695e-06, "loss": 0.1101, "step": 236500 }, { "epoch": 0.65, "learning_rate": 3.5366339225814193e-06, "loss": 0.1089, "step": 237000 }, { "epoch": 0.65, "learning_rate": 3.522998129169144e-06, "loss": 0.1059, "step": 237500 }, { "epoch": 0.65, "learning_rate": 3.5093623357568685e-06, "loss": 0.1097, "step": 238000 }, { "epoch": 0.65, "learning_rate": 3.495726542344593e-06, "loss": 0.106, "step": 238500 }, { "epoch": 0.65, "learning_rate": 3.4820907489323173e-06, "loss": 0.1073, "step": 239000 }, { "epoch": 0.65, "learning_rate": 3.468454955520042e-06, "loss": 0.1076, "step": 239500 }, { "epoch": 0.65, "learning_rate": 3.4548191621077665e-06, "loss": 0.1095, "step": 240000 }, { "epoch": 0.66, "learning_rate": 3.441183368695491e-06, "loss": 0.1095, "step": 240500 }, { "epoch": 0.66, "learning_rate": 3.4275475752832153e-06, "loss": 0.109, "step": 241000 }, { "epoch": 0.66, "learning_rate": 3.41391178187094e-06, "loss": 0.1085, "step": 241500 }, { "epoch": 0.66, "learning_rate": 3.4002759884586645e-06, "loss": 0.1118, "step": 242000 }, { "epoch": 0.66, "learning_rate": 3.3866401950463896e-06, "loss": 0.1109, "step": 242500 }, { "epoch": 0.66, "learning_rate": 3.373004401634114e-06, "loss": 0.1059, "step": 243000 }, { "epoch": 0.66, "learning_rate": 3.3593686082218384e-06, "loss": 0.1047, "step": 243500 }, { "epoch": 0.67, "learning_rate": 3.345732814809563e-06, "loss": 0.1117, "step": 244000 }, { "epoch": 0.67, "learning_rate": 3.3320970213972876e-06, "loss": 0.1095, "step": 244500 }, { "epoch": 0.67, "learning_rate": 3.3184612279850122e-06, "loss": 0.1145, "step": 245000 }, { "epoch": 0.67, "learning_rate": 3.3048254345727364e-06, "loss": 0.1086, "step": 245500 }, { "epoch": 0.67, "learning_rate": 3.291189641160461e-06, "loss": 0.1085, "step": 246000 }, { "epoch": 0.67, "learning_rate": 3.2775538477481856e-06, "loss": 0.1085, "step": 246500 }, { "epoch": 0.67, "learning_rate": 3.26391805433591e-06, "loss": 0.1146, "step": 247000 }, { "epoch": 0.67, "learning_rate": 3.2502822609236344e-06, "loss": 0.1133, "step": 247500 }, { "epoch": 0.68, "learning_rate": 3.236646467511359e-06, "loss": 0.1073, "step": 248000 }, { "epoch": 0.68, "learning_rate": 3.2230106740990837e-06, "loss": 0.1117, "step": 248500 }, { "epoch": 0.68, "learning_rate": 3.209374880686808e-06, "loss": 0.1107, "step": 249000 }, { "epoch": 0.68, "learning_rate": 3.1957390872745325e-06, "loss": 0.1089, "step": 249500 }, { "epoch": 0.68, "learning_rate": 3.182103293862257e-06, "loss": 0.1049, "step": 250000 }, { "epoch": 0.68, "learning_rate": 3.1684675004499817e-06, "loss": 0.1106, "step": 250500 }, { "epoch": 0.68, "learning_rate": 3.154831707037706e-06, "loss": 0.1123, "step": 251000 }, { "epoch": 0.69, "learning_rate": 3.1411959136254305e-06, "loss": 0.1074, "step": 251500 }, { "epoch": 0.69, "learning_rate": 3.127560120213155e-06, "loss": 0.1069, "step": 252000 }, { "epoch": 0.69, "learning_rate": 3.1139243268008797e-06, "loss": 0.1085, "step": 252500 }, { "epoch": 0.69, "learning_rate": 3.100288533388604e-06, "loss": 0.1049, "step": 253000 }, { "epoch": 0.69, "learning_rate": 3.0866527399763285e-06, "loss": 0.1105, "step": 253500 }, { "epoch": 0.69, "learning_rate": 3.073016946564053e-06, "loss": 0.1105, "step": 254000 }, { "epoch": 0.69, "learning_rate": 3.0593811531517777e-06, "loss": 0.1052, "step": 254500 }, { "epoch": 0.7, "learning_rate": 3.045745359739502e-06, "loss": 0.106, "step": 255000 }, { "epoch": 0.7, "learning_rate": 3.0321095663272265e-06, "loss": 0.106, "step": 255500 }, { "epoch": 0.7, "learning_rate": 3.018473772914951e-06, "loss": 0.1086, "step": 256000 }, { "epoch": 0.7, "learning_rate": 3.0048379795026753e-06, "loss": 0.1086, "step": 256500 }, { "epoch": 0.7, "learning_rate": 2.9912021860904e-06, "loss": 0.1058, "step": 257000 }, { "epoch": 0.7, "learning_rate": 2.9775663926781246e-06, "loss": 0.1068, "step": 257500 }, { "epoch": 0.7, "learning_rate": 2.963930599265849e-06, "loss": 0.1048, "step": 258000 }, { "epoch": 0.7, "learning_rate": 2.9502948058535734e-06, "loss": 0.1066, "step": 258500 }, { "epoch": 0.71, "learning_rate": 2.936659012441298e-06, "loss": 0.106, "step": 259000 }, { "epoch": 0.71, "learning_rate": 2.9230232190290226e-06, "loss": 0.1055, "step": 259500 }, { "epoch": 0.71, "learning_rate": 2.909387425616747e-06, "loss": 0.1089, "step": 260000 }, { "epoch": 0.71, "learning_rate": 2.8957516322044714e-06, "loss": 0.1055, "step": 260500 }, { "epoch": 0.71, "learning_rate": 2.882115838792196e-06, "loss": 0.1031, "step": 261000 }, { "epoch": 0.71, "learning_rate": 2.8684800453799206e-06, "loss": 0.107, "step": 261500 }, { "epoch": 0.71, "learning_rate": 2.8548442519676452e-06, "loss": 0.1067, "step": 262000 }, { "epoch": 0.72, "learning_rate": 2.8412084585553694e-06, "loss": 0.1101, "step": 262500 }, { "epoch": 0.72, "learning_rate": 2.827572665143094e-06, "loss": 0.1071, "step": 263000 }, { "epoch": 0.72, "learning_rate": 2.8139368717308186e-06, "loss": 0.1102, "step": 263500 }, { "epoch": 0.72, "learning_rate": 2.800301078318543e-06, "loss": 0.1058, "step": 264000 }, { "epoch": 0.72, "learning_rate": 2.7866652849062674e-06, "loss": 0.1094, "step": 264500 }, { "epoch": 0.72, "learning_rate": 2.773029491493992e-06, "loss": 0.1056, "step": 265000 }, { "epoch": 0.72, "learning_rate": 2.7593936980817167e-06, "loss": 0.1058, "step": 265500 }, { "epoch": 0.73, "learning_rate": 2.7457579046694417e-06, "loss": 0.1093, "step": 266000 }, { "epoch": 0.73, "learning_rate": 2.732122111257166e-06, "loss": 0.1066, "step": 266500 }, { "epoch": 0.73, "learning_rate": 2.7184863178448905e-06, "loss": 0.1076, "step": 267000 }, { "epoch": 0.73, "learning_rate": 2.704850524432615e-06, "loss": 0.1103, "step": 267500 }, { "epoch": 0.73, "learning_rate": 2.6912147310203397e-06, "loss": 0.1093, "step": 268000 }, { "epoch": 0.73, "learning_rate": 2.677578937608064e-06, "loss": 0.1031, "step": 268500 }, { "epoch": 0.73, "learning_rate": 2.6639431441957885e-06, "loss": 0.1072, "step": 269000 }, { "epoch": 0.73, "learning_rate": 2.650307350783513e-06, "loss": 0.1097, "step": 269500 }, { "epoch": 0.74, "learning_rate": 2.6366715573712378e-06, "loss": 0.1088, "step": 270000 }, { "epoch": 0.74, "learning_rate": 2.623035763958962e-06, "loss": 0.1101, "step": 270500 }, { "epoch": 0.74, "learning_rate": 2.6093999705466866e-06, "loss": 0.1027, "step": 271000 }, { "epoch": 0.74, "learning_rate": 2.595764177134411e-06, "loss": 0.1106, "step": 271500 }, { "epoch": 0.74, "learning_rate": 2.5821283837221358e-06, "loss": 0.1048, "step": 272000 }, { "epoch": 0.74, "learning_rate": 2.56849259030986e-06, "loss": 0.1082, "step": 272500 }, { "epoch": 0.74, "learning_rate": 2.5548567968975846e-06, "loss": 0.1051, "step": 273000 }, { "epoch": 0.75, "learning_rate": 2.541221003485309e-06, "loss": 0.1113, "step": 273500 }, { "epoch": 0.75, "learning_rate": 2.5275852100730334e-06, "loss": 0.1066, "step": 274000 }, { "epoch": 0.75, "learning_rate": 2.513949416660758e-06, "loss": 0.1061, "step": 274500 }, { "epoch": 0.75, "learning_rate": 2.5003136232484826e-06, "loss": 0.1109, "step": 275000 }, { "epoch": 0.75, "learning_rate": 2.4866778298362072e-06, "loss": 0.1052, "step": 275500 }, { "epoch": 0.75, "learning_rate": 2.4730420364239314e-06, "loss": 0.1053, "step": 276000 }, { "epoch": 0.75, "learning_rate": 2.459406243011656e-06, "loss": 0.1062, "step": 276500 }, { "epoch": 0.76, "learning_rate": 2.4457704495993806e-06, "loss": 0.1033, "step": 277000 }, { "epoch": 0.76, "learning_rate": 2.4321346561871052e-06, "loss": 0.1049, "step": 277500 }, { "epoch": 0.76, "learning_rate": 2.4184988627748294e-06, "loss": 0.1097, "step": 278000 }, { "epoch": 0.76, "learning_rate": 2.404863069362554e-06, "loss": 0.1078, "step": 278500 }, { "epoch": 0.76, "learning_rate": 2.3912272759502787e-06, "loss": 0.1084, "step": 279000 }, { "epoch": 0.76, "learning_rate": 2.3775914825380033e-06, "loss": 0.1101, "step": 279500 }, { "epoch": 0.76, "learning_rate": 2.3639556891257275e-06, "loss": 0.1118, "step": 280000 }, { "epoch": 0.76, "learning_rate": 2.350319895713452e-06, "loss": 0.1077, "step": 280500 }, { "epoch": 0.77, "learning_rate": 2.3366841023011767e-06, "loss": 0.1079, "step": 281000 }, { "epoch": 0.77, "learning_rate": 2.3230483088889013e-06, "loss": 0.1064, "step": 281500 }, { "epoch": 0.77, "learning_rate": 2.3094125154766255e-06, "loss": 0.1064, "step": 282000 }, { "epoch": 0.77, "learning_rate": 2.29577672206435e-06, "loss": 0.1103, "step": 282500 }, { "epoch": 0.77, "learning_rate": 2.2821409286520747e-06, "loss": 0.1113, "step": 283000 }, { "epoch": 0.77, "learning_rate": 2.2685051352397993e-06, "loss": 0.1069, "step": 283500 }, { "epoch": 0.77, "learning_rate": 2.254869341827524e-06, "loss": 0.1084, "step": 284000 }, { "epoch": 0.78, "learning_rate": 2.2412335484152486e-06, "loss": 0.1067, "step": 284500 }, { "epoch": 0.78, "learning_rate": 2.2275977550029727e-06, "loss": 0.1068, "step": 285000 }, { "epoch": 0.78, "learning_rate": 2.2139619615906974e-06, "loss": 0.1063, "step": 285500 }, { "epoch": 0.78, "learning_rate": 2.200326168178422e-06, "loss": 0.1118, "step": 286000 }, { "epoch": 0.78, "learning_rate": 2.1866903747661466e-06, "loss": 0.1043, "step": 286500 }, { "epoch": 0.78, "learning_rate": 2.1730545813538708e-06, "loss": 0.1102, "step": 287000 }, { "epoch": 0.78, "learning_rate": 2.1594187879415954e-06, "loss": 0.1097, "step": 287500 }, { "epoch": 0.79, "learning_rate": 2.14578299452932e-06, "loss": 0.1103, "step": 288000 }, { "epoch": 0.79, "learning_rate": 2.132147201117044e-06, "loss": 0.1054, "step": 288500 }, { "epoch": 0.79, "learning_rate": 2.118511407704769e-06, "loss": 0.1078, "step": 289000 }, { "epoch": 0.79, "learning_rate": 2.1048756142924934e-06, "loss": 0.1072, "step": 289500 }, { "epoch": 0.79, "learning_rate": 2.091239820880218e-06, "loss": 0.108, "step": 290000 }, { "epoch": 0.79, "learning_rate": 2.077604027467942e-06, "loss": 0.1056, "step": 290500 }, { "epoch": 0.79, "learning_rate": 2.063968234055667e-06, "loss": 0.1079, "step": 291000 }, { "epoch": 0.79, "learning_rate": 2.0503324406433914e-06, "loss": 0.1059, "step": 291500 }, { "epoch": 0.8, "learning_rate": 2.036696647231116e-06, "loss": 0.1106, "step": 292000 }, { "epoch": 0.8, "learning_rate": 2.0230608538188402e-06, "loss": 0.1062, "step": 292500 }, { "epoch": 0.8, "learning_rate": 2.009425060406565e-06, "loss": 0.1068, "step": 293000 }, { "epoch": 0.8, "learning_rate": 1.9957892669942895e-06, "loss": 0.1076, "step": 293500 }, { "epoch": 0.8, "learning_rate": 1.982153473582014e-06, "loss": 0.1034, "step": 294000 }, { "epoch": 0.8, "learning_rate": 1.9685176801697387e-06, "loss": 0.1083, "step": 294500 }, { "epoch": 0.8, "learning_rate": 1.9548818867574633e-06, "loss": 0.1071, "step": 295000 }, { "epoch": 0.81, "learning_rate": 1.9412460933451875e-06, "loss": 0.1048, "step": 295500 }, { "epoch": 0.81, "learning_rate": 1.927610299932912e-06, "loss": 0.1096, "step": 296000 }, { "epoch": 0.81, "learning_rate": 1.9139745065206367e-06, "loss": 0.1077, "step": 296500 }, { "epoch": 0.81, "learning_rate": 1.9003387131083611e-06, "loss": 0.108, "step": 297000 }, { "epoch": 0.81, "learning_rate": 1.8867029196960857e-06, "loss": 0.1052, "step": 297500 }, { "epoch": 0.81, "learning_rate": 1.8730671262838101e-06, "loss": 0.1065, "step": 298000 }, { "epoch": 0.81, "learning_rate": 1.8594313328715347e-06, "loss": 0.1068, "step": 298500 }, { "epoch": 0.82, "learning_rate": 1.8457955394592591e-06, "loss": 0.1091, "step": 299000 }, { "epoch": 0.82, "learning_rate": 1.8321597460469837e-06, "loss": 0.108, "step": 299500 }, { "epoch": 0.82, "learning_rate": 1.8185239526347081e-06, "loss": 0.1081, "step": 300000 }, { "epoch": 0.82, "learning_rate": 1.8048881592224328e-06, "loss": 0.1066, "step": 300500 }, { "epoch": 0.82, "learning_rate": 1.7912523658101572e-06, "loss": 0.1022, "step": 301000 }, { "epoch": 0.82, "learning_rate": 1.7776165723978816e-06, "loss": 0.1037, "step": 301500 }, { "epoch": 0.82, "learning_rate": 1.7639807789856062e-06, "loss": 0.1056, "step": 302000 }, { "epoch": 0.82, "learning_rate": 1.7503449855733306e-06, "loss": 0.1038, "step": 302500 }, { "epoch": 0.83, "learning_rate": 1.7367091921610552e-06, "loss": 0.1108, "step": 303000 }, { "epoch": 0.83, "learning_rate": 1.7230733987487796e-06, "loss": 0.106, "step": 303500 }, { "epoch": 0.83, "learning_rate": 1.7094376053365042e-06, "loss": 0.1068, "step": 304000 }, { "epoch": 0.83, "learning_rate": 1.6958018119242286e-06, "loss": 0.107, "step": 304500 }, { "epoch": 0.83, "learning_rate": 1.6821660185119532e-06, "loss": 0.1067, "step": 305000 }, { "epoch": 0.83, "learning_rate": 1.6685302250996776e-06, "loss": 0.1095, "step": 305500 }, { "epoch": 0.83, "learning_rate": 1.6548944316874024e-06, "loss": 0.1052, "step": 306000 }, { "epoch": 0.84, "learning_rate": 1.6412586382751268e-06, "loss": 0.1054, "step": 306500 }, { "epoch": 0.84, "learning_rate": 1.6276228448628515e-06, "loss": 0.1038, "step": 307000 }, { "epoch": 0.84, "learning_rate": 1.6139870514505759e-06, "loss": 0.1066, "step": 307500 }, { "epoch": 0.84, "learning_rate": 1.6003512580383005e-06, "loss": 0.1035, "step": 308000 }, { "epoch": 0.84, "learning_rate": 1.5867154646260249e-06, "loss": 0.1067, "step": 308500 }, { "epoch": 0.84, "learning_rate": 1.5730796712137495e-06, "loss": 0.102, "step": 309000 }, { "epoch": 0.84, "learning_rate": 1.5594438778014739e-06, "loss": 0.1068, "step": 309500 }, { "epoch": 0.85, "learning_rate": 1.5458080843891985e-06, "loss": 0.1078, "step": 310000 }, { "epoch": 0.85, "learning_rate": 1.5321722909769229e-06, "loss": 0.1074, "step": 310500 }, { "epoch": 0.85, "learning_rate": 1.5185364975646475e-06, "loss": 0.1078, "step": 311000 }, { "epoch": 0.85, "learning_rate": 1.504900704152372e-06, "loss": 0.1056, "step": 311500 }, { "epoch": 0.85, "learning_rate": 1.4912649107400965e-06, "loss": 0.1061, "step": 312000 }, { "epoch": 0.85, "learning_rate": 1.477629117327821e-06, "loss": 0.1044, "step": 312500 }, { "epoch": 0.85, "learning_rate": 1.4639933239155455e-06, "loss": 0.1108, "step": 313000 }, { "epoch": 0.85, "learning_rate": 1.45035753050327e-06, "loss": 0.1104, "step": 313500 }, { "epoch": 0.86, "learning_rate": 1.4367217370909945e-06, "loss": 0.1073, "step": 314000 }, { "epoch": 0.86, "learning_rate": 1.423085943678719e-06, "loss": 0.1039, "step": 314500 }, { "epoch": 0.86, "learning_rate": 1.4094501502664433e-06, "loss": 0.1022, "step": 315000 }, { "epoch": 0.86, "learning_rate": 1.395814356854168e-06, "loss": 0.1046, "step": 315500 }, { "epoch": 0.86, "learning_rate": 1.3821785634418924e-06, "loss": 0.104, "step": 316000 }, { "epoch": 0.86, "learning_rate": 1.368542770029617e-06, "loss": 0.1065, "step": 316500 }, { "epoch": 0.86, "learning_rate": 1.3549069766173414e-06, "loss": 0.1081, "step": 317000 }, { "epoch": 0.87, "learning_rate": 1.3412711832050662e-06, "loss": 0.1071, "step": 317500 }, { "epoch": 0.87, "learning_rate": 1.3276353897927908e-06, "loss": 0.1056, "step": 318000 }, { "epoch": 0.87, "learning_rate": 1.3139995963805152e-06, "loss": 0.1068, "step": 318500 }, { "epoch": 0.87, "learning_rate": 1.3003638029682398e-06, "loss": 0.1036, "step": 319000 }, { "epoch": 0.87, "learning_rate": 1.2867280095559642e-06, "loss": 0.1077, "step": 319500 }, { "epoch": 0.87, "learning_rate": 1.2730922161436886e-06, "loss": 0.1064, "step": 320000 }, { "epoch": 0.87, "learning_rate": 1.2594564227314132e-06, "loss": 0.1092, "step": 320500 }, { "epoch": 0.88, "learning_rate": 1.2458206293191376e-06, "loss": 0.1073, "step": 321000 }, { "epoch": 0.88, "learning_rate": 1.2321848359068622e-06, "loss": 0.1086, "step": 321500 }, { "epoch": 0.88, "learning_rate": 1.2185490424945866e-06, "loss": 0.1039, "step": 322000 }, { "epoch": 0.88, "learning_rate": 1.2049132490823113e-06, "loss": 0.1025, "step": 322500 }, { "epoch": 0.88, "learning_rate": 1.1912774556700357e-06, "loss": 0.104, "step": 323000 }, { "epoch": 0.88, "learning_rate": 1.1776416622577603e-06, "loss": 0.1067, "step": 323500 }, { "epoch": 0.88, "learning_rate": 1.1640058688454847e-06, "loss": 0.1095, "step": 324000 }, { "epoch": 0.88, "learning_rate": 1.1503700754332093e-06, "loss": 0.1045, "step": 324500 }, { "epoch": 0.89, "learning_rate": 1.1367342820209337e-06, "loss": 0.1096, "step": 325000 }, { "epoch": 0.89, "learning_rate": 1.1230984886086583e-06, "loss": 0.1059, "step": 325500 }, { "epoch": 0.89, "learning_rate": 1.1094626951963827e-06, "loss": 0.1111, "step": 326000 }, { "epoch": 0.89, "learning_rate": 1.0958269017841073e-06, "loss": 0.1079, "step": 326500 }, { "epoch": 0.89, "learning_rate": 1.082191108371832e-06, "loss": 0.104, "step": 327000 }, { "epoch": 0.89, "learning_rate": 1.0685553149595563e-06, "loss": 0.1039, "step": 327500 }, { "epoch": 0.89, "learning_rate": 1.054919521547281e-06, "loss": 0.1007, "step": 328000 }, { "epoch": 0.9, "learning_rate": 1.0412837281350053e-06, "loss": 0.1078, "step": 328500 }, { "epoch": 0.9, "learning_rate": 1.02764793472273e-06, "loss": 0.1044, "step": 329000 }, { "epoch": 0.9, "learning_rate": 1.0140121413104544e-06, "loss": 0.1051, "step": 329500 }, { "epoch": 0.9, "learning_rate": 1.000376347898179e-06, "loss": 0.1051, "step": 330000 }, { "epoch": 0.9, "learning_rate": 9.867405544859034e-07, "loss": 0.1074, "step": 330500 }, { "epoch": 0.9, "learning_rate": 9.731047610736278e-07, "loss": 0.1019, "step": 331000 }, { "epoch": 0.9, "learning_rate": 9.594689676613524e-07, "loss": 0.1039, "step": 331500 }, { "epoch": 0.91, "learning_rate": 9.45833174249077e-07, "loss": 0.1047, "step": 332000 }, { "epoch": 0.91, "learning_rate": 9.321973808368015e-07, "loss": 0.1059, "step": 332500 }, { "epoch": 0.91, "learning_rate": 9.18561587424526e-07, "loss": 0.1042, "step": 333000 }, { "epoch": 0.91, "learning_rate": 9.049257940122505e-07, "loss": 0.1081, "step": 333500 }, { "epoch": 0.91, "learning_rate": 8.91290000599975e-07, "loss": 0.101, "step": 334000 }, { "epoch": 0.91, "learning_rate": 8.776542071876995e-07, "loss": 0.1042, "step": 334500 }, { "epoch": 0.91, "learning_rate": 8.64018413775424e-07, "loss": 0.1096, "step": 335000 }, { "epoch": 0.91, "learning_rate": 8.503826203631485e-07, "loss": 0.1042, "step": 335500 }, { "epoch": 0.92, "learning_rate": 8.36746826950873e-07, "loss": 0.1027, "step": 336000 }, { "epoch": 0.92, "learning_rate": 8.231110335385976e-07, "loss": 0.1073, "step": 336500 }, { "epoch": 0.92, "learning_rate": 8.09475240126322e-07, "loss": 0.1096, "step": 337000 }, { "epoch": 0.92, "learning_rate": 7.958394467140465e-07, "loss": 0.099, "step": 337500 }, { "epoch": 0.92, "learning_rate": 7.822036533017712e-07, "loss": 0.1059, "step": 338000 }, { "epoch": 0.92, "learning_rate": 7.685678598894957e-07, "loss": 0.109, "step": 338500 }, { "epoch": 0.92, "learning_rate": 7.549320664772202e-07, "loss": 0.1081, "step": 339000 }, { "epoch": 0.93, "learning_rate": 7.412962730649447e-07, "loss": 0.1031, "step": 339500 }, { "epoch": 0.93, "learning_rate": 7.276604796526691e-07, "loss": 0.1074, "step": 340000 }, { "epoch": 0.93, "learning_rate": 7.140246862403936e-07, "loss": 0.1066, "step": 340500 }, { "epoch": 0.93, "learning_rate": 7.003888928281181e-07, "loss": 0.1077, "step": 341000 }, { "epoch": 0.93, "learning_rate": 6.867530994158426e-07, "loss": 0.102, "step": 341500 }, { "epoch": 0.93, "learning_rate": 6.731173060035671e-07, "loss": 0.1054, "step": 342000 }, { "epoch": 0.93, "learning_rate": 6.594815125912916e-07, "loss": 0.1051, "step": 342500 }, { "epoch": 0.94, "learning_rate": 6.458457191790161e-07, "loss": 0.1083, "step": 343000 }, { "epoch": 0.94, "learning_rate": 6.322099257667407e-07, "loss": 0.1045, "step": 343500 }, { "epoch": 0.94, "learning_rate": 6.185741323544653e-07, "loss": 0.1042, "step": 344000 }, { "epoch": 0.94, "learning_rate": 6.049383389421898e-07, "loss": 0.1038, "step": 344500 }, { "epoch": 0.94, "learning_rate": 5.913025455299143e-07, "loss": 0.104, "step": 345000 }, { "epoch": 0.94, "learning_rate": 5.776667521176388e-07, "loss": 0.1053, "step": 345500 }, { "epoch": 0.94, "learning_rate": 5.640309587053633e-07, "loss": 0.1046, "step": 346000 }, { "epoch": 0.94, "learning_rate": 5.503951652930878e-07, "loss": 0.1048, "step": 346500 }, { "epoch": 0.95, "learning_rate": 5.367593718808123e-07, "loss": 0.1079, "step": 347000 }, { "epoch": 0.95, "learning_rate": 5.231235784685368e-07, "loss": 0.1072, "step": 347500 }, { "epoch": 0.95, "learning_rate": 5.094877850562613e-07, "loss": 0.1055, "step": 348000 }, { "epoch": 0.95, "learning_rate": 4.958519916439858e-07, "loss": 0.1065, "step": 348500 }, { "epoch": 0.95, "learning_rate": 4.822161982317103e-07, "loss": 0.1031, "step": 349000 }, { "epoch": 0.95, "learning_rate": 4.685804048194349e-07, "loss": 0.1058, "step": 349500 }, { "epoch": 0.95, "learning_rate": 4.5494461140715933e-07, "loss": 0.1038, "step": 350000 }, { "epoch": 0.96, "learning_rate": 4.4130881799488384e-07, "loss": 0.1048, "step": 350500 }, { "epoch": 0.96, "learning_rate": 4.276730245826084e-07, "loss": 0.1035, "step": 351000 }, { "epoch": 0.96, "learning_rate": 4.140372311703329e-07, "loss": 0.1037, "step": 351500 }, { "epoch": 0.96, "learning_rate": 4.004014377580574e-07, "loss": 0.1057, "step": 352000 }, { "epoch": 0.96, "learning_rate": 3.867656443457819e-07, "loss": 0.1087, "step": 352500 }, { "epoch": 0.96, "learning_rate": 3.7312985093350643e-07, "loss": 0.1064, "step": 353000 }, { "epoch": 0.96, "learning_rate": 3.5949405752123093e-07, "loss": 0.1024, "step": 353500 }, { "epoch": 0.97, "learning_rate": 3.458582641089555e-07, "loss": 0.1014, "step": 354000 }, { "epoch": 0.97, "learning_rate": 3.3222247069668e-07, "loss": 0.1012, "step": 354500 }, { "epoch": 0.97, "learning_rate": 3.185866772844045e-07, "loss": 0.1032, "step": 355000 }, { "epoch": 0.97, "learning_rate": 3.04950883872129e-07, "loss": 0.1055, "step": 355500 }, { "epoch": 0.97, "learning_rate": 2.913150904598535e-07, "loss": 0.105, "step": 356000 }, { "epoch": 0.97, "learning_rate": 2.77679297047578e-07, "loss": 0.1033, "step": 356500 }, { "epoch": 0.97, "learning_rate": 2.6404350363530253e-07, "loss": 0.108, "step": 357000 }, { "epoch": 0.97, "learning_rate": 2.504077102230271e-07, "loss": 0.1015, "step": 357500 }, { "epoch": 0.98, "learning_rate": 2.3677191681075157e-07, "loss": 0.1052, "step": 358000 }, { "epoch": 0.98, "learning_rate": 2.2313612339847608e-07, "loss": 0.1095, "step": 358500 }, { "epoch": 0.98, "learning_rate": 2.095003299862006e-07, "loss": 0.1066, "step": 359000 }, { "epoch": 0.98, "learning_rate": 1.958645365739251e-07, "loss": 0.1114, "step": 359500 }, { "epoch": 0.98, "learning_rate": 1.822287431616496e-07, "loss": 0.1022, "step": 360000 }, { "epoch": 0.98, "learning_rate": 1.6859294974937413e-07, "loss": 0.1087, "step": 360500 }, { "epoch": 0.98, "learning_rate": 1.5495715633709864e-07, "loss": 0.104, "step": 361000 }, { "epoch": 0.99, "learning_rate": 1.4132136292482314e-07, "loss": 0.1079, "step": 361500 }, { "epoch": 0.99, "learning_rate": 1.2768556951254768e-07, "loss": 0.1021, "step": 362000 }, { "epoch": 0.99, "learning_rate": 1.1404977610027217e-07, "loss": 0.1052, "step": 362500 }, { "epoch": 0.99, "learning_rate": 1.0041398268799669e-07, "loss": 0.1057, "step": 363000 }, { "epoch": 0.99, "learning_rate": 8.677818927572121e-08, "loss": 0.1079, "step": 363500 }, { "epoch": 0.99, "learning_rate": 7.314239586344572e-08, "loss": 0.1086, "step": 364000 }, { "epoch": 0.99, "learning_rate": 5.9506602451170224e-08, "loss": 0.109, "step": 364500 }, { "epoch": 1.0, "learning_rate": 4.5870809038894743e-08, "loss": 0.1024, "step": 365000 }, { "epoch": 1.0, "learning_rate": 3.223501562661925e-08, "loss": 0.1052, "step": 365500 }, { "epoch": 1.0, "learning_rate": 1.8599222214343766e-08, "loss": 0.1085, "step": 366000 }, { "epoch": 1.0, "learning_rate": 4.963428802068278e-09, "loss": 0.1071, "step": 366500 }, { "epoch": 1.0, "step": 366682, "total_flos": 1.2550121084963002e+17, "train_loss": 0.11798365104506718, "train_runtime": 23878.2895, "train_samples_per_second": 122.85, "train_steps_per_second": 15.356 } ], "max_steps": 366682, "num_train_epochs": 1, "total_flos": 1.2550121084963002e+17, "trial_name": null, "trial_params": null }