| { |
| "best_metric": 0.080692358314991, |
| "best_model_checkpoint": "autotrain-sec4/checkpoint-10605", |
| "epoch": 3.0, |
| "eval_steps": 500, |
| "global_step": 10605, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.007072135785007072, |
| "grad_norm": 75.34838104248047, |
| "learning_rate": 6.504524886877829e-07, |
| "loss": 3.7557, |
| "step": 25 |
| }, |
| { |
| "epoch": 0.014144271570014143, |
| "grad_norm": 15.420210838317871, |
| "learning_rate": 1.3574660633484164e-06, |
| "loss": 2.4958, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.021216407355021217, |
| "grad_norm": 4.017354965209961, |
| "learning_rate": 2.0361990950226245e-06, |
| "loss": 1.1167, |
| "step": 75 |
| }, |
| { |
| "epoch": 0.028288543140028287, |
| "grad_norm": 3.0722529888153076, |
| "learning_rate": 2.743212669683258e-06, |
| "loss": 0.7577, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.03536067892503536, |
| "grad_norm": 4.105304718017578, |
| "learning_rate": 3.450226244343892e-06, |
| "loss": 0.6, |
| "step": 125 |
| }, |
| { |
| "epoch": 0.042432814710042434, |
| "grad_norm": 2.796954870223999, |
| "learning_rate": 4.157239819004525e-06, |
| "loss": 0.5213, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.04950495049504951, |
| "grad_norm": 2.0742604732513428, |
| "learning_rate": 4.864253393665159e-06, |
| "loss": 0.4437, |
| "step": 175 |
| }, |
| { |
| "epoch": 0.056577086280056574, |
| "grad_norm": 1.6229099035263062, |
| "learning_rate": 5.5712669683257925e-06, |
| "loss": 0.3827, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.06364922206506365, |
| "grad_norm": 2.0883893966674805, |
| "learning_rate": 6.278280542986425e-06, |
| "loss": 0.3385, |
| "step": 225 |
| }, |
| { |
| "epoch": 0.07072135785007072, |
| "grad_norm": 2.5166773796081543, |
| "learning_rate": 6.985294117647059e-06, |
| "loss": 0.2949, |
| "step": 250 |
| }, |
| { |
| "epoch": 0.07779349363507779, |
| "grad_norm": 2.4907472133636475, |
| "learning_rate": 7.692307692307694e-06, |
| "loss": 0.2776, |
| "step": 275 |
| }, |
| { |
| "epoch": 0.08486562942008487, |
| "grad_norm": 2.4089417457580566, |
| "learning_rate": 8.399321266968327e-06, |
| "loss": 0.2867, |
| "step": 300 |
| }, |
| { |
| "epoch": 0.09193776520509193, |
| "grad_norm": 1.0878061056137085, |
| "learning_rate": 9.10633484162896e-06, |
| "loss": 0.2567, |
| "step": 325 |
| }, |
| { |
| "epoch": 0.09900990099009901, |
| "grad_norm": 1.353794813156128, |
| "learning_rate": 9.813348416289593e-06, |
| "loss": 0.2377, |
| "step": 350 |
| }, |
| { |
| "epoch": 0.10608203677510608, |
| "grad_norm": 1.3143301010131836, |
| "learning_rate": 1.0520361990950227e-05, |
| "loss": 0.2446, |
| "step": 375 |
| }, |
| { |
| "epoch": 0.11315417256011315, |
| "grad_norm": 1.1061075925827026, |
| "learning_rate": 1.122737556561086e-05, |
| "loss": 0.2136, |
| "step": 400 |
| }, |
| { |
| "epoch": 0.12022630834512023, |
| "grad_norm": 4.132507801055908, |
| "learning_rate": 1.1934389140271494e-05, |
| "loss": 0.2089, |
| "step": 425 |
| }, |
| { |
| "epoch": 0.1272984441301273, |
| "grad_norm": 2.015286922454834, |
| "learning_rate": 1.2641402714932126e-05, |
| "loss": 0.206, |
| "step": 450 |
| }, |
| { |
| "epoch": 0.13437057991513437, |
| "grad_norm": 1.3202295303344727, |
| "learning_rate": 1.3348416289592761e-05, |
| "loss": 0.2135, |
| "step": 475 |
| }, |
| { |
| "epoch": 0.14144271570014144, |
| "grad_norm": 1.005441665649414, |
| "learning_rate": 1.4055429864253395e-05, |
| "loss": 0.198, |
| "step": 500 |
| }, |
| { |
| "epoch": 0.1485148514851485, |
| "grad_norm": 1.0634604692459106, |
| "learning_rate": 1.4762443438914029e-05, |
| "loss": 0.1869, |
| "step": 525 |
| }, |
| { |
| "epoch": 0.15558698727015557, |
| "grad_norm": 1.5153967142105103, |
| "learning_rate": 1.5469457013574662e-05, |
| "loss": 0.1938, |
| "step": 550 |
| }, |
| { |
| "epoch": 0.16265912305516267, |
| "grad_norm": 1.5015569925308228, |
| "learning_rate": 1.6176470588235296e-05, |
| "loss": 0.1916, |
| "step": 575 |
| }, |
| { |
| "epoch": 0.16973125884016974, |
| "grad_norm": 1.3801521062850952, |
| "learning_rate": 1.688348416289593e-05, |
| "loss": 0.1859, |
| "step": 600 |
| }, |
| { |
| "epoch": 0.1768033946251768, |
| "grad_norm": 1.4514083862304688, |
| "learning_rate": 1.7590497737556563e-05, |
| "loss": 0.1794, |
| "step": 625 |
| }, |
| { |
| "epoch": 0.18387553041018387, |
| "grad_norm": 1.1229180097579956, |
| "learning_rate": 1.8297511312217194e-05, |
| "loss": 0.181, |
| "step": 650 |
| }, |
| { |
| "epoch": 0.19094766619519093, |
| "grad_norm": 0.9584263563156128, |
| "learning_rate": 1.9004524886877827e-05, |
| "loss": 0.1702, |
| "step": 675 |
| }, |
| { |
| "epoch": 0.19801980198019803, |
| "grad_norm": 1.5446968078613281, |
| "learning_rate": 1.971153846153846e-05, |
| "loss": 0.1655, |
| "step": 700 |
| }, |
| { |
| "epoch": 0.2050919377652051, |
| "grad_norm": 1.180188775062561, |
| "learning_rate": 2.0418552036199095e-05, |
| "loss": 0.1783, |
| "step": 725 |
| }, |
| { |
| "epoch": 0.21216407355021216, |
| "grad_norm": 1.4047558307647705, |
| "learning_rate": 2.112556561085973e-05, |
| "loss": 0.1654, |
| "step": 750 |
| }, |
| { |
| "epoch": 0.21923620933521923, |
| "grad_norm": 1.6405079364776611, |
| "learning_rate": 2.1832579185520362e-05, |
| "loss": 0.1676, |
| "step": 775 |
| }, |
| { |
| "epoch": 0.2263083451202263, |
| "grad_norm": 0.799575924873352, |
| "learning_rate": 2.2539592760180996e-05, |
| "loss": 0.1529, |
| "step": 800 |
| }, |
| { |
| "epoch": 0.2333804809052334, |
| "grad_norm": 1.0612698793411255, |
| "learning_rate": 2.324660633484163e-05, |
| "loss": 0.158, |
| "step": 825 |
| }, |
| { |
| "epoch": 0.24045261669024046, |
| "grad_norm": 0.8970231413841248, |
| "learning_rate": 2.3953619909502263e-05, |
| "loss": 0.1556, |
| "step": 850 |
| }, |
| { |
| "epoch": 0.24752475247524752, |
| "grad_norm": 1.1954729557037354, |
| "learning_rate": 2.4660633484162897e-05, |
| "loss": 0.1512, |
| "step": 875 |
| }, |
| { |
| "epoch": 0.2545968882602546, |
| "grad_norm": 2.2156217098236084, |
| "learning_rate": 2.536764705882353e-05, |
| "loss": 0.1528, |
| "step": 900 |
| }, |
| { |
| "epoch": 0.26166902404526166, |
| "grad_norm": 1.4057120084762573, |
| "learning_rate": 2.6074660633484164e-05, |
| "loss": 0.1562, |
| "step": 925 |
| }, |
| { |
| "epoch": 0.26874115983026875, |
| "grad_norm": 1.9392261505126953, |
| "learning_rate": 2.6781674208144798e-05, |
| "loss": 0.1601, |
| "step": 950 |
| }, |
| { |
| "epoch": 0.2758132956152758, |
| "grad_norm": 1.114762306213379, |
| "learning_rate": 2.7488687782805435e-05, |
| "loss": 0.1616, |
| "step": 975 |
| }, |
| { |
| "epoch": 0.2828854314002829, |
| "grad_norm": 0.6488351225852966, |
| "learning_rate": 2.8195701357466065e-05, |
| "loss": 0.1564, |
| "step": 1000 |
| }, |
| { |
| "epoch": 0.28995756718529, |
| "grad_norm": 0.7562068700790405, |
| "learning_rate": 2.8902714932126696e-05, |
| "loss": 0.1447, |
| "step": 1025 |
| }, |
| { |
| "epoch": 0.297029702970297, |
| "grad_norm": 1.5005004405975342, |
| "learning_rate": 2.9609728506787333e-05, |
| "loss": 0.1411, |
| "step": 1050 |
| }, |
| { |
| "epoch": 0.3041018387553041, |
| "grad_norm": 0.9700536131858826, |
| "learning_rate": 3.0316742081447963e-05, |
| "loss": 0.1628, |
| "step": 1075 |
| }, |
| { |
| "epoch": 0.31117397454031115, |
| "grad_norm": 0.9135770797729492, |
| "learning_rate": 3.10237556561086e-05, |
| "loss": 0.1429, |
| "step": 1100 |
| }, |
| { |
| "epoch": 0.31824611032531824, |
| "grad_norm": 0.5135190486907959, |
| "learning_rate": 3.1730769230769234e-05, |
| "loss": 0.1349, |
| "step": 1125 |
| }, |
| { |
| "epoch": 0.32531824611032534, |
| "grad_norm": 0.7440194487571716, |
| "learning_rate": 3.243778280542987e-05, |
| "loss": 0.1511, |
| "step": 1150 |
| }, |
| { |
| "epoch": 0.3323903818953324, |
| "grad_norm": 0.774247407913208, |
| "learning_rate": 3.31447963800905e-05, |
| "loss": 0.1433, |
| "step": 1175 |
| }, |
| { |
| "epoch": 0.33946251768033947, |
| "grad_norm": 0.9245631098747253, |
| "learning_rate": 3.3851809954751135e-05, |
| "loss": 0.136, |
| "step": 1200 |
| }, |
| { |
| "epoch": 0.3465346534653465, |
| "grad_norm": 0.5323517918586731, |
| "learning_rate": 3.455882352941177e-05, |
| "loss": 0.1402, |
| "step": 1225 |
| }, |
| { |
| "epoch": 0.3536067892503536, |
| "grad_norm": 0.5974646210670471, |
| "learning_rate": 3.52658371040724e-05, |
| "loss": 0.139, |
| "step": 1250 |
| }, |
| { |
| "epoch": 0.3606789250353607, |
| "grad_norm": 0.7609033584594727, |
| "learning_rate": 3.5972850678733036e-05, |
| "loss": 0.1414, |
| "step": 1275 |
| }, |
| { |
| "epoch": 0.36775106082036774, |
| "grad_norm": 0.6598523259162903, |
| "learning_rate": 3.667986425339366e-05, |
| "loss": 0.1279, |
| "step": 1300 |
| }, |
| { |
| "epoch": 0.37482319660537483, |
| "grad_norm": 0.646514892578125, |
| "learning_rate": 3.73868778280543e-05, |
| "loss": 0.1318, |
| "step": 1325 |
| }, |
| { |
| "epoch": 0.38189533239038187, |
| "grad_norm": 0.6801839470863342, |
| "learning_rate": 3.809389140271493e-05, |
| "loss": 0.135, |
| "step": 1350 |
| }, |
| { |
| "epoch": 0.38896746817538896, |
| "grad_norm": 0.43455860018730164, |
| "learning_rate": 3.880090497737557e-05, |
| "loss": 0.1229, |
| "step": 1375 |
| }, |
| { |
| "epoch": 0.39603960396039606, |
| "grad_norm": 0.7492242455482483, |
| "learning_rate": 3.95079185520362e-05, |
| "loss": 0.1336, |
| "step": 1400 |
| }, |
| { |
| "epoch": 0.4031117397454031, |
| "grad_norm": 0.7798008918762207, |
| "learning_rate": 4.021493212669684e-05, |
| "loss": 0.1278, |
| "step": 1425 |
| }, |
| { |
| "epoch": 0.4101838755304102, |
| "grad_norm": 0.5796477794647217, |
| "learning_rate": 4.0921945701357465e-05, |
| "loss": 0.1231, |
| "step": 1450 |
| }, |
| { |
| "epoch": 0.41725601131541723, |
| "grad_norm": 0.8398697376251221, |
| "learning_rate": 4.1628959276018105e-05, |
| "loss": 0.1305, |
| "step": 1475 |
| }, |
| { |
| "epoch": 0.4243281471004243, |
| "grad_norm": 1.1518373489379883, |
| "learning_rate": 4.233597285067873e-05, |
| "loss": 0.1345, |
| "step": 1500 |
| }, |
| { |
| "epoch": 0.4314002828854314, |
| "grad_norm": 0.6339196562767029, |
| "learning_rate": 4.304298642533937e-05, |
| "loss": 0.1388, |
| "step": 1525 |
| }, |
| { |
| "epoch": 0.43847241867043846, |
| "grad_norm": 0.6023873090744019, |
| "learning_rate": 4.375e-05, |
| "loss": 0.1343, |
| "step": 1550 |
| }, |
| { |
| "epoch": 0.44554455445544555, |
| "grad_norm": 0.6470553874969482, |
| "learning_rate": 4.445701357466063e-05, |
| "loss": 0.1353, |
| "step": 1575 |
| }, |
| { |
| "epoch": 0.4526166902404526, |
| "grad_norm": 0.6824638843536377, |
| "learning_rate": 4.516402714932127e-05, |
| "loss": 0.1233, |
| "step": 1600 |
| }, |
| { |
| "epoch": 0.4596888260254597, |
| "grad_norm": 0.7944373488426208, |
| "learning_rate": 4.58710407239819e-05, |
| "loss": 0.1235, |
| "step": 1625 |
| }, |
| { |
| "epoch": 0.4667609618104668, |
| "grad_norm": 0.576715350151062, |
| "learning_rate": 4.6578054298642534e-05, |
| "loss": 0.1225, |
| "step": 1650 |
| }, |
| { |
| "epoch": 0.4738330975954738, |
| "grad_norm": 0.5373973846435547, |
| "learning_rate": 4.728506787330317e-05, |
| "loss": 0.1379, |
| "step": 1675 |
| }, |
| { |
| "epoch": 0.4809052333804809, |
| "grad_norm": 0.6271287798881531, |
| "learning_rate": 4.79920814479638e-05, |
| "loss": 0.1295, |
| "step": 1700 |
| }, |
| { |
| "epoch": 0.48797736916548795, |
| "grad_norm": 0.776996374130249, |
| "learning_rate": 4.8699095022624435e-05, |
| "loss": 0.1284, |
| "step": 1725 |
| }, |
| { |
| "epoch": 0.49504950495049505, |
| "grad_norm": 0.7526723742485046, |
| "learning_rate": 4.940610859728507e-05, |
| "loss": 0.1215, |
| "step": 1750 |
| }, |
| { |
| "epoch": 0.5021216407355021, |
| "grad_norm": 0.5230354070663452, |
| "learning_rate": 4.99999921989483e-05, |
| "loss": 0.1198, |
| "step": 1775 |
| }, |
| { |
| "epoch": 0.5091937765205092, |
| "grad_norm": 0.9118756651878357, |
| "learning_rate": 4.999958995831941e-05, |
| "loss": 0.134, |
| "step": 1800 |
| }, |
| { |
| "epoch": 0.5162659123055162, |
| "grad_norm": 0.7880003452301025, |
| "learning_rate": 4.9998578271728684e-05, |
| "loss": 0.1271, |
| "step": 1825 |
| }, |
| { |
| "epoch": 0.5233380480905233, |
| "grad_norm": 0.47186046838760376, |
| "learning_rate": 4.999695716383928e-05, |
| "loss": 0.1188, |
| "step": 1850 |
| }, |
| { |
| "epoch": 0.5304101838755304, |
| "grad_norm": 0.5088236331939697, |
| "learning_rate": 4.9994726674170947e-05, |
| "loss": 0.1179, |
| "step": 1875 |
| }, |
| { |
| "epoch": 0.5374823196605375, |
| "grad_norm": 0.6230493187904358, |
| "learning_rate": 4.999188685709908e-05, |
| "loss": 0.1262, |
| "step": 1900 |
| }, |
| { |
| "epoch": 0.5445544554455446, |
| "grad_norm": 0.9874520301818848, |
| "learning_rate": 4.998843778185343e-05, |
| "loss": 0.1244, |
| "step": 1925 |
| }, |
| { |
| "epoch": 0.5516265912305516, |
| "grad_norm": 0.5979165434837341, |
| "learning_rate": 4.998437953251637e-05, |
| "loss": 0.1291, |
| "step": 1950 |
| }, |
| { |
| "epoch": 0.5586987270155587, |
| "grad_norm": 0.8219141364097595, |
| "learning_rate": 4.997971220802088e-05, |
| "loss": 0.1155, |
| "step": 1975 |
| }, |
| { |
| "epoch": 0.5657708628005658, |
| "grad_norm": 0.41394418478012085, |
| "learning_rate": 4.997443592214809e-05, |
| "loss": 0.1198, |
| "step": 2000 |
| }, |
| { |
| "epoch": 0.5728429985855729, |
| "grad_norm": 1.253273844718933, |
| "learning_rate": 4.996855080352457e-05, |
| "loss": 0.1191, |
| "step": 2025 |
| }, |
| { |
| "epoch": 0.57991513437058, |
| "grad_norm": 0.9971447587013245, |
| "learning_rate": 4.9962056995619135e-05, |
| "loss": 0.1328, |
| "step": 2050 |
| }, |
| { |
| "epoch": 0.5869872701555869, |
| "grad_norm": 0.49998825788497925, |
| "learning_rate": 4.995495465673939e-05, |
| "loss": 0.115, |
| "step": 2075 |
| }, |
| { |
| "epoch": 0.594059405940594, |
| "grad_norm": 0.9231402277946472, |
| "learning_rate": 4.994724396002783e-05, |
| "loss": 0.1131, |
| "step": 2100 |
| }, |
| { |
| "epoch": 0.6011315417256011, |
| "grad_norm": 0.5812481641769409, |
| "learning_rate": 4.9938925093457684e-05, |
| "loss": 0.103, |
| "step": 2125 |
| }, |
| { |
| "epoch": 0.6082036775106082, |
| "grad_norm": 0.602626383304596, |
| "learning_rate": 4.992999825982824e-05, |
| "loss": 0.1097, |
| "step": 2150 |
| }, |
| { |
| "epoch": 0.6152758132956153, |
| "grad_norm": 0.4413115084171295, |
| "learning_rate": 4.9920463676759975e-05, |
| "loss": 0.1081, |
| "step": 2175 |
| }, |
| { |
| "epoch": 0.6223479490806223, |
| "grad_norm": 0.796340823173523, |
| "learning_rate": 4.991032157668924e-05, |
| "loss": 0.1134, |
| "step": 2200 |
| }, |
| { |
| "epoch": 0.6294200848656294, |
| "grad_norm": 0.8647779822349548, |
| "learning_rate": 4.989957220686257e-05, |
| "loss": 0.1169, |
| "step": 2225 |
| }, |
| { |
| "epoch": 0.6364922206506365, |
| "grad_norm": 0.5049439072608948, |
| "learning_rate": 4.988821582933065e-05, |
| "loss": 0.1155, |
| "step": 2250 |
| }, |
| { |
| "epoch": 0.6435643564356436, |
| "grad_norm": 0.5249194502830505, |
| "learning_rate": 4.9876252720941974e-05, |
| "loss": 0.1076, |
| "step": 2275 |
| }, |
| { |
| "epoch": 0.6506364922206507, |
| "grad_norm": 0.9044466018676758, |
| "learning_rate": 4.986368317333603e-05, |
| "loss": 0.1171, |
| "step": 2300 |
| }, |
| { |
| "epoch": 0.6577086280056577, |
| "grad_norm": 0.6795557141304016, |
| "learning_rate": 4.985050749293626e-05, |
| "loss": 0.1099, |
| "step": 2325 |
| }, |
| { |
| "epoch": 0.6647807637906648, |
| "grad_norm": 0.5548400282859802, |
| "learning_rate": 4.983672600094253e-05, |
| "loss": 0.1145, |
| "step": 2350 |
| }, |
| { |
| "epoch": 0.6718528995756718, |
| "grad_norm": 0.8753097057342529, |
| "learning_rate": 4.982233903332335e-05, |
| "loss": 0.1086, |
| "step": 2375 |
| }, |
| { |
| "epoch": 0.6789250353606789, |
| "grad_norm": 0.7015976309776306, |
| "learning_rate": 4.980734694080763e-05, |
| "loss": 0.1134, |
| "step": 2400 |
| }, |
| { |
| "epoch": 0.685997171145686, |
| "grad_norm": 0.3631691336631775, |
| "learning_rate": 4.979175008887619e-05, |
| "loss": 0.1102, |
| "step": 2425 |
| }, |
| { |
| "epoch": 0.693069306930693, |
| "grad_norm": 0.512153685092926, |
| "learning_rate": 4.977554885775278e-05, |
| "loss": 0.1135, |
| "step": 2450 |
| }, |
| { |
| "epoch": 0.7001414427157001, |
| "grad_norm": 0.62693190574646, |
| "learning_rate": 4.975874364239491e-05, |
| "loss": 0.1104, |
| "step": 2475 |
| }, |
| { |
| "epoch": 0.7072135785007072, |
| "grad_norm": 0.8053261637687683, |
| "learning_rate": 4.974133485248409e-05, |
| "loss": 0.1092, |
| "step": 2500 |
| }, |
| { |
| "epoch": 0.7142857142857143, |
| "grad_norm": 0.8354981541633606, |
| "learning_rate": 4.972332291241597e-05, |
| "loss": 0.1155, |
| "step": 2525 |
| }, |
| { |
| "epoch": 0.7213578500707214, |
| "grad_norm": 1.4328135251998901, |
| "learning_rate": 4.97047082612899e-05, |
| "loss": 0.1071, |
| "step": 2550 |
| }, |
| { |
| "epoch": 0.7284299858557284, |
| "grad_norm": 0.5748372077941895, |
| "learning_rate": 4.96854913528983e-05, |
| "loss": 0.105, |
| "step": 2575 |
| }, |
| { |
| "epoch": 0.7355021216407355, |
| "grad_norm": 0.5140411257743835, |
| "learning_rate": 4.966567265571553e-05, |
| "loss": 0.1076, |
| "step": 2600 |
| }, |
| { |
| "epoch": 0.7425742574257426, |
| "grad_norm": 0.42150813341140747, |
| "learning_rate": 4.9645252652886524e-05, |
| "loss": 0.1076, |
| "step": 2625 |
| }, |
| { |
| "epoch": 0.7496463932107497, |
| "grad_norm": 0.7284517288208008, |
| "learning_rate": 4.9624231842214966e-05, |
| "loss": 0.1147, |
| "step": 2650 |
| }, |
| { |
| "epoch": 0.7567185289957568, |
| "grad_norm": 0.3435399830341339, |
| "learning_rate": 4.960261073615119e-05, |
| "loss": 0.117, |
| "step": 2675 |
| }, |
| { |
| "epoch": 0.7637906647807637, |
| "grad_norm": 0.5726285576820374, |
| "learning_rate": 4.9580389861779676e-05, |
| "loss": 0.1028, |
| "step": 2700 |
| }, |
| { |
| "epoch": 0.7708628005657708, |
| "grad_norm": 0.6155900955200195, |
| "learning_rate": 4.955756976080619e-05, |
| "loss": 0.1059, |
| "step": 2725 |
| }, |
| { |
| "epoch": 0.7779349363507779, |
| "grad_norm": 0.5489698052406311, |
| "learning_rate": 4.95341509895446e-05, |
| "loss": 0.1085, |
| "step": 2750 |
| }, |
| { |
| "epoch": 0.785007072135785, |
| "grad_norm": 0.4485224783420563, |
| "learning_rate": 4.951013411890329e-05, |
| "loss": 0.1046, |
| "step": 2775 |
| }, |
| { |
| "epoch": 0.7920792079207921, |
| "grad_norm": 1.4330599308013916, |
| "learning_rate": 4.9485519734371254e-05, |
| "loss": 0.1066, |
| "step": 2800 |
| }, |
| { |
| "epoch": 0.7991513437057991, |
| "grad_norm": 0.5611838102340698, |
| "learning_rate": 4.946030843600382e-05, |
| "loss": 0.108, |
| "step": 2825 |
| }, |
| { |
| "epoch": 0.8062234794908062, |
| "grad_norm": 0.44172587990760803, |
| "learning_rate": 4.943450083840804e-05, |
| "loss": 0.1115, |
| "step": 2850 |
| }, |
| { |
| "epoch": 0.8132956152758133, |
| "grad_norm": 0.5805100798606873, |
| "learning_rate": 4.940809757072767e-05, |
| "loss": 0.1017, |
| "step": 2875 |
| }, |
| { |
| "epoch": 0.8203677510608204, |
| "grad_norm": 0.42033523321151733, |
| "learning_rate": 4.938109927662786e-05, |
| "loss": 0.0987, |
| "step": 2900 |
| }, |
| { |
| "epoch": 0.8274398868458275, |
| "grad_norm": 0.519852876663208, |
| "learning_rate": 4.9353506614279465e-05, |
| "loss": 0.1085, |
| "step": 2925 |
| }, |
| { |
| "epoch": 0.8345120226308345, |
| "grad_norm": 0.41062799096107483, |
| "learning_rate": 4.932532025634298e-05, |
| "loss": 0.1122, |
| "step": 2950 |
| }, |
| { |
| "epoch": 0.8415841584158416, |
| "grad_norm": 0.4441488981246948, |
| "learning_rate": 4.929654088995216e-05, |
| "loss": 0.1075, |
| "step": 2975 |
| }, |
| { |
| "epoch": 0.8486562942008486, |
| "grad_norm": 0.5983813405036926, |
| "learning_rate": 4.926716921669724e-05, |
| "loss": 0.105, |
| "step": 3000 |
| }, |
| { |
| "epoch": 0.8557284299858557, |
| "grad_norm": 0.4088096022605896, |
| "learning_rate": 4.92372059526079e-05, |
| "loss": 0.1055, |
| "step": 3025 |
| }, |
| { |
| "epoch": 0.8628005657708628, |
| "grad_norm": 0.3722389042377472, |
| "learning_rate": 4.92066518281357e-05, |
| "loss": 0.1016, |
| "step": 3050 |
| }, |
| { |
| "epoch": 0.8698727015558698, |
| "grad_norm": 0.3515225946903229, |
| "learning_rate": 4.9175507588136395e-05, |
| "loss": 0.1061, |
| "step": 3075 |
| }, |
| { |
| "epoch": 0.8769448373408769, |
| "grad_norm": 0.6641696691513062, |
| "learning_rate": 4.914377399185167e-05, |
| "loss": 0.0976, |
| "step": 3100 |
| }, |
| { |
| "epoch": 0.884016973125884, |
| "grad_norm": 1.0824599266052246, |
| "learning_rate": 4.911145181289072e-05, |
| "loss": 0.0993, |
| "step": 3125 |
| }, |
| { |
| "epoch": 0.8910891089108911, |
| "grad_norm": 0.9281112551689148, |
| "learning_rate": 4.90785418392113e-05, |
| "loss": 0.1049, |
| "step": 3150 |
| }, |
| { |
| "epoch": 0.8981612446958982, |
| "grad_norm": 0.6602789163589478, |
| "learning_rate": 4.904504487310061e-05, |
| "loss": 0.1105, |
| "step": 3175 |
| }, |
| { |
| "epoch": 0.9052333804809052, |
| "grad_norm": 0.522657036781311, |
| "learning_rate": 4.901096173115567e-05, |
| "loss": 0.098, |
| "step": 3200 |
| }, |
| { |
| "epoch": 0.9123055162659123, |
| "grad_norm": 0.38893359899520874, |
| "learning_rate": 4.897629324426343e-05, |
| "loss": 0.0992, |
| "step": 3225 |
| }, |
| { |
| "epoch": 0.9193776520509194, |
| "grad_norm": 0.9329886436462402, |
| "learning_rate": 4.894104025758054e-05, |
| "loss": 0.1048, |
| "step": 3250 |
| }, |
| { |
| "epoch": 0.9264497878359265, |
| "grad_norm": 0.4844260811805725, |
| "learning_rate": 4.890520363051269e-05, |
| "loss": 0.1019, |
| "step": 3275 |
| }, |
| { |
| "epoch": 0.9335219236209336, |
| "grad_norm": 0.39830756187438965, |
| "learning_rate": 4.886878423669373e-05, |
| "loss": 0.0996, |
| "step": 3300 |
| }, |
| { |
| "epoch": 0.9405940594059405, |
| "grad_norm": 0.3621814548969269, |
| "learning_rate": 4.8831782963964314e-05, |
| "loss": 0.0959, |
| "step": 3325 |
| }, |
| { |
| "epoch": 0.9476661951909476, |
| "grad_norm": 0.43947598338127136, |
| "learning_rate": 4.87942007143503e-05, |
| "loss": 0.1028, |
| "step": 3350 |
| }, |
| { |
| "epoch": 0.9547383309759547, |
| "grad_norm": 0.4322707951068878, |
| "learning_rate": 4.87560384040407e-05, |
| "loss": 0.0957, |
| "step": 3375 |
| }, |
| { |
| "epoch": 0.9618104667609618, |
| "grad_norm": 0.7152990102767944, |
| "learning_rate": 4.8717296963365446e-05, |
| "loss": 0.0983, |
| "step": 3400 |
| }, |
| { |
| "epoch": 0.9688826025459689, |
| "grad_norm": 0.36375129222869873, |
| "learning_rate": 4.8677977336772576e-05, |
| "loss": 0.0986, |
| "step": 3425 |
| }, |
| { |
| "epoch": 0.9759547383309759, |
| "grad_norm": 0.3541475534439087, |
| "learning_rate": 4.8638080482805334e-05, |
| "loss": 0.1039, |
| "step": 3450 |
| }, |
| { |
| "epoch": 0.983026874115983, |
| "grad_norm": 0.5277055501937866, |
| "learning_rate": 4.8597607374078734e-05, |
| "loss": 0.1009, |
| "step": 3475 |
| }, |
| { |
| "epoch": 0.9900990099009901, |
| "grad_norm": 0.32760316133499146, |
| "learning_rate": 4.855655899725587e-05, |
| "loss": 0.0967, |
| "step": 3500 |
| }, |
| { |
| "epoch": 0.9971711456859972, |
| "grad_norm": 0.4430937170982361, |
| "learning_rate": 4.851493635302384e-05, |
| "loss": 0.0959, |
| "step": 3525 |
| }, |
| { |
| "epoch": 1.0, |
| "eval_accuracy": 0.9641816937160642, |
| "eval_f1": 0.855812468110255, |
| "eval_loss": 0.10302519798278809, |
| "eval_precision": 0.8459347469557771, |
| "eval_recall": 0.8659235929015533, |
| "eval_runtime": 65.6322, |
| "eval_samples_per_second": 765.966, |
| "eval_steps_per_second": 3.002, |
| "step": 3535 |
| }, |
| { |
| "epoch": 1.0042432814710043, |
| "grad_norm": 0.5774176120758057, |
| "learning_rate": 4.8472740456069404e-05, |
| "loss": 0.0883, |
| "step": 3550 |
| }, |
| { |
| "epoch": 1.0113154172560113, |
| "grad_norm": 0.6957634687423706, |
| "learning_rate": 4.842997233505418e-05, |
| "loss": 0.0774, |
| "step": 3575 |
| }, |
| { |
| "epoch": 1.0183875530410185, |
| "grad_norm": 0.3464818000793457, |
| "learning_rate": 4.838663303258961e-05, |
| "loss": 0.0823, |
| "step": 3600 |
| }, |
| { |
| "epoch": 1.0254596888260255, |
| "grad_norm": 0.6941576600074768, |
| "learning_rate": 4.8342723605211556e-05, |
| "loss": 0.0826, |
| "step": 3625 |
| }, |
| { |
| "epoch": 1.0325318246110324, |
| "grad_norm": 0.44520071148872375, |
| "learning_rate": 4.829824512335449e-05, |
| "loss": 0.0867, |
| "step": 3650 |
| }, |
| { |
| "epoch": 1.0396039603960396, |
| "grad_norm": 0.3814884126186371, |
| "learning_rate": 4.825319867132545e-05, |
| "loss": 0.0869, |
| "step": 3675 |
| }, |
| { |
| "epoch": 1.0466760961810466, |
| "grad_norm": 0.44684168696403503, |
| "learning_rate": 4.8207585347277574e-05, |
| "loss": 0.0842, |
| "step": 3700 |
| }, |
| { |
| "epoch": 1.0537482319660538, |
| "grad_norm": 0.7008658051490784, |
| "learning_rate": 4.816140626318334e-05, |
| "loss": 0.0864, |
| "step": 3725 |
| }, |
| { |
| "epoch": 1.0608203677510608, |
| "grad_norm": 0.45034533739089966, |
| "learning_rate": 4.8114662544807476e-05, |
| "loss": 0.0801, |
| "step": 3750 |
| }, |
| { |
| "epoch": 1.0678925035360678, |
| "grad_norm": 0.6562340259552002, |
| "learning_rate": 4.8067355331679485e-05, |
| "loss": 0.0814, |
| "step": 3775 |
| }, |
| { |
| "epoch": 1.074964639321075, |
| "grad_norm": 0.7696852684020996, |
| "learning_rate": 4.801948577706588e-05, |
| "loss": 0.0837, |
| "step": 3800 |
| }, |
| { |
| "epoch": 1.082036775106082, |
| "grad_norm": 0.38005831837654114, |
| "learning_rate": 4.7971055047942074e-05, |
| "loss": 0.0836, |
| "step": 3825 |
| }, |
| { |
| "epoch": 1.0891089108910892, |
| "grad_norm": 0.5856048464775085, |
| "learning_rate": 4.792206432496392e-05, |
| "loss": 0.0776, |
| "step": 3850 |
| }, |
| { |
| "epoch": 1.0961810466760962, |
| "grad_norm": 0.3622016906738281, |
| "learning_rate": 4.787251480243895e-05, |
| "loss": 0.084, |
| "step": 3875 |
| }, |
| { |
| "epoch": 1.1032531824611032, |
| "grad_norm": 0.3826795816421509, |
| "learning_rate": 4.782240768829722e-05, |
| "loss": 0.0822, |
| "step": 3900 |
| }, |
| { |
| "epoch": 1.1103253182461104, |
| "grad_norm": 0.4321337938308716, |
| "learning_rate": 4.77717442040619e-05, |
| "loss": 0.0797, |
| "step": 3925 |
| }, |
| { |
| "epoch": 1.1173974540311173, |
| "grad_norm": 0.3895370364189148, |
| "learning_rate": 4.772052558481949e-05, |
| "loss": 0.0808, |
| "step": 3950 |
| }, |
| { |
| "epoch": 1.1244695898161245, |
| "grad_norm": 0.3963691294193268, |
| "learning_rate": 4.766875307918969e-05, |
| "loss": 0.0829, |
| "step": 3975 |
| }, |
| { |
| "epoch": 1.1315417256011315, |
| "grad_norm": 1.1154179573059082, |
| "learning_rate": 4.7616427949294985e-05, |
| "loss": 0.0864, |
| "step": 4000 |
| }, |
| { |
| "epoch": 1.1386138613861387, |
| "grad_norm": 0.6903427839279175, |
| "learning_rate": 4.756355147072986e-05, |
| "loss": 0.0816, |
| "step": 4025 |
| }, |
| { |
| "epoch": 1.1456859971711457, |
| "grad_norm": 0.3236379325389862, |
| "learning_rate": 4.751012493252971e-05, |
| "loss": 0.085, |
| "step": 4050 |
| }, |
| { |
| "epoch": 1.1527581329561527, |
| "grad_norm": 0.31708064675331116, |
| "learning_rate": 4.745614963713941e-05, |
| "loss": 0.0824, |
| "step": 4075 |
| }, |
| { |
| "epoch": 1.15983026874116, |
| "grad_norm": 0.47363677620887756, |
| "learning_rate": 4.740162690038159e-05, |
| "loss": 0.08, |
| "step": 4100 |
| }, |
| { |
| "epoch": 1.166902404526167, |
| "grad_norm": 0.5583330988883972, |
| "learning_rate": 4.7346558051424505e-05, |
| "loss": 0.0864, |
| "step": 4125 |
| }, |
| { |
| "epoch": 1.1739745403111739, |
| "grad_norm": 0.4270670413970947, |
| "learning_rate": 4.729094443274969e-05, |
| "loss": 0.0774, |
| "step": 4150 |
| }, |
| { |
| "epoch": 1.181046676096181, |
| "grad_norm": 0.4069669544696808, |
| "learning_rate": 4.7234787400119176e-05, |
| "loss": 0.0856, |
| "step": 4175 |
| }, |
| { |
| "epoch": 1.188118811881188, |
| "grad_norm": 0.3747076988220215, |
| "learning_rate": 4.717808832254251e-05, |
| "loss": 0.09, |
| "step": 4200 |
| }, |
| { |
| "epoch": 1.1951909476661953, |
| "grad_norm": 0.45713725686073303, |
| "learning_rate": 4.71208485822433e-05, |
| "loss": 0.0758, |
| "step": 4225 |
| }, |
| { |
| "epoch": 1.2022630834512023, |
| "grad_norm": 0.41270920634269714, |
| "learning_rate": 4.7063069574625595e-05, |
| "loss": 0.0849, |
| "step": 4250 |
| }, |
| { |
| "epoch": 1.2093352192362095, |
| "grad_norm": 0.7367244362831116, |
| "learning_rate": 4.70047527082398e-05, |
| "loss": 0.0938, |
| "step": 4275 |
| }, |
| { |
| "epoch": 1.2164073550212164, |
| "grad_norm": 0.8561830520629883, |
| "learning_rate": 4.6945899404748376e-05, |
| "loss": 0.0822, |
| "step": 4300 |
| }, |
| { |
| "epoch": 1.2234794908062234, |
| "grad_norm": 0.8913972973823547, |
| "learning_rate": 4.6886511098891194e-05, |
| "loss": 0.0898, |
| "step": 4325 |
| }, |
| { |
| "epoch": 1.2305516265912306, |
| "grad_norm": 2.0065510272979736, |
| "learning_rate": 4.682658923845052e-05, |
| "loss": 0.0784, |
| "step": 4350 |
| }, |
| { |
| "epoch": 1.2376237623762376, |
| "grad_norm": 0.36617517471313477, |
| "learning_rate": 4.676613528421575e-05, |
| "loss": 0.0775, |
| "step": 4375 |
| }, |
| { |
| "epoch": 1.2446958981612446, |
| "grad_norm": 0.3758735954761505, |
| "learning_rate": 4.67051507099478e-05, |
| "loss": 0.081, |
| "step": 4400 |
| }, |
| { |
| "epoch": 1.2517680339462518, |
| "grad_norm": 0.4398355782032013, |
| "learning_rate": 4.664363700234312e-05, |
| "loss": 0.0847, |
| "step": 4425 |
| }, |
| { |
| "epoch": 1.2588401697312588, |
| "grad_norm": 0.4554755687713623, |
| "learning_rate": 4.658159566099757e-05, |
| "loss": 0.0828, |
| "step": 4450 |
| }, |
| { |
| "epoch": 1.265912305516266, |
| "grad_norm": 0.3456956446170807, |
| "learning_rate": 4.651902819836974e-05, |
| "loss": 0.0825, |
| "step": 4475 |
| }, |
| { |
| "epoch": 1.272984441301273, |
| "grad_norm": 0.38322913646698, |
| "learning_rate": 4.645593613974415e-05, |
| "loss": 0.0812, |
| "step": 4500 |
| }, |
| { |
| "epoch": 1.2800565770862802, |
| "grad_norm": 0.5935035943984985, |
| "learning_rate": 4.639232102319403e-05, |
| "loss": 0.0813, |
| "step": 4525 |
| }, |
| { |
| "epoch": 1.2871287128712872, |
| "grad_norm": 0.39350178837776184, |
| "learning_rate": 4.632818439954386e-05, |
| "loss": 0.0828, |
| "step": 4550 |
| }, |
| { |
| "epoch": 1.2942008486562941, |
| "grad_norm": 0.36052772402763367, |
| "learning_rate": 4.626352783233154e-05, |
| "loss": 0.0826, |
| "step": 4575 |
| }, |
| { |
| "epoch": 1.3012729844413014, |
| "grad_norm": 0.3978725075721741, |
| "learning_rate": 4.619835289777027e-05, |
| "loss": 0.0831, |
| "step": 4600 |
| }, |
| { |
| "epoch": 1.3083451202263083, |
| "grad_norm": 0.4236008822917938, |
| "learning_rate": 4.6132661184710137e-05, |
| "loss": 0.0895, |
| "step": 4625 |
| }, |
| { |
| "epoch": 1.3154172560113153, |
| "grad_norm": 0.4994755983352661, |
| "learning_rate": 4.606645429459937e-05, |
| "loss": 0.0882, |
| "step": 4650 |
| }, |
| { |
| "epoch": 1.3224893917963225, |
| "grad_norm": 1.0472633838653564, |
| "learning_rate": 4.5999733841445325e-05, |
| "loss": 0.0751, |
| "step": 4675 |
| }, |
| { |
| "epoch": 1.3295615275813295, |
| "grad_norm": 0.29904720187187195, |
| "learning_rate": 4.593250145177511e-05, |
| "loss": 0.0859, |
| "step": 4700 |
| }, |
| { |
| "epoch": 1.3366336633663367, |
| "grad_norm": 0.3832816779613495, |
| "learning_rate": 4.5864758764595927e-05, |
| "loss": 0.0785, |
| "step": 4725 |
| }, |
| { |
| "epoch": 1.3437057991513437, |
| "grad_norm": 0.49846336245536804, |
| "learning_rate": 4.5796507431355174e-05, |
| "loss": 0.075, |
| "step": 4750 |
| }, |
| { |
| "epoch": 1.350777934936351, |
| "grad_norm": 0.379282146692276, |
| "learning_rate": 4.5727749115900094e-05, |
| "loss": 0.0858, |
| "step": 4775 |
| }, |
| { |
| "epoch": 1.3578500707213579, |
| "grad_norm": 0.5583311915397644, |
| "learning_rate": 4.56584854944373e-05, |
| "loss": 0.0951, |
| "step": 4800 |
| }, |
| { |
| "epoch": 1.3649222065063649, |
| "grad_norm": 0.48655304312705994, |
| "learning_rate": 4.5588718255491865e-05, |
| "loss": 0.0884, |
| "step": 4825 |
| }, |
| { |
| "epoch": 1.371994342291372, |
| "grad_norm": 0.6549155116081238, |
| "learning_rate": 4.5518449099866155e-05, |
| "loss": 0.0858, |
| "step": 4850 |
| }, |
| { |
| "epoch": 1.379066478076379, |
| "grad_norm": 0.3174663186073303, |
| "learning_rate": 4.5447679740598403e-05, |
| "loss": 0.0837, |
| "step": 4875 |
| }, |
| { |
| "epoch": 1.386138613861386, |
| "grad_norm": 0.36421680450439453, |
| "learning_rate": 4.537641190292091e-05, |
| "loss": 0.0841, |
| "step": 4900 |
| }, |
| { |
| "epoch": 1.3932107496463932, |
| "grad_norm": 0.3484024107456207, |
| "learning_rate": 4.530464732421801e-05, |
| "loss": 0.0886, |
| "step": 4925 |
| }, |
| { |
| "epoch": 1.4002828854314002, |
| "grad_norm": 0.7408681511878967, |
| "learning_rate": 4.523238775398371e-05, |
| "loss": 0.0813, |
| "step": 4950 |
| }, |
| { |
| "epoch": 1.4073550212164074, |
| "grad_norm": 0.35419762134552, |
| "learning_rate": 4.515963495377902e-05, |
| "loss": 0.0768, |
| "step": 4975 |
| }, |
| { |
| "epoch": 1.4144271570014144, |
| "grad_norm": 0.3253655731678009, |
| "learning_rate": 4.508639069718904e-05, |
| "loss": 0.0776, |
| "step": 5000 |
| }, |
| { |
| "epoch": 1.4214992927864216, |
| "grad_norm": 0.3978010416030884, |
| "learning_rate": 4.501265676977974e-05, |
| "loss": 0.077, |
| "step": 5025 |
| }, |
| { |
| "epoch": 1.4285714285714286, |
| "grad_norm": 0.2976684868335724, |
| "learning_rate": 4.493843496905434e-05, |
| "loss": 0.0727, |
| "step": 5050 |
| }, |
| { |
| "epoch": 1.4356435643564356, |
| "grad_norm": 0.3433378040790558, |
| "learning_rate": 4.486372710440961e-05, |
| "loss": 0.0804, |
| "step": 5075 |
| }, |
| { |
| "epoch": 1.4427157001414428, |
| "grad_norm": 0.5507691502571106, |
| "learning_rate": 4.478853499709167e-05, |
| "loss": 0.0852, |
| "step": 5100 |
| }, |
| { |
| "epoch": 1.4497878359264498, |
| "grad_norm": 0.36808139085769653, |
| "learning_rate": 4.471286048015163e-05, |
| "loss": 0.0832, |
| "step": 5125 |
| }, |
| { |
| "epoch": 1.4568599717114568, |
| "grad_norm": 0.3539692163467407, |
| "learning_rate": 4.46367053984009e-05, |
| "loss": 0.0817, |
| "step": 5150 |
| }, |
| { |
| "epoch": 1.463932107496464, |
| "grad_norm": 0.8253048658370972, |
| "learning_rate": 4.4560071608366216e-05, |
| "loss": 0.0826, |
| "step": 5175 |
| }, |
| { |
| "epoch": 1.471004243281471, |
| "grad_norm": 0.4476601779460907, |
| "learning_rate": 4.448296097824437e-05, |
| "loss": 0.088, |
| "step": 5200 |
| }, |
| { |
| "epoch": 1.4780763790664782, |
| "grad_norm": 1.5917980670928955, |
| "learning_rate": 4.440537538785668e-05, |
| "loss": 0.0828, |
| "step": 5225 |
| }, |
| { |
| "epoch": 1.4851485148514851, |
| "grad_norm": 0.6799459457397461, |
| "learning_rate": 4.432731672860316e-05, |
| "loss": 0.0786, |
| "step": 5250 |
| }, |
| { |
| "epoch": 1.4922206506364923, |
| "grad_norm": 0.34660467505455017, |
| "learning_rate": 4.424878690341642e-05, |
| "loss": 0.0749, |
| "step": 5275 |
| }, |
| { |
| "epoch": 1.4992927864214993, |
| "grad_norm": 0.45074212551116943, |
| "learning_rate": 4.4169787826715225e-05, |
| "loss": 0.0904, |
| "step": 5300 |
| }, |
| { |
| "epoch": 1.5063649222065063, |
| "grad_norm": 0.41240620613098145, |
| "learning_rate": 4.4090321424357914e-05, |
| "loss": 0.078, |
| "step": 5325 |
| }, |
| { |
| "epoch": 1.5134370579915135, |
| "grad_norm": 0.3282380998134613, |
| "learning_rate": 4.401038963359538e-05, |
| "loss": 0.0835, |
| "step": 5350 |
| }, |
| { |
| "epoch": 1.5205091937765205, |
| "grad_norm": 0.3657848536968231, |
| "learning_rate": 4.3929994403023865e-05, |
| "loss": 0.0801, |
| "step": 5375 |
| }, |
| { |
| "epoch": 1.5275813295615275, |
| "grad_norm": 0.3543812930583954, |
| "learning_rate": 4.384913769253746e-05, |
| "loss": 0.081, |
| "step": 5400 |
| }, |
| { |
| "epoch": 1.5346534653465347, |
| "grad_norm": 0.8204165697097778, |
| "learning_rate": 4.376782147328031e-05, |
| "loss": 0.0757, |
| "step": 5425 |
| }, |
| { |
| "epoch": 1.541725601131542, |
| "grad_norm": 0.4716034233570099, |
| "learning_rate": 4.368604772759861e-05, |
| "loss": 0.0818, |
| "step": 5450 |
| }, |
| { |
| "epoch": 1.5487977369165487, |
| "grad_norm": 0.3865686357021332, |
| "learning_rate": 4.360381844899221e-05, |
| "loss": 0.076, |
| "step": 5475 |
| }, |
| { |
| "epoch": 1.5558698727015559, |
| "grad_norm": 0.5982063412666321, |
| "learning_rate": 4.352113564206606e-05, |
| "loss": 0.0861, |
| "step": 5500 |
| }, |
| { |
| "epoch": 1.562942008486563, |
| "grad_norm": 0.4285977780818939, |
| "learning_rate": 4.343800132248132e-05, |
| "loss": 0.0767, |
| "step": 5525 |
| }, |
| { |
| "epoch": 1.57001414427157, |
| "grad_norm": 0.542254626750946, |
| "learning_rate": 4.3354417516906274e-05, |
| "loss": 0.0814, |
| "step": 5550 |
| }, |
| { |
| "epoch": 1.577086280056577, |
| "grad_norm": 0.273899644613266, |
| "learning_rate": 4.327038626296682e-05, |
| "loss": 0.0759, |
| "step": 5575 |
| }, |
| { |
| "epoch": 1.5841584158415842, |
| "grad_norm": 0.3528795540332794, |
| "learning_rate": 4.318590960919692e-05, |
| "loss": 0.0785, |
| "step": 5600 |
| }, |
| { |
| "epoch": 1.5912305516265912, |
| "grad_norm": 0.4583662748336792, |
| "learning_rate": 4.3100989614988566e-05, |
| "loss": 0.0831, |
| "step": 5625 |
| }, |
| { |
| "epoch": 1.5983026874115982, |
| "grad_norm": 0.4005572497844696, |
| "learning_rate": 4.3015628350541595e-05, |
| "loss": 0.0781, |
| "step": 5650 |
| }, |
| { |
| "epoch": 1.6053748231966054, |
| "grad_norm": 0.3729841411113739, |
| "learning_rate": 4.2929827896813274e-05, |
| "loss": 0.0849, |
| "step": 5675 |
| }, |
| { |
| "epoch": 1.6124469589816126, |
| "grad_norm": 0.37740039825439453, |
| "learning_rate": 4.284359034546751e-05, |
| "loss": 0.0819, |
| "step": 5700 |
| }, |
| { |
| "epoch": 1.6195190947666194, |
| "grad_norm": 0.45638662576675415, |
| "learning_rate": 4.2756917798823874e-05, |
| "loss": 0.0749, |
| "step": 5725 |
| }, |
| { |
| "epoch": 1.6265912305516266, |
| "grad_norm": 0.3588665723800659, |
| "learning_rate": 4.2669812369806374e-05, |
| "loss": 0.0715, |
| "step": 5750 |
| }, |
| { |
| "epoch": 1.6336633663366338, |
| "grad_norm": 0.3274244964122772, |
| "learning_rate": 4.258227618189191e-05, |
| "loss": 0.0707, |
| "step": 5775 |
| }, |
| { |
| "epoch": 1.6407355021216408, |
| "grad_norm": 0.36558204889297485, |
| "learning_rate": 4.249431136905853e-05, |
| "loss": 0.0732, |
| "step": 5800 |
| }, |
| { |
| "epoch": 1.6478076379066477, |
| "grad_norm": 0.32020628452301025, |
| "learning_rate": 4.240592007573341e-05, |
| "loss": 0.0787, |
| "step": 5825 |
| }, |
| { |
| "epoch": 1.654879773691655, |
| "grad_norm": 0.4697955846786499, |
| "learning_rate": 4.231710445674056e-05, |
| "loss": 0.0707, |
| "step": 5850 |
| }, |
| { |
| "epoch": 1.661951909476662, |
| "grad_norm": 0.3986113965511322, |
| "learning_rate": 4.222786667724832e-05, |
| "loss": 0.0746, |
| "step": 5875 |
| }, |
| { |
| "epoch": 1.669024045261669, |
| "grad_norm": 0.35733580589294434, |
| "learning_rate": 4.213820891271652e-05, |
| "loss": 0.0787, |
| "step": 5900 |
| }, |
| { |
| "epoch": 1.6760961810466761, |
| "grad_norm": 0.6812814474105835, |
| "learning_rate": 4.204813334884353e-05, |
| "loss": 0.0921, |
| "step": 5925 |
| }, |
| { |
| "epoch": 1.6831683168316833, |
| "grad_norm": 0.3964853584766388, |
| "learning_rate": 4.19576421815129e-05, |
| "loss": 0.0786, |
| "step": 5950 |
| }, |
| { |
| "epoch": 1.69024045261669, |
| "grad_norm": 0.4422667324542999, |
| "learning_rate": 4.186673761673989e-05, |
| "loss": 0.0752, |
| "step": 5975 |
| }, |
| { |
| "epoch": 1.6973125884016973, |
| "grad_norm": 0.5387901663780212, |
| "learning_rate": 4.177542187061763e-05, |
| "loss": 0.0829, |
| "step": 6000 |
| }, |
| { |
| "epoch": 1.7043847241867045, |
| "grad_norm": 0.30730950832366943, |
| "learning_rate": 4.168369716926315e-05, |
| "loss": 0.0782, |
| "step": 6025 |
| }, |
| { |
| "epoch": 1.7114568599717115, |
| "grad_norm": 0.3543797731399536, |
| "learning_rate": 4.159156574876306e-05, |
| "loss": 0.0842, |
| "step": 6050 |
| }, |
| { |
| "epoch": 1.7185289957567185, |
| "grad_norm": 0.44837215542793274, |
| "learning_rate": 4.1499029855119106e-05, |
| "loss": 0.0754, |
| "step": 6075 |
| }, |
| { |
| "epoch": 1.7256011315417257, |
| "grad_norm": 0.4181993007659912, |
| "learning_rate": 4.1406091744193354e-05, |
| "loss": 0.0833, |
| "step": 6100 |
| }, |
| { |
| "epoch": 1.7326732673267327, |
| "grad_norm": 0.47836604714393616, |
| "learning_rate": 4.1312753681653213e-05, |
| "loss": 0.0819, |
| "step": 6125 |
| }, |
| { |
| "epoch": 1.7397454031117396, |
| "grad_norm": 0.42279157042503357, |
| "learning_rate": 4.121901794291622e-05, |
| "loss": 0.0748, |
| "step": 6150 |
| }, |
| { |
| "epoch": 1.7468175388967468, |
| "grad_norm": 0.40630263090133667, |
| "learning_rate": 4.112488681309455e-05, |
| "loss": 0.0812, |
| "step": 6175 |
| }, |
| { |
| "epoch": 1.753889674681754, |
| "grad_norm": 0.414064884185791, |
| "learning_rate": 4.103036258693932e-05, |
| "loss": 0.0815, |
| "step": 6200 |
| }, |
| { |
| "epoch": 1.7609618104667608, |
| "grad_norm": 0.3953958749771118, |
| "learning_rate": 4.0935447568784646e-05, |
| "loss": 0.0839, |
| "step": 6225 |
| }, |
| { |
| "epoch": 1.768033946251768, |
| "grad_norm": 0.4421181082725525, |
| "learning_rate": 4.084014407249145e-05, |
| "loss": 0.0784, |
| "step": 6250 |
| }, |
| { |
| "epoch": 1.7751060820367752, |
| "grad_norm": 0.6101165413856506, |
| "learning_rate": 4.0744454421391095e-05, |
| "loss": 0.084, |
| "step": 6275 |
| }, |
| { |
| "epoch": 1.7821782178217822, |
| "grad_norm": 0.8179222941398621, |
| "learning_rate": 4.064838094822868e-05, |
| "loss": 0.0792, |
| "step": 6300 |
| }, |
| { |
| "epoch": 1.7892503536067892, |
| "grad_norm": 0.40432754158973694, |
| "learning_rate": 4.055192599510624e-05, |
| "loss": 0.0769, |
| "step": 6325 |
| }, |
| { |
| "epoch": 1.7963224893917964, |
| "grad_norm": 0.36589494347572327, |
| "learning_rate": 4.0455091913425606e-05, |
| "loss": 0.0783, |
| "step": 6350 |
| }, |
| { |
| "epoch": 1.8033946251768034, |
| "grad_norm": 0.3545513451099396, |
| "learning_rate": 4.0357881063831105e-05, |
| "loss": 0.0723, |
| "step": 6375 |
| }, |
| { |
| "epoch": 1.8104667609618104, |
| "grad_norm": 0.5380228161811829, |
| "learning_rate": 4.026029581615198e-05, |
| "loss": 0.0771, |
| "step": 6400 |
| }, |
| { |
| "epoch": 1.8175388967468176, |
| "grad_norm": 0.6094365119934082, |
| "learning_rate": 4.016233854934468e-05, |
| "loss": 0.084, |
| "step": 6425 |
| }, |
| { |
| "epoch": 1.8246110325318248, |
| "grad_norm": 0.6324317455291748, |
| "learning_rate": 4.0064011651434776e-05, |
| "loss": 0.0794, |
| "step": 6450 |
| }, |
| { |
| "epoch": 1.8316831683168315, |
| "grad_norm": 0.2819110155105591, |
| "learning_rate": 3.996531751945883e-05, |
| "loss": 0.0853, |
| "step": 6475 |
| }, |
| { |
| "epoch": 1.8387553041018387, |
| "grad_norm": 0.3535131514072418, |
| "learning_rate": 3.986625855940591e-05, |
| "loss": 0.0712, |
| "step": 6500 |
| }, |
| { |
| "epoch": 1.845827439886846, |
| "grad_norm": 0.32027262449264526, |
| "learning_rate": 3.9766837186158944e-05, |
| "loss": 0.0751, |
| "step": 6525 |
| }, |
| { |
| "epoch": 1.852899575671853, |
| "grad_norm": 0.4550008475780487, |
| "learning_rate": 3.9667055823435885e-05, |
| "loss": 0.0744, |
| "step": 6550 |
| }, |
| { |
| "epoch": 1.85997171145686, |
| "grad_norm": 0.30715903639793396, |
| "learning_rate": 3.956691690373055e-05, |
| "loss": 0.0745, |
| "step": 6575 |
| }, |
| { |
| "epoch": 1.8670438472418671, |
| "grad_norm": 0.5177515745162964, |
| "learning_rate": 3.946642286825339e-05, |
| "loss": 0.0715, |
| "step": 6600 |
| }, |
| { |
| "epoch": 1.874115983026874, |
| "grad_norm": 0.5210005640983582, |
| "learning_rate": 3.9365576166871964e-05, |
| "loss": 0.0696, |
| "step": 6625 |
| }, |
| { |
| "epoch": 1.881188118811881, |
| "grad_norm": 0.28931137919425964, |
| "learning_rate": 3.926437925805118e-05, |
| "loss": 0.0749, |
| "step": 6650 |
| }, |
| { |
| "epoch": 1.8882602545968883, |
| "grad_norm": 0.5680564641952515, |
| "learning_rate": 3.916283460879341e-05, |
| "loss": 0.074, |
| "step": 6675 |
| }, |
| { |
| "epoch": 1.8953323903818955, |
| "grad_norm": 0.44310295581817627, |
| "learning_rate": 3.90609446945783e-05, |
| "loss": 0.0805, |
| "step": 6700 |
| }, |
| { |
| "epoch": 1.9024045261669023, |
| "grad_norm": 0.6425299644470215, |
| "learning_rate": 3.8958711999302475e-05, |
| "loss": 0.0842, |
| "step": 6725 |
| }, |
| { |
| "epoch": 1.9094766619519095, |
| "grad_norm": 0.300752729177475, |
| "learning_rate": 3.885613901521893e-05, |
| "loss": 0.0698, |
| "step": 6750 |
| }, |
| { |
| "epoch": 1.9165487977369167, |
| "grad_norm": 0.43727684020996094, |
| "learning_rate": 3.875322824287633e-05, |
| "loss": 0.0767, |
| "step": 6775 |
| }, |
| { |
| "epoch": 1.9236209335219236, |
| "grad_norm": 0.41746985912323, |
| "learning_rate": 3.8649982191058e-05, |
| "loss": 0.0795, |
| "step": 6800 |
| }, |
| { |
| "epoch": 1.9306930693069306, |
| "grad_norm": 0.35224172472953796, |
| "learning_rate": 3.8546403376720786e-05, |
| "loss": 0.0765, |
| "step": 6825 |
| }, |
| { |
| "epoch": 1.9377652050919378, |
| "grad_norm": 0.295906662940979, |
| "learning_rate": 3.8442494324933724e-05, |
| "loss": 0.076, |
| "step": 6850 |
| }, |
| { |
| "epoch": 1.9448373408769448, |
| "grad_norm": 0.2558325529098511, |
| "learning_rate": 3.833825756881645e-05, |
| "loss": 0.0807, |
| "step": 6875 |
| }, |
| { |
| "epoch": 1.9519094766619518, |
| "grad_norm": 0.3434934616088867, |
| "learning_rate": 3.8233695649477426e-05, |
| "loss": 0.0686, |
| "step": 6900 |
| }, |
| { |
| "epoch": 1.958981612446959, |
| "grad_norm": 0.23843282461166382, |
| "learning_rate": 3.8128811115952044e-05, |
| "loss": 0.0795, |
| "step": 6925 |
| }, |
| { |
| "epoch": 1.9660537482319662, |
| "grad_norm": 0.7551613450050354, |
| "learning_rate": 3.802360652514047e-05, |
| "loss": 0.0681, |
| "step": 6950 |
| }, |
| { |
| "epoch": 1.973125884016973, |
| "grad_norm": 0.4647892713546753, |
| "learning_rate": 3.7918084441745275e-05, |
| "loss": 0.0798, |
| "step": 6975 |
| }, |
| { |
| "epoch": 1.9801980198019802, |
| "grad_norm": 0.3910640478134155, |
| "learning_rate": 3.781224743820897e-05, |
| "loss": 0.077, |
| "step": 7000 |
| }, |
| { |
| "epoch": 1.9872701555869874, |
| "grad_norm": 0.3788031339645386, |
| "learning_rate": 3.770609809465124e-05, |
| "loss": 0.0805, |
| "step": 7025 |
| }, |
| { |
| "epoch": 1.9943422913719944, |
| "grad_norm": 0.6797524690628052, |
| "learning_rate": 3.759963899880609e-05, |
| "loss": 0.0726, |
| "step": 7050 |
| }, |
| { |
| "epoch": 2.0, |
| "eval_accuracy": 0.9707863671603164, |
| "eval_f1": 0.8914057811793163, |
| "eval_loss": 0.08405517786741257, |
| "eval_precision": 0.8863308401537767, |
| "eval_recall": 0.8965391729373575, |
| "eval_runtime": 54.724, |
| "eval_samples_per_second": 918.646, |
| "eval_steps_per_second": 3.6, |
| "step": 7070 |
| }, |
| { |
| "epoch": 2.0014144271570014, |
| "grad_norm": 0.22099077701568604, |
| "learning_rate": 3.749287274595872e-05, |
| "loss": 0.0755, |
| "step": 7075 |
| }, |
| { |
| "epoch": 2.0084865629420086, |
| "grad_norm": 0.4921034872531891, |
| "learning_rate": 3.738580193888228e-05, |
| "loss": 0.0501, |
| "step": 7100 |
| }, |
| { |
| "epoch": 2.0155586987270158, |
| "grad_norm": 0.23433761298656464, |
| "learning_rate": 3.7278429187774436e-05, |
| "loss": 0.0501, |
| "step": 7125 |
| }, |
| { |
| "epoch": 2.0226308345120225, |
| "grad_norm": 0.4263274371623993, |
| "learning_rate": 3.7170757110193697e-05, |
| "loss": 0.0513, |
| "step": 7150 |
| }, |
| { |
| "epoch": 2.0297029702970297, |
| "grad_norm": 0.2480274736881256, |
| "learning_rate": 3.7062788330995635e-05, |
| "loss": 0.0509, |
| "step": 7175 |
| }, |
| { |
| "epoch": 2.036775106082037, |
| "grad_norm": 0.5820468664169312, |
| "learning_rate": 3.695452548226887e-05, |
| "loss": 0.0573, |
| "step": 7200 |
| }, |
| { |
| "epoch": 2.0438472418670437, |
| "grad_norm": 0.42300423979759216, |
| "learning_rate": 3.684597120327094e-05, |
| "loss": 0.0558, |
| "step": 7225 |
| }, |
| { |
| "epoch": 2.050919377652051, |
| "grad_norm": 0.41470280289649963, |
| "learning_rate": 3.673712814036392e-05, |
| "loss": 0.0519, |
| "step": 7250 |
| }, |
| { |
| "epoch": 2.057991513437058, |
| "grad_norm": 0.2917075753211975, |
| "learning_rate": 3.662799894694995e-05, |
| "loss": 0.0518, |
| "step": 7275 |
| }, |
| { |
| "epoch": 2.065063649222065, |
| "grad_norm": 0.41207775473594666, |
| "learning_rate": 3.651858628340649e-05, |
| "loss": 0.0474, |
| "step": 7300 |
| }, |
| { |
| "epoch": 2.072135785007072, |
| "grad_norm": 0.31983888149261475, |
| "learning_rate": 3.640889281702155e-05, |
| "loss": 0.0583, |
| "step": 7325 |
| }, |
| { |
| "epoch": 2.0792079207920793, |
| "grad_norm": 0.65561842918396, |
| "learning_rate": 3.629892122192859e-05, |
| "loss": 0.057, |
| "step": 7350 |
| }, |
| { |
| "epoch": 2.0862800565770865, |
| "grad_norm": 1.253188133239746, |
| "learning_rate": 3.6188674179041356e-05, |
| "loss": 0.0524, |
| "step": 7375 |
| }, |
| { |
| "epoch": 2.0933521923620932, |
| "grad_norm": 0.34525951743125916, |
| "learning_rate": 3.607815437598853e-05, |
| "loss": 0.0577, |
| "step": 7400 |
| }, |
| { |
| "epoch": 2.1004243281471005, |
| "grad_norm": 0.5648230910301208, |
| "learning_rate": 3.5967364507048226e-05, |
| "loss": 0.0564, |
| "step": 7425 |
| }, |
| { |
| "epoch": 2.1074964639321077, |
| "grad_norm": 0.2504655420780182, |
| "learning_rate": 3.585630727308227e-05, |
| "loss": 0.0476, |
| "step": 7450 |
| }, |
| { |
| "epoch": 2.1145685997171144, |
| "grad_norm": 0.3219181299209595, |
| "learning_rate": 3.574498538147035e-05, |
| "loss": 0.0538, |
| "step": 7475 |
| }, |
| { |
| "epoch": 2.1216407355021216, |
| "grad_norm": 0.36033445596694946, |
| "learning_rate": 3.563340154604411e-05, |
| "loss": 0.0576, |
| "step": 7500 |
| }, |
| { |
| "epoch": 2.128712871287129, |
| "grad_norm": 0.3835320472717285, |
| "learning_rate": 3.552155848702086e-05, |
| "loss": 0.0529, |
| "step": 7525 |
| }, |
| { |
| "epoch": 2.1357850070721356, |
| "grad_norm": 0.40226078033447266, |
| "learning_rate": 3.5409458930937346e-05, |
| "loss": 0.0507, |
| "step": 7550 |
| }, |
| { |
| "epoch": 2.142857142857143, |
| "grad_norm": 0.34018775820732117, |
| "learning_rate": 3.5297105610583265e-05, |
| "loss": 0.0514, |
| "step": 7575 |
| }, |
| { |
| "epoch": 2.14992927864215, |
| "grad_norm": 0.7404220700263977, |
| "learning_rate": 3.518450126493464e-05, |
| "loss": 0.0511, |
| "step": 7600 |
| }, |
| { |
| "epoch": 2.157001414427157, |
| "grad_norm": 0.2098894715309143, |
| "learning_rate": 3.507164863908704e-05, |
| "loss": 0.0543, |
| "step": 7625 |
| }, |
| { |
| "epoch": 2.164073550212164, |
| "grad_norm": 0.43998876214027405, |
| "learning_rate": 3.4958550484188675e-05, |
| "loss": 0.0547, |
| "step": 7650 |
| }, |
| { |
| "epoch": 2.171145685997171, |
| "grad_norm": 0.32174915075302124, |
| "learning_rate": 3.4845209557373315e-05, |
| "loss": 0.0528, |
| "step": 7675 |
| }, |
| { |
| "epoch": 2.1782178217821784, |
| "grad_norm": 0.8568967580795288, |
| "learning_rate": 3.4731628621693094e-05, |
| "loss": 0.0511, |
| "step": 7700 |
| }, |
| { |
| "epoch": 2.185289957567185, |
| "grad_norm": 0.39218711853027344, |
| "learning_rate": 3.461781044605112e-05, |
| "loss": 0.0556, |
| "step": 7725 |
| }, |
| { |
| "epoch": 2.1923620933521923, |
| "grad_norm": 0.2832966446876526, |
| "learning_rate": 3.4503757805134004e-05, |
| "loss": 0.0553, |
| "step": 7750 |
| }, |
| { |
| "epoch": 2.1994342291371995, |
| "grad_norm": 0.3094783425331116, |
| "learning_rate": 3.438947347934421e-05, |
| "loss": 0.0482, |
| "step": 7775 |
| }, |
| { |
| "epoch": 2.2065063649222063, |
| "grad_norm": 0.40016797184944153, |
| "learning_rate": 3.427496025473227e-05, |
| "loss": 0.0508, |
| "step": 7800 |
| }, |
| { |
| "epoch": 2.2135785007072135, |
| "grad_norm": 0.3726707398891449, |
| "learning_rate": 3.416022092292887e-05, |
| "loss": 0.0507, |
| "step": 7825 |
| }, |
| { |
| "epoch": 2.2206506364922207, |
| "grad_norm": 0.4792294502258301, |
| "learning_rate": 3.4045258281076766e-05, |
| "loss": 0.0549, |
| "step": 7850 |
| }, |
| { |
| "epoch": 2.227722772277228, |
| "grad_norm": 1.009504795074463, |
| "learning_rate": 3.393007513176265e-05, |
| "loss": 0.0558, |
| "step": 7875 |
| }, |
| { |
| "epoch": 2.2347949080622347, |
| "grad_norm": 0.40913182497024536, |
| "learning_rate": 3.38146742829488e-05, |
| "loss": 0.058, |
| "step": 7900 |
| }, |
| { |
| "epoch": 2.241867043847242, |
| "grad_norm": 0.436496376991272, |
| "learning_rate": 3.369905854790458e-05, |
| "loss": 0.0546, |
| "step": 7925 |
| }, |
| { |
| "epoch": 2.248939179632249, |
| "grad_norm": 0.2754940986633301, |
| "learning_rate": 3.358323074513795e-05, |
| "loss": 0.0498, |
| "step": 7950 |
| }, |
| { |
| "epoch": 2.256011315417256, |
| "grad_norm": 0.7202574610710144, |
| "learning_rate": 3.3467193698326696e-05, |
| "loss": 0.0542, |
| "step": 7975 |
| }, |
| { |
| "epoch": 2.263083451202263, |
| "grad_norm": 0.3375069200992584, |
| "learning_rate": 3.335095023624958e-05, |
| "loss": 0.0501, |
| "step": 8000 |
| }, |
| { |
| "epoch": 2.2701555869872703, |
| "grad_norm": 0.34498855471611023, |
| "learning_rate": 3.323450319271744e-05, |
| "loss": 0.0551, |
| "step": 8025 |
| }, |
| { |
| "epoch": 2.2772277227722775, |
| "grad_norm": 0.427866131067276, |
| "learning_rate": 3.311785540650405e-05, |
| "loss": 0.0517, |
| "step": 8050 |
| }, |
| { |
| "epoch": 2.2842998585572842, |
| "grad_norm": 0.2636614739894867, |
| "learning_rate": 3.300100972127694e-05, |
| "loss": 0.0503, |
| "step": 8075 |
| }, |
| { |
| "epoch": 2.2913719943422914, |
| "grad_norm": 0.34367385506629944, |
| "learning_rate": 3.288396898552807e-05, |
| "loss": 0.0512, |
| "step": 8100 |
| }, |
| { |
| "epoch": 2.298444130127298, |
| "grad_norm": 0.27327075600624084, |
| "learning_rate": 3.2766736052504385e-05, |
| "loss": 0.0539, |
| "step": 8125 |
| }, |
| { |
| "epoch": 2.3055162659123054, |
| "grad_norm": 0.3980356454849243, |
| "learning_rate": 3.264931378013827e-05, |
| "loss": 0.0583, |
| "step": 8150 |
| }, |
| { |
| "epoch": 2.3125884016973126, |
| "grad_norm": 0.24006566405296326, |
| "learning_rate": 3.2531705030977864e-05, |
| "loss": 0.0511, |
| "step": 8175 |
| }, |
| { |
| "epoch": 2.31966053748232, |
| "grad_norm": 0.5592175722122192, |
| "learning_rate": 3.241391267211729e-05, |
| "loss": 0.0502, |
| "step": 8200 |
| }, |
| { |
| "epoch": 2.3267326732673266, |
| "grad_norm": 1.0215811729431152, |
| "learning_rate": 3.229593957512672e-05, |
| "loss": 0.0554, |
| "step": 8225 |
| }, |
| { |
| "epoch": 2.333804809052334, |
| "grad_norm": 0.27189400792121887, |
| "learning_rate": 3.2177788615982454e-05, |
| "loss": 0.0501, |
| "step": 8250 |
| }, |
| { |
| "epoch": 2.340876944837341, |
| "grad_norm": 0.3555835485458374, |
| "learning_rate": 3.205946267499672e-05, |
| "loss": 0.0563, |
| "step": 8275 |
| }, |
| { |
| "epoch": 2.3479490806223478, |
| "grad_norm": 0.36138617992401123, |
| "learning_rate": 3.19409646367475e-05, |
| "loss": 0.0567, |
| "step": 8300 |
| }, |
| { |
| "epoch": 2.355021216407355, |
| "grad_norm": 0.8795189261436462, |
| "learning_rate": 3.182229739000822e-05, |
| "loss": 0.0599, |
| "step": 8325 |
| }, |
| { |
| "epoch": 2.362093352192362, |
| "grad_norm": 0.26847535371780396, |
| "learning_rate": 3.170346382767731e-05, |
| "loss": 0.0496, |
| "step": 8350 |
| }, |
| { |
| "epoch": 2.3691654879773694, |
| "grad_norm": 0.29600900411605835, |
| "learning_rate": 3.158446684670765e-05, |
| "loss": 0.0485, |
| "step": 8375 |
| }, |
| { |
| "epoch": 2.376237623762376, |
| "grad_norm": 0.2916370630264282, |
| "learning_rate": 3.1465309348036016e-05, |
| "loss": 0.0499, |
| "step": 8400 |
| }, |
| { |
| "epoch": 2.3833097595473833, |
| "grad_norm": 0.5006932616233826, |
| "learning_rate": 3.1345994236512305e-05, |
| "loss": 0.0485, |
| "step": 8425 |
| }, |
| { |
| "epoch": 2.3903818953323905, |
| "grad_norm": 0.3413617014884949, |
| "learning_rate": 3.122652442082874e-05, |
| "loss": 0.0506, |
| "step": 8450 |
| }, |
| { |
| "epoch": 2.3974540311173973, |
| "grad_norm": 0.3974776268005371, |
| "learning_rate": 3.1106902813448956e-05, |
| "loss": 0.05, |
| "step": 8475 |
| }, |
| { |
| "epoch": 2.4045261669024045, |
| "grad_norm": 0.3416425883769989, |
| "learning_rate": 3.098713233053702e-05, |
| "loss": 0.0532, |
| "step": 8500 |
| }, |
| { |
| "epoch": 2.4115983026874117, |
| "grad_norm": 0.5350253582000732, |
| "learning_rate": 3.086721589188628e-05, |
| "loss": 0.0505, |
| "step": 8525 |
| }, |
| { |
| "epoch": 2.418670438472419, |
| "grad_norm": 0.6094595789909363, |
| "learning_rate": 3.0747156420848275e-05, |
| "loss": 0.0488, |
| "step": 8550 |
| }, |
| { |
| "epoch": 2.4257425742574257, |
| "grad_norm": 0.2579383850097656, |
| "learning_rate": 3.0626956844261404e-05, |
| "loss": 0.0531, |
| "step": 8575 |
| }, |
| { |
| "epoch": 2.432814710042433, |
| "grad_norm": 0.3322567939758301, |
| "learning_rate": 3.050662009237959e-05, |
| "loss": 0.0521, |
| "step": 8600 |
| }, |
| { |
| "epoch": 2.4398868458274396, |
| "grad_norm": 0.2817617654800415, |
| "learning_rate": 3.038614909880084e-05, |
| "loss": 0.0583, |
| "step": 8625 |
| }, |
| { |
| "epoch": 2.446958981612447, |
| "grad_norm": 0.36568090319633484, |
| "learning_rate": 3.026554680039575e-05, |
| "loss": 0.0505, |
| "step": 8650 |
| }, |
| { |
| "epoch": 2.454031117397454, |
| "grad_norm": 0.32663673162460327, |
| "learning_rate": 3.0144816137235892e-05, |
| "loss": 0.0521, |
| "step": 8675 |
| }, |
| { |
| "epoch": 2.4611032531824613, |
| "grad_norm": 0.4173208475112915, |
| "learning_rate": 3.0023960052522148e-05, |
| "loss": 0.0586, |
| "step": 8700 |
| }, |
| { |
| "epoch": 2.468175388967468, |
| "grad_norm": 0.2725406289100647, |
| "learning_rate": 2.9902981492512945e-05, |
| "loss": 0.0484, |
| "step": 8725 |
| }, |
| { |
| "epoch": 2.4752475247524752, |
| "grad_norm": 0.4088152348995209, |
| "learning_rate": 2.9781883406452453e-05, |
| "loss": 0.0477, |
| "step": 8750 |
| }, |
| { |
| "epoch": 2.4823196605374824, |
| "grad_norm": 0.35629716515541077, |
| "learning_rate": 2.966066874649869e-05, |
| "loss": 0.0484, |
| "step": 8775 |
| }, |
| { |
| "epoch": 2.489391796322489, |
| "grad_norm": 0.4508310854434967, |
| "learning_rate": 2.9539340467651494e-05, |
| "loss": 0.0525, |
| "step": 8800 |
| }, |
| { |
| "epoch": 2.4964639321074964, |
| "grad_norm": 0.49331003427505493, |
| "learning_rate": 2.9417901527680582e-05, |
| "loss": 0.049, |
| "step": 8825 |
| }, |
| { |
| "epoch": 2.5035360678925036, |
| "grad_norm": 0.3239150941371918, |
| "learning_rate": 2.9296354887053367e-05, |
| "loss": 0.0522, |
| "step": 8850 |
| }, |
| { |
| "epoch": 2.510608203677511, |
| "grad_norm": 0.2788864076137543, |
| "learning_rate": 2.9174703508862828e-05, |
| "loss": 0.0549, |
| "step": 8875 |
| }, |
| { |
| "epoch": 2.5176803394625176, |
| "grad_norm": 0.3619994819164276, |
| "learning_rate": 2.9057822401556116e-05, |
| "loss": 0.0525, |
| "step": 8900 |
| }, |
| { |
| "epoch": 2.5247524752475248, |
| "grad_norm": 0.25908371806144714, |
| "learning_rate": 2.89359743427919e-05, |
| "loss": 0.057, |
| "step": 8925 |
| }, |
| { |
| "epoch": 2.531824611032532, |
| "grad_norm": 0.3243955373764038, |
| "learning_rate": 2.8814030331907134e-05, |
| "loss": 0.052, |
| "step": 8950 |
| }, |
| { |
| "epoch": 2.5388967468175387, |
| "grad_norm": 0.3788680136203766, |
| "learning_rate": 2.8691993341681998e-05, |
| "loss": 0.0526, |
| "step": 8975 |
| }, |
| { |
| "epoch": 2.545968882602546, |
| "grad_norm": 0.27258631587028503, |
| "learning_rate": 2.856986634716332e-05, |
| "loss": 0.0555, |
| "step": 9000 |
| }, |
| { |
| "epoch": 2.553041018387553, |
| "grad_norm": 1.2522977590560913, |
| "learning_rate": 2.8447652325592066e-05, |
| "loss": 0.0574, |
| "step": 9025 |
| }, |
| { |
| "epoch": 2.5601131541725604, |
| "grad_norm": 0.23333418369293213, |
| "learning_rate": 2.8325354256330787e-05, |
| "loss": 0.0531, |
| "step": 9050 |
| }, |
| { |
| "epoch": 2.567185289957567, |
| "grad_norm": 0.32750052213668823, |
| "learning_rate": 2.8202975120790963e-05, |
| "loss": 0.0505, |
| "step": 9075 |
| }, |
| { |
| "epoch": 2.5742574257425743, |
| "grad_norm": 0.3817848861217499, |
| "learning_rate": 2.8080517902360316e-05, |
| "loss": 0.0533, |
| "step": 9100 |
| }, |
| { |
| "epoch": 2.581329561527581, |
| "grad_norm": 1.9594416618347168, |
| "learning_rate": 2.7957985586330106e-05, |
| "loss": 0.0499, |
| "step": 9125 |
| }, |
| { |
| "epoch": 2.5884016973125883, |
| "grad_norm": 0.3222751021385193, |
| "learning_rate": 2.7835381159822336e-05, |
| "loss": 0.0553, |
| "step": 9150 |
| }, |
| { |
| "epoch": 2.5954738330975955, |
| "grad_norm": 0.3441489040851593, |
| "learning_rate": 2.7712707611716938e-05, |
| "loss": 0.0533, |
| "step": 9175 |
| }, |
| { |
| "epoch": 2.6025459688826027, |
| "grad_norm": 0.39582306146621704, |
| "learning_rate": 2.75899679325789e-05, |
| "loss": 0.0516, |
| "step": 9200 |
| }, |
| { |
| "epoch": 2.6096181046676095, |
| "grad_norm": 0.32057181000709534, |
| "learning_rate": 2.7467165114585358e-05, |
| "loss": 0.0573, |
| "step": 9225 |
| }, |
| { |
| "epoch": 2.6166902404526167, |
| "grad_norm": 0.49784281849861145, |
| "learning_rate": 2.734430215145269e-05, |
| "loss": 0.0508, |
| "step": 9250 |
| }, |
| { |
| "epoch": 2.623762376237624, |
| "grad_norm": 0.3396826386451721, |
| "learning_rate": 2.7221382038363492e-05, |
| "loss": 0.053, |
| "step": 9275 |
| }, |
| { |
| "epoch": 2.6308345120226306, |
| "grad_norm": 0.30675604939460754, |
| "learning_rate": 2.7098407771893563e-05, |
| "loss": 0.0527, |
| "step": 9300 |
| }, |
| { |
| "epoch": 2.637906647807638, |
| "grad_norm": 0.7486620545387268, |
| "learning_rate": 2.6975382349938892e-05, |
| "loss": 0.0466, |
| "step": 9325 |
| }, |
| { |
| "epoch": 2.644978783592645, |
| "grad_norm": 0.2760941684246063, |
| "learning_rate": 2.685230877164253e-05, |
| "loss": 0.0544, |
| "step": 9350 |
| }, |
| { |
| "epoch": 2.6520509193776522, |
| "grad_norm": 0.4320007264614105, |
| "learning_rate": 2.672919003732149e-05, |
| "loss": 0.0516, |
| "step": 9375 |
| }, |
| { |
| "epoch": 2.659123055162659, |
| "grad_norm": 0.390225350856781, |
| "learning_rate": 2.6606029148393625e-05, |
| "loss": 0.0491, |
| "step": 9400 |
| }, |
| { |
| "epoch": 2.666195190947666, |
| "grad_norm": 0.3701136112213135, |
| "learning_rate": 2.648282910730444e-05, |
| "loss": 0.0531, |
| "step": 9425 |
| }, |
| { |
| "epoch": 2.6732673267326734, |
| "grad_norm": 0.554530918598175, |
| "learning_rate": 2.6359592917453897e-05, |
| "loss": 0.0514, |
| "step": 9450 |
| }, |
| { |
| "epoch": 2.68033946251768, |
| "grad_norm": 0.7218428254127502, |
| "learning_rate": 2.623632358312319e-05, |
| "loss": 0.0531, |
| "step": 9475 |
| }, |
| { |
| "epoch": 2.6874115983026874, |
| "grad_norm": 0.4088691174983978, |
| "learning_rate": 2.6113024109401547e-05, |
| "loss": 0.0517, |
| "step": 9500 |
| }, |
| { |
| "epoch": 2.6944837340876946, |
| "grad_norm": 0.2585553228855133, |
| "learning_rate": 2.5989697502112903e-05, |
| "loss": 0.0495, |
| "step": 9525 |
| }, |
| { |
| "epoch": 2.701555869872702, |
| "grad_norm": 0.3183114528656006, |
| "learning_rate": 2.58663467677427e-05, |
| "loss": 0.0488, |
| "step": 9550 |
| }, |
| { |
| "epoch": 2.7086280056577086, |
| "grad_norm": 0.35523074865341187, |
| "learning_rate": 2.574297491336452e-05, |
| "loss": 0.0507, |
| "step": 9575 |
| }, |
| { |
| "epoch": 2.7157001414427158, |
| "grad_norm": 0.3662826120853424, |
| "learning_rate": 2.5619584946566844e-05, |
| "loss": 0.0492, |
| "step": 9600 |
| }, |
| { |
| "epoch": 2.7227722772277225, |
| "grad_norm": 0.5980022549629211, |
| "learning_rate": 2.549617987537968e-05, |
| "loss": 0.056, |
| "step": 9625 |
| }, |
| { |
| "epoch": 2.7298444130127297, |
| "grad_norm": 0.3375484347343445, |
| "learning_rate": 2.537276270820127e-05, |
| "loss": 0.0545, |
| "step": 9650 |
| }, |
| { |
| "epoch": 2.736916548797737, |
| "grad_norm": 0.2702608108520508, |
| "learning_rate": 2.524933645372472e-05, |
| "loss": 0.0483, |
| "step": 9675 |
| }, |
| { |
| "epoch": 2.743988684582744, |
| "grad_norm": 0.20587654411792755, |
| "learning_rate": 2.512590412086468e-05, |
| "loss": 0.048, |
| "step": 9700 |
| }, |
| { |
| "epoch": 2.751060820367751, |
| "grad_norm": 0.40114825963974, |
| "learning_rate": 2.5002468718683985e-05, |
| "loss": 0.0478, |
| "step": 9725 |
| }, |
| { |
| "epoch": 2.758132956152758, |
| "grad_norm": 0.25009649991989136, |
| "learning_rate": 2.4879033256320268e-05, |
| "loss": 0.0528, |
| "step": 9750 |
| }, |
| { |
| "epoch": 2.7652050919377653, |
| "grad_norm": 0.4038143455982208, |
| "learning_rate": 2.475560074291266e-05, |
| "loss": 0.0536, |
| "step": 9775 |
| }, |
| { |
| "epoch": 2.772277227722772, |
| "grad_norm": 0.3600768744945526, |
| "learning_rate": 2.463217418752838e-05, |
| "loss": 0.0513, |
| "step": 9800 |
| }, |
| { |
| "epoch": 2.7793493635077793, |
| "grad_norm": 0.2662784457206726, |
| "learning_rate": 2.4508756599089423e-05, |
| "loss": 0.0528, |
| "step": 9825 |
| }, |
| { |
| "epoch": 2.7864214992927865, |
| "grad_norm": 0.4926321506500244, |
| "learning_rate": 2.4385350986299162e-05, |
| "loss": 0.0502, |
| "step": 9850 |
| }, |
| { |
| "epoch": 2.7934936350777937, |
| "grad_norm": 0.3003157675266266, |
| "learning_rate": 2.4261960357569036e-05, |
| "loss": 0.0491, |
| "step": 9875 |
| }, |
| { |
| "epoch": 2.8005657708628005, |
| "grad_norm": 0.3073439598083496, |
| "learning_rate": 2.413858772094521e-05, |
| "loss": 0.051, |
| "step": 9900 |
| }, |
| { |
| "epoch": 2.8076379066478077, |
| "grad_norm": 0.4092470109462738, |
| "learning_rate": 2.4015236084035204e-05, |
| "loss": 0.0474, |
| "step": 9925 |
| }, |
| { |
| "epoch": 2.814710042432815, |
| "grad_norm": 0.5402696132659912, |
| "learning_rate": 2.389190845393463e-05, |
| "loss": 0.0492, |
| "step": 9950 |
| }, |
| { |
| "epoch": 2.8217821782178216, |
| "grad_norm": 0.3162129521369934, |
| "learning_rate": 2.376860783715384e-05, |
| "loss": 0.0509, |
| "step": 9975 |
| }, |
| { |
| "epoch": 2.828854314002829, |
| "grad_norm": 0.3184404671192169, |
| "learning_rate": 2.3645337239544658e-05, |
| "loss": 0.0493, |
| "step": 10000 |
| }, |
| { |
| "epoch": 2.835926449787836, |
| "grad_norm": 0.33763962984085083, |
| "learning_rate": 2.3522099666227087e-05, |
| "loss": 0.0489, |
| "step": 10025 |
| }, |
| { |
| "epoch": 2.8429985855728432, |
| "grad_norm": 0.4636186361312866, |
| "learning_rate": 2.339889812151606e-05, |
| "loss": 0.0545, |
| "step": 10050 |
| }, |
| { |
| "epoch": 2.85007072135785, |
| "grad_norm": 0.3926536738872528, |
| "learning_rate": 2.3275735608848208e-05, |
| "loss": 0.0501, |
| "step": 10075 |
| }, |
| { |
| "epoch": 2.857142857142857, |
| "grad_norm": 0.4206957519054413, |
| "learning_rate": 2.3152615130708592e-05, |
| "loss": 0.0502, |
| "step": 10100 |
| }, |
| { |
| "epoch": 2.864214992927864, |
| "grad_norm": 0.729640781879425, |
| "learning_rate": 2.302953968855759e-05, |
| "loss": 0.0469, |
| "step": 10125 |
| }, |
| { |
| "epoch": 2.871287128712871, |
| "grad_norm": 0.37660130858421326, |
| "learning_rate": 2.2906512282757644e-05, |
| "loss": 0.0524, |
| "step": 10150 |
| }, |
| { |
| "epoch": 2.8783592644978784, |
| "grad_norm": 0.5026111602783203, |
| "learning_rate": 2.2783535912500187e-05, |
| "loss": 0.0587, |
| "step": 10175 |
| }, |
| { |
| "epoch": 2.8854314002828856, |
| "grad_norm": 0.3521290123462677, |
| "learning_rate": 2.266061357573248e-05, |
| "loss": 0.0507, |
| "step": 10200 |
| }, |
| { |
| "epoch": 2.8925035360678923, |
| "grad_norm": 0.32747042179107666, |
| "learning_rate": 2.253774826908453e-05, |
| "loss": 0.0531, |
| "step": 10225 |
| }, |
| { |
| "epoch": 2.8995756718528995, |
| "grad_norm": 0.3167445659637451, |
| "learning_rate": 2.2414942987796084e-05, |
| "loss": 0.0468, |
| "step": 10250 |
| }, |
| { |
| "epoch": 2.9066478076379068, |
| "grad_norm": 0.30890071392059326, |
| "learning_rate": 2.2292200725643534e-05, |
| "loss": 0.0535, |
| "step": 10275 |
| }, |
| { |
| "epoch": 2.9137199434229135, |
| "grad_norm": 0.4126497209072113, |
| "learning_rate": 2.216952447486701e-05, |
| "loss": 0.0491, |
| "step": 10300 |
| }, |
| { |
| "epoch": 2.9207920792079207, |
| "grad_norm": 1.0146815776824951, |
| "learning_rate": 2.2046917226097387e-05, |
| "loss": 0.0507, |
| "step": 10325 |
| }, |
| { |
| "epoch": 2.927864214992928, |
| "grad_norm": 0.3795275390148163, |
| "learning_rate": 2.1924381968283394e-05, |
| "loss": 0.0501, |
| "step": 10350 |
| }, |
| { |
| "epoch": 2.934936350777935, |
| "grad_norm": 0.3325127065181732, |
| "learning_rate": 2.1801921688618758e-05, |
| "loss": 0.0438, |
| "step": 10375 |
| }, |
| { |
| "epoch": 2.942008486562942, |
| "grad_norm": 0.358235239982605, |
| "learning_rate": 2.167953937246934e-05, |
| "loss": 0.0519, |
| "step": 10400 |
| }, |
| { |
| "epoch": 2.949080622347949, |
| "grad_norm": 0.37195077538490295, |
| "learning_rate": 2.1557238003300438e-05, |
| "loss": 0.0472, |
| "step": 10425 |
| }, |
| { |
| "epoch": 2.9561527581329563, |
| "grad_norm": 0.3947463929653168, |
| "learning_rate": 2.1435020562603944e-05, |
| "loss": 0.0469, |
| "step": 10450 |
| }, |
| { |
| "epoch": 2.963224893917963, |
| "grad_norm": 0.3517344892024994, |
| "learning_rate": 2.1312890029825765e-05, |
| "loss": 0.0451, |
| "step": 10475 |
| }, |
| { |
| "epoch": 2.9702970297029703, |
| "grad_norm": 0.4573589861392975, |
| "learning_rate": 2.1190849382293153e-05, |
| "loss": 0.0478, |
| "step": 10500 |
| }, |
| { |
| "epoch": 2.9773691654879775, |
| "grad_norm": 0.4348020553588867, |
| "learning_rate": 2.1068901595142076e-05, |
| "loss": 0.0471, |
| "step": 10525 |
| }, |
| { |
| "epoch": 2.9844413012729847, |
| "grad_norm": 0.27447789907455444, |
| "learning_rate": 2.0947049641244763e-05, |
| "loss": 0.0525, |
| "step": 10550 |
| }, |
| { |
| "epoch": 2.9915134370579914, |
| "grad_norm": 0.5210450291633606, |
| "learning_rate": 2.0825296491137178e-05, |
| "loss": 0.049, |
| "step": 10575 |
| }, |
| { |
| "epoch": 2.9985855728429986, |
| "grad_norm": 0.4803789556026459, |
| "learning_rate": 2.0703645112946632e-05, |
| "loss": 0.048, |
| "step": 10600 |
| }, |
| { |
| "epoch": 3.0, |
| "eval_accuracy": 0.9731636716504077, |
| "eval_f1": 0.9040792973909716, |
| "eval_loss": 0.080692358314991, |
| "eval_precision": 0.8956920811279763, |
| "eval_recall": 0.9126250733540829, |
| "eval_runtime": 54.8402, |
| "eval_samples_per_second": 916.701, |
| "eval_steps_per_second": 3.592, |
| "step": 10605 |
| } |
| ], |
| "logging_steps": 25, |
| "max_steps": 17675, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 5, |
| "save_steps": 500, |
| "stateful_callbacks": { |
| "EarlyStoppingCallback": { |
| "args": { |
| "early_stopping_patience": 5, |
| "early_stopping_threshold": 0.01 |
| }, |
| "attributes": { |
| "early_stopping_patience_counter": 1 |
| } |
| }, |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": false |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 3.589143261681162e+17, |
| "train_batch_size": 128, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
