{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 0.2, "eval_steps": 500, "global_step": 4000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "grad_norm": 0.15012216567993164, "learning_rate": 9e-07, "loss": 1.2086, "step": 10 }, { "grad_norm": 0.13072419166564941, "learning_rate": 1.9e-06, "loss": 1.2071, "step": 20 }, { "grad_norm": 0.12423919886350632, "learning_rate": 2.9e-06, "loss": 1.2035, "step": 30 }, { "grad_norm": 0.15493424236774445, "learning_rate": 3.9e-06, "loss": 1.1841, "step": 40 }, { "grad_norm": 0.2636421024799347, "learning_rate": 4.9000000000000005e-06, "loss": 1.1534, "step": 50 }, { "grad_norm": 0.1940675526857376, "learning_rate": 5.9e-06, "loss": 1.1253, "step": 60 }, { "grad_norm": 0.42185041308403015, "learning_rate": 6.900000000000001e-06, "loss": 1.11, "step": 70 }, { "grad_norm": 0.16094836592674255, "learning_rate": 7.9e-06, "loss": 1.1014, "step": 80 }, { "grad_norm": 0.21415859460830688, "learning_rate": 8.9e-06, "loss": 1.0986, "step": 90 }, { "grad_norm": 0.26082614064216614, "learning_rate": 9.900000000000002e-06, "loss": 1.0897, "step": 100 }, { "grad_norm": 0.24920392036437988, "learning_rate": 1.09e-05, "loss": 1.0845, "step": 110 }, { "grad_norm": 0.19736778736114502, "learning_rate": 1.19e-05, "loss": 1.0756, "step": 120 }, { "grad_norm": 0.17664603888988495, "learning_rate": 1.29e-05, "loss": 1.0589, "step": 130 }, { "grad_norm": 0.2617020905017853, "learning_rate": 1.3900000000000002e-05, "loss": 1.0399, "step": 140 }, { "grad_norm": 0.5237306952476501, "learning_rate": 1.49e-05, "loss": 1.0118, "step": 150 }, { "grad_norm": 0.29383063316345215, "learning_rate": 1.59e-05, "loss": 0.9761, "step": 160 }, { "grad_norm": 0.6056665778160095, "learning_rate": 1.69e-05, "loss": 0.9334, "step": 170 }, { "grad_norm": 0.527864396572113, "learning_rate": 1.79e-05, "loss": 0.8957, "step": 180 }, { "grad_norm": 0.7024576663970947, "learning_rate": 1.8900000000000002e-05, "loss": 0.8583, "step": 190 }, { "grad_norm": 0.6996781826019287, "learning_rate": 1.9900000000000003e-05, "loss": 0.8179, "step": 200 }, { "grad_norm": 0.8450202941894531, "learning_rate": 2.09e-05, "loss": 0.7829, "step": 210 }, { "grad_norm": 0.9583392143249512, "learning_rate": 2.19e-05, "loss": 0.745, "step": 220 }, { "grad_norm": 1.7142727375030518, "learning_rate": 2.29e-05, "loss": 0.7135, "step": 230 }, { "grad_norm": 0.8069450855255127, "learning_rate": 2.39e-05, "loss": 0.6912, "step": 240 }, { "grad_norm": 0.8630396723747253, "learning_rate": 2.4900000000000002e-05, "loss": 0.6688, "step": 250 }, { "grad_norm": 1.0455894470214844, "learning_rate": 2.5900000000000003e-05, "loss": 0.6463, "step": 260 }, { "grad_norm": 1.3055671453475952, "learning_rate": 2.6900000000000003e-05, "loss": 0.6241, "step": 270 }, { "grad_norm": 1.0451176166534424, "learning_rate": 2.7900000000000004e-05, "loss": 0.5995, "step": 280 }, { "grad_norm": 0.8477664589881897, "learning_rate": 2.8899999999999998e-05, "loss": 0.5817, "step": 290 }, { "grad_norm": 1.3464720249176025, "learning_rate": 2.9900000000000002e-05, "loss": 0.571, "step": 300 }, { "grad_norm": 0.90577632188797, "learning_rate": 3.09e-05, "loss": 0.5596, "step": 310 }, { "grad_norm": 0.9414107203483582, "learning_rate": 3.19e-05, "loss": 0.5423, "step": 320 }, { "grad_norm": 1.1524189710617065, "learning_rate": 3.29e-05, "loss": 0.5275, "step": 330 }, { "grad_norm": 0.7501125931739807, "learning_rate": 3.3900000000000004e-05, "loss": 0.5178, "step": 340 }, { "grad_norm": 1.073644995689392, "learning_rate": 3.49e-05, "loss": 0.4947, "step": 350 }, { "grad_norm": 1.12371027469635, "learning_rate": 3.59e-05, "loss": 0.4778, "step": 360 }, { "grad_norm": 1.2945274114608765, "learning_rate": 3.69e-05, "loss": 0.4703, "step": 370 }, { "grad_norm": 0.9674573540687561, "learning_rate": 3.79e-05, "loss": 0.4568, "step": 380 }, { "grad_norm": 1.2989366054534912, "learning_rate": 3.8900000000000004e-05, "loss": 0.4433, "step": 390 }, { "grad_norm": 1.2226121425628662, "learning_rate": 3.99e-05, "loss": 0.4272, "step": 400 }, { "grad_norm": 1.2388277053833008, "learning_rate": 4.09e-05, "loss": 0.4054, "step": 410 }, { "grad_norm": 1.370849609375, "learning_rate": 4.19e-05, "loss": 0.3921, "step": 420 }, { "grad_norm": 1.1839030981063843, "learning_rate": 4.29e-05, "loss": 0.3823, "step": 430 }, { "grad_norm": 0.9793280363082886, "learning_rate": 4.39e-05, "loss": 0.362, "step": 440 }, { "grad_norm": 1.08059823513031, "learning_rate": 4.49e-05, "loss": 0.3544, "step": 450 }, { "grad_norm": 1.2476121187210083, "learning_rate": 4.5900000000000004e-05, "loss": 0.3359, "step": 460 }, { "grad_norm": 1.5092262029647827, "learning_rate": 4.69e-05, "loss": 0.3235, "step": 470 }, { "grad_norm": 1.4717587232589722, "learning_rate": 4.79e-05, "loss": 0.31, "step": 480 }, { "grad_norm": 1.3592662811279297, "learning_rate": 4.89e-05, "loss": 0.2975, "step": 490 }, { "grad_norm": 1.1353168487548828, "learning_rate": 4.99e-05, "loss": 0.2877, "step": 500 }, { "grad_norm": 1.9561127424240112, "learning_rate": 5.0900000000000004e-05, "loss": 0.2692, "step": 510 }, { "grad_norm": 1.9443875551223755, "learning_rate": 5.19e-05, "loss": 0.253, "step": 520 }, { "grad_norm": 1.329230546951294, "learning_rate": 5.2900000000000005e-05, "loss": 0.2338, "step": 530 }, { "grad_norm": 1.4158693552017212, "learning_rate": 5.390000000000001e-05, "loss": 0.2161, "step": 540 }, { "grad_norm": 1.59760582447052, "learning_rate": 5.4900000000000006e-05, "loss": 0.2021, "step": 550 }, { "grad_norm": 1.1889493465423584, "learning_rate": 5.590000000000001e-05, "loss": 0.1897, "step": 560 }, { "grad_norm": 1.6009477376937866, "learning_rate": 5.69e-05, "loss": 0.1795, "step": 570 }, { "grad_norm": 1.6399871110916138, "learning_rate": 5.79e-05, "loss": 0.172, "step": 580 }, { "grad_norm": 1.272241473197937, "learning_rate": 5.89e-05, "loss": 0.1638, "step": 590 }, { "grad_norm": 1.4894850254058838, "learning_rate": 5.99e-05, "loss": 0.1546, "step": 600 }, { "grad_norm": 1.1664891242980957, "learning_rate": 6.09e-05, "loss": 0.1439, "step": 610 }, { "grad_norm": 1.2341861724853516, "learning_rate": 6.19e-05, "loss": 0.138, "step": 620 }, { "grad_norm": 1.2932838201522827, "learning_rate": 6.29e-05, "loss": 0.1372, "step": 630 }, { "grad_norm": 1.2948802709579468, "learning_rate": 6.390000000000001e-05, "loss": 0.1345, "step": 640 }, { "grad_norm": 1.1293632984161377, "learning_rate": 6.49e-05, "loss": 0.1281, "step": 650 }, { "grad_norm": 1.3699828386306763, "learning_rate": 6.59e-05, "loss": 0.1278, "step": 660 }, { "grad_norm": 1.3181347846984863, "learning_rate": 6.690000000000001e-05, "loss": 0.1265, "step": 670 }, { "grad_norm": 1.2064861059188843, "learning_rate": 6.790000000000001e-05, "loss": 0.1236, "step": 680 }, { "grad_norm": 1.164475440979004, "learning_rate": 6.89e-05, "loss": 0.1187, "step": 690 }, { "grad_norm": 1.2478479146957397, "learning_rate": 6.99e-05, "loss": 0.1205, "step": 700 }, { "grad_norm": 1.1712687015533447, "learning_rate": 7.09e-05, "loss": 0.1221, "step": 710 }, { "grad_norm": 1.3739711046218872, "learning_rate": 7.19e-05, "loss": 0.1204, "step": 720 }, { "grad_norm": 1.3635663986206055, "learning_rate": 7.29e-05, "loss": 0.1165, "step": 730 }, { "grad_norm": 1.1010574102401733, "learning_rate": 7.390000000000001e-05, "loss": 0.118, "step": 740 }, { "grad_norm": 1.2504950761795044, "learning_rate": 7.49e-05, "loss": 0.113, "step": 750 }, { "grad_norm": 1.144167184829712, "learning_rate": 7.59e-05, "loss": 0.111, "step": 760 }, { "grad_norm": 1.1996899843215942, "learning_rate": 7.69e-05, "loss": 0.1075, "step": 770 }, { "grad_norm": 1.0457687377929688, "learning_rate": 7.790000000000001e-05, "loss": 0.1131, "step": 780 }, { "grad_norm": 1.4280327558517456, "learning_rate": 7.890000000000001e-05, "loss": 0.1115, "step": 790 }, { "grad_norm": 1.16835618019104, "learning_rate": 7.99e-05, "loss": 0.1105, "step": 800 }, { "grad_norm": 1.403944492340088, "learning_rate": 8.090000000000001e-05, "loss": 0.1083, "step": 810 }, { "grad_norm": 1.2651947736740112, "learning_rate": 8.19e-05, "loss": 0.1099, "step": 820 }, { "grad_norm": 1.1956747770309448, "learning_rate": 8.29e-05, "loss": 0.1127, "step": 830 }, { "grad_norm": 1.141771912574768, "learning_rate": 8.39e-05, "loss": 0.1075, "step": 840 }, { "grad_norm": 1.2325164079666138, "learning_rate": 8.49e-05, "loss": 0.1075, "step": 850 }, { "grad_norm": 1.240704894065857, "learning_rate": 8.59e-05, "loss": 0.1038, "step": 860 }, { "grad_norm": 1.3105555772781372, "learning_rate": 8.69e-05, "loss": 0.1065, "step": 870 }, { "grad_norm": 1.3029900789260864, "learning_rate": 8.790000000000001e-05, "loss": 0.1032, "step": 880 }, { "grad_norm": 1.1861451864242554, "learning_rate": 8.89e-05, "loss": 0.1028, "step": 890 }, { "grad_norm": 1.3183419704437256, "learning_rate": 8.99e-05, "loss": 0.1032, "step": 900 }, { "grad_norm": 1.0315393209457397, "learning_rate": 9.090000000000001e-05, "loss": 0.0994, "step": 910 }, { "grad_norm": 1.1772143840789795, "learning_rate": 9.190000000000001e-05, "loss": 0.1014, "step": 920 }, { "grad_norm": 1.1044213771820068, "learning_rate": 9.290000000000001e-05, "loss": 0.0996, "step": 930 }, { "grad_norm": 1.0565844774246216, "learning_rate": 9.39e-05, "loss": 0.0997, "step": 940 }, { "grad_norm": 1.066573143005371, "learning_rate": 9.49e-05, "loss": 0.0996, "step": 950 }, { "grad_norm": 1.405110239982605, "learning_rate": 9.59e-05, "loss": 0.0963, "step": 960 }, { "grad_norm": 1.2910921573638916, "learning_rate": 9.69e-05, "loss": 0.0944, "step": 970 }, { "grad_norm": 1.1059085130691528, "learning_rate": 9.790000000000001e-05, "loss": 0.0967, "step": 980 }, { "grad_norm": 1.2163500785827637, "learning_rate": 9.89e-05, "loss": 0.0947, "step": 990 }, { "grad_norm": 0.8789966106414795, "learning_rate": 9.99e-05, "loss": 0.1027, "step": 1000 }, { "grad_norm": 1.1948117017745972, "learning_rate": 9.999994463727085e-05, "loss": 0.098, "step": 1010 }, { "grad_norm": 1.0235472917556763, "learning_rate": 9.999975326009292e-05, "loss": 0.0924, "step": 1020 }, { "grad_norm": 1.262692928314209, "learning_rate": 9.999942518549879e-05, "loss": 0.0937, "step": 1030 }, { "grad_norm": 1.0914654731750488, "learning_rate": 9.999896041438544e-05, "loss": 0.0937, "step": 1040 }, { "grad_norm": 0.9506348967552185, "learning_rate": 9.999835894802353e-05, "loss": 0.09, "step": 1050 }, { "grad_norm": 1.0387773513793945, "learning_rate": 9.999762078805743e-05, "loss": 0.0891, "step": 1060 }, { "grad_norm": 1.1099857091903687, "learning_rate": 9.999674593650526e-05, "loss": 0.0898, "step": 1070 }, { "grad_norm": 0.9533413648605347, "learning_rate": 9.99957343957588e-05, "loss": 0.09, "step": 1080 }, { "grad_norm": 0.8956478238105774, "learning_rate": 9.99945861685836e-05, "loss": 0.0866, "step": 1090 }, { "grad_norm": 0.8327865600585938, "learning_rate": 9.999330125811884e-05, "loss": 0.0911, "step": 1100 }, { "grad_norm": 1.0309516191482544, "learning_rate": 9.999187966787744e-05, "loss": 0.0888, "step": 1110 }, { "grad_norm": 1.0671669244766235, "learning_rate": 9.999032140174595e-05, "loss": 0.0853, "step": 1120 }, { "grad_norm": 1.0329535007476807, "learning_rate": 9.998862646398464e-05, "loss": 0.0847, "step": 1130 }, { "grad_norm": 1.174817442893982, "learning_rate": 9.998679485922739e-05, "loss": 0.0867, "step": 1140 }, { "grad_norm": 0.9075472950935364, "learning_rate": 9.998482659248174e-05, "loss": 0.0866, "step": 1150 }, { "grad_norm": 0.9608914852142334, "learning_rate": 9.998272166912883e-05, "loss": 0.0853, "step": 1160 }, { "grad_norm": 0.9264180064201355, "learning_rate": 9.998048009492347e-05, "loss": 0.0855, "step": 1170 }, { "grad_norm": 0.8515126705169678, "learning_rate": 9.997810187599403e-05, "loss": 0.0873, "step": 1180 }, { "grad_norm": 1.0085526704788208, "learning_rate": 9.997558701884249e-05, "loss": 0.084, "step": 1190 }, { "grad_norm": 0.8803794384002686, "learning_rate": 9.997293553034433e-05, "loss": 0.0857, "step": 1200 }, { "grad_norm": 0.8993883728981018, "learning_rate": 9.997014741774866e-05, "loss": 0.085, "step": 1210 }, { "grad_norm": 0.9325219988822937, "learning_rate": 9.996722268867803e-05, "loss": 0.08, "step": 1220 }, { "grad_norm": 0.7938389778137207, "learning_rate": 9.996416135112858e-05, "loss": 0.0804, "step": 1230 }, { "grad_norm": 1.0101356506347656, "learning_rate": 9.996096341346988e-05, "loss": 0.082, "step": 1240 }, { "grad_norm": 0.7713646292686462, "learning_rate": 9.995762888444495e-05, "loss": 0.0825, "step": 1250 }, { "grad_norm": 0.8901200890541077, "learning_rate": 9.995415777317027e-05, "loss": 0.0802, "step": 1260 }, { "grad_norm": 0.8909335732460022, "learning_rate": 9.995055008913574e-05, "loss": 0.078, "step": 1270 }, { "grad_norm": 0.8464662432670593, "learning_rate": 9.994680584220463e-05, "loss": 0.0768, "step": 1280 }, { "grad_norm": 0.8395467400550842, "learning_rate": 9.994292504261355e-05, "loss": 0.0803, "step": 1290 }, { "grad_norm": 0.9191763401031494, "learning_rate": 9.993890770097247e-05, "loss": 0.0778, "step": 1300 }, { "grad_norm": 0.9382724165916443, "learning_rate": 9.993475382826467e-05, "loss": 0.0778, "step": 1310 }, { "grad_norm": 0.8167632818222046, "learning_rate": 9.993046343584664e-05, "loss": 0.0768, "step": 1320 }, { "grad_norm": 0.964731752872467, "learning_rate": 9.992603653544816e-05, "loss": 0.0749, "step": 1330 }, { "grad_norm": 0.8039717674255371, "learning_rate": 9.992147313917222e-05, "loss": 0.0731, "step": 1340 }, { "grad_norm": 0.9752305746078491, "learning_rate": 9.991677325949497e-05, "loss": 0.0746, "step": 1350 }, { "grad_norm": 0.7187494039535522, "learning_rate": 9.991193690926568e-05, "loss": 0.0735, "step": 1360 }, { "grad_norm": 0.8025557398796082, "learning_rate": 9.990696410170678e-05, "loss": 0.0732, "step": 1370 }, { "grad_norm": 0.7585386633872986, "learning_rate": 9.990185485041371e-05, "loss": 0.0753, "step": 1380 }, { "grad_norm": 0.8446924090385437, "learning_rate": 9.989660916935498e-05, "loss": 0.0752, "step": 1390 }, { "grad_norm": 0.8321685791015625, "learning_rate": 9.989122707287208e-05, "loss": 0.0774, "step": 1400 }, { "grad_norm": 0.9697508215904236, "learning_rate": 9.988570857567945e-05, "loss": 0.0764, "step": 1410 }, { "grad_norm": 0.8045315742492676, "learning_rate": 9.988005369286446e-05, "loss": 0.0766, "step": 1420 }, { "grad_norm": 0.7006632089614868, "learning_rate": 9.987426243988734e-05, "loss": 0.0729, "step": 1430 }, { "grad_norm": 0.7440106272697449, "learning_rate": 9.986833483258114e-05, "loss": 0.0745, "step": 1440 }, { "grad_norm": 0.8354970812797546, "learning_rate": 9.986227088715173e-05, "loss": 0.0737, "step": 1450 }, { "grad_norm": 0.6917275190353394, "learning_rate": 9.98560706201777e-05, "loss": 0.0712, "step": 1460 }, { "grad_norm": 0.7299842834472656, "learning_rate": 9.984973404861036e-05, "loss": 0.0718, "step": 1470 }, { "grad_norm": 0.7468602657318115, "learning_rate": 9.984326118977361e-05, "loss": 0.0731, "step": 1480 }, { "grad_norm": 0.8825043439865112, "learning_rate": 9.983665206136406e-05, "loss": 0.0738, "step": 1490 }, { "grad_norm": 0.8674785494804382, "learning_rate": 9.982990668145075e-05, "loss": 0.0705, "step": 1500 }, { "grad_norm": 0.8119779229164124, "learning_rate": 9.982302506847534e-05, "loss": 0.0744, "step": 1510 }, { "grad_norm": 0.7920791506767273, "learning_rate": 9.981600724125189e-05, "loss": 0.0719, "step": 1520 }, { "grad_norm": 0.740151047706604, "learning_rate": 9.980885321896685e-05, "loss": 0.0735, "step": 1530 }, { "grad_norm": 0.7403775453567505, "learning_rate": 9.980156302117905e-05, "loss": 0.071, "step": 1540 }, { "grad_norm": 0.8703851699829102, "learning_rate": 9.979413666781963e-05, "loss": 0.0694, "step": 1550 }, { "grad_norm": 0.7278032302856445, "learning_rate": 9.978657417919193e-05, "loss": 0.0708, "step": 1560 }, { "grad_norm": 0.8014971017837524, "learning_rate": 9.977887557597153e-05, "loss": 0.0646, "step": 1570 }, { "grad_norm": 0.7528615593910217, "learning_rate": 9.97710408792061e-05, "loss": 0.0682, "step": 1580 }, { "grad_norm": 0.7048172354698181, "learning_rate": 9.976307011031542e-05, "loss": 0.0694, "step": 1590 }, { "grad_norm": 0.7487463355064392, "learning_rate": 9.975496329109126e-05, "loss": 0.07, "step": 1600 }, { "grad_norm": 0.8928484916687012, "learning_rate": 9.974672044369732e-05, "loss": 0.0736, "step": 1610 }, { "grad_norm": 0.8624300956726074, "learning_rate": 9.97383415906693e-05, "loss": 0.0692, "step": 1620 }, { "grad_norm": 0.7228650450706482, "learning_rate": 9.97298267549146e-05, "loss": 0.0719, "step": 1630 }, { "grad_norm": 0.9539430737495422, "learning_rate": 9.972117595971249e-05, "loss": 0.0692, "step": 1640 }, { "grad_norm": 0.6346696019172668, "learning_rate": 9.971238922871391e-05, "loss": 0.0688, "step": 1650 }, { "grad_norm": 0.7980417013168335, "learning_rate": 9.970346658594142e-05, "loss": 0.0747, "step": 1660 }, { "grad_norm": 0.8892179131507874, "learning_rate": 9.969440805578923e-05, "loss": 0.0692, "step": 1670 }, { "grad_norm": 0.8207274079322815, "learning_rate": 9.968521366302298e-05, "loss": 0.0677, "step": 1680 }, { "grad_norm": 0.8967740535736084, "learning_rate": 9.967588343277981e-05, "loss": 0.071, "step": 1690 }, { "grad_norm": 0.680429995059967, "learning_rate": 9.966641739056818e-05, "loss": 0.0671, "step": 1700 }, { "grad_norm": 0.5896378755569458, "learning_rate": 9.965681556226793e-05, "loss": 0.0695, "step": 1710 }, { "grad_norm": 0.6645721197128296, "learning_rate": 9.964707797413006e-05, "loss": 0.067, "step": 1720 }, { "grad_norm": 0.8198447227478027, "learning_rate": 9.963720465277679e-05, "loss": 0.067, "step": 1730 }, { "grad_norm": 0.7200887203216553, "learning_rate": 9.96271956252014e-05, "loss": 0.0658, "step": 1740 }, { "grad_norm": 0.7971823215484619, "learning_rate": 9.961705091876816e-05, "loss": 0.0649, "step": 1750 }, { "grad_norm": 0.7375441789627075, "learning_rate": 9.960677056121235e-05, "loss": 0.0657, "step": 1760 }, { "grad_norm": 0.7519879341125488, "learning_rate": 9.959635458064005e-05, "loss": 0.0646, "step": 1770 }, { "grad_norm": 0.7177464962005615, "learning_rate": 9.958580300552815e-05, "loss": 0.0659, "step": 1780 }, { "grad_norm": 0.8185701966285706, "learning_rate": 9.957511586472426e-05, "loss": 0.0645, "step": 1790 }, { "grad_norm": 0.7226177453994751, "learning_rate": 9.956429318744662e-05, "loss": 0.0664, "step": 1800 }, { "grad_norm": 0.6147686839103699, "learning_rate": 9.955333500328404e-05, "loss": 0.0663, "step": 1810 }, { "grad_norm": 0.8226826786994934, "learning_rate": 9.95422413421957e-05, "loss": 0.0654, "step": 1820 }, { "grad_norm": 0.6308465600013733, "learning_rate": 9.953101223451133e-05, "loss": 0.0671, "step": 1830 }, { "grad_norm": 0.6037870645523071, "learning_rate": 9.951964771093085e-05, "loss": 0.0652, "step": 1840 }, { "grad_norm": 0.679835319519043, "learning_rate": 9.950814780252442e-05, "loss": 0.067, "step": 1850 }, { "grad_norm": 0.6414503455162048, "learning_rate": 9.949651254073236e-05, "loss": 0.0625, "step": 1860 }, { "grad_norm": 0.6243733167648315, "learning_rate": 9.948474195736504e-05, "loss": 0.0631, "step": 1870 }, { "grad_norm": 0.7260075807571411, "learning_rate": 9.947283608460277e-05, "loss": 0.0616, "step": 1880 }, { "grad_norm": 0.6100785732269287, "learning_rate": 9.946079495499577e-05, "loss": 0.0657, "step": 1890 }, { "grad_norm": 0.6928042769432068, "learning_rate": 9.944861860146401e-05, "loss": 0.0619, "step": 1900 }, { "grad_norm": 0.6407603621482849, "learning_rate": 9.943630705729719e-05, "loss": 0.0636, "step": 1910 }, { "grad_norm": 0.6835041642189026, "learning_rate": 9.942386035615459e-05, "loss": 0.0659, "step": 1920 }, { "grad_norm": 0.674142599105835, "learning_rate": 9.941127853206503e-05, "loss": 0.0635, "step": 1930 }, { "grad_norm": 0.6852268576622009, "learning_rate": 9.939856161942673e-05, "loss": 0.0624, "step": 1940 }, { "grad_norm": 0.6523897647857666, "learning_rate": 9.938570965300724e-05, "loss": 0.0623, "step": 1950 }, { "grad_norm": 0.6527503132820129, "learning_rate": 9.937272266794335e-05, "loss": 0.0611, "step": 1960 }, { "grad_norm": 0.6659324169158936, "learning_rate": 9.935960069974096e-05, "loss": 0.0636, "step": 1970 }, { "grad_norm": 0.6519555449485779, "learning_rate": 9.934634378427506e-05, "loss": 0.0653, "step": 1980 }, { "grad_norm": 0.7410769462585449, "learning_rate": 9.933295195778954e-05, "loss": 0.0667, "step": 1990 }, { "grad_norm": 0.5415292382240295, "learning_rate": 9.931942525689715e-05, "loss": 0.0678, "step": 2000 }, { "grad_norm": 0.5994923710823059, "learning_rate": 9.930576371857936e-05, "loss": 0.0628, "step": 2010 }, { "grad_norm": 0.670676589012146, "learning_rate": 9.929196738018629e-05, "loss": 0.0624, "step": 2020 }, { "grad_norm": 0.5595282912254333, "learning_rate": 9.927803627943662e-05, "loss": 0.0614, "step": 2030 }, { "grad_norm": 0.5828508138656616, "learning_rate": 9.926397045441744e-05, "loss": 0.0604, "step": 2040 }, { "grad_norm": 0.5831713676452637, "learning_rate": 9.924976994358417e-05, "loss": 0.0605, "step": 2050 }, { "grad_norm": 0.6327888369560242, "learning_rate": 9.923543478576048e-05, "loss": 0.0592, "step": 2060 }, { "grad_norm": 0.6445356607437134, "learning_rate": 9.922096502013813e-05, "loss": 0.0605, "step": 2070 }, { "grad_norm": 0.5931155681610107, "learning_rate": 9.92063606862769e-05, "loss": 0.0608, "step": 2080 }, { "grad_norm": 0.601141095161438, "learning_rate": 9.919162182410453e-05, "loss": 0.059, "step": 2090 }, { "grad_norm": 0.559126615524292, "learning_rate": 9.917674847391645e-05, "loss": 0.0571, "step": 2100 }, { "grad_norm": 0.6964930295944214, "learning_rate": 9.916174067637584e-05, "loss": 0.0592, "step": 2110 }, { "grad_norm": 0.7474396824836731, "learning_rate": 9.914659847251348e-05, "loss": 0.0628, "step": 2120 }, { "grad_norm": 0.6328505277633667, "learning_rate": 9.913132190372753e-05, "loss": 0.0616, "step": 2130 }, { "grad_norm": 0.6215328574180603, "learning_rate": 9.911591101178359e-05, "loss": 0.0605, "step": 2140 }, { "grad_norm": 0.8151555061340332, "learning_rate": 9.910036583881443e-05, "loss": 0.0622, "step": 2150 }, { "grad_norm": 0.6360267400741577, "learning_rate": 9.908468642731995e-05, "loss": 0.0633, "step": 2160 }, { "grad_norm": 0.6325832605361938, "learning_rate": 9.906887282016707e-05, "loss": 0.0646, "step": 2170 }, { "grad_norm": 0.616167426109314, "learning_rate": 9.90529250605896e-05, "loss": 0.0586, "step": 2180 }, { "grad_norm": 0.5955585837364197, "learning_rate": 9.903684319218809e-05, "loss": 0.0586, "step": 2190 }, { "grad_norm": 0.712861180305481, "learning_rate": 9.902062725892976e-05, "loss": 0.0616, "step": 2200 }, { "grad_norm": 0.6084709167480469, "learning_rate": 9.900427730514834e-05, "loss": 0.0613, "step": 2210 }, { "grad_norm": 0.5973120331764221, "learning_rate": 9.8987793375544e-05, "loss": 0.0607, "step": 2220 }, { "grad_norm": 0.6146500706672668, "learning_rate": 9.897117551518318e-05, "loss": 0.0608, "step": 2230 }, { "grad_norm": 0.5947669148445129, "learning_rate": 9.895442376949844e-05, "loss": 0.0619, "step": 2240 }, { "grad_norm": 0.6229901909828186, "learning_rate": 9.893753818428845e-05, "loss": 0.0632, "step": 2250 }, { "grad_norm": 0.5743619203567505, "learning_rate": 9.892051880571773e-05, "loss": 0.062, "step": 2260 }, { "grad_norm": 0.7406059503555298, "learning_rate": 9.890336568031663e-05, "loss": 0.0597, "step": 2270 }, { "grad_norm": 0.6080294251441956, "learning_rate": 9.888607885498113e-05, "loss": 0.0575, "step": 2280 }, { "grad_norm": 0.5063527226448059, "learning_rate": 9.886865837697275e-05, "loss": 0.0575, "step": 2290 }, { "grad_norm": 0.5277106165885925, "learning_rate": 9.88511042939184e-05, "loss": 0.0588, "step": 2300 }, { "grad_norm": 0.6051730513572693, "learning_rate": 9.883341665381028e-05, "loss": 0.058, "step": 2310 }, { "grad_norm": 0.5652435421943665, "learning_rate": 9.881559550500575e-05, "loss": 0.0605, "step": 2320 }, { "grad_norm": 0.5267813801765442, "learning_rate": 9.879764089622712e-05, "loss": 0.0569, "step": 2330 }, { "grad_norm": 0.5755409002304077, "learning_rate": 9.87795528765616e-05, "loss": 0.0605, "step": 2340 }, { "grad_norm": 0.6392557621002197, "learning_rate": 9.876133149546118e-05, "loss": 0.0589, "step": 2350 }, { "grad_norm": 0.5705186724662781, "learning_rate": 9.874297680274238e-05, "loss": 0.0583, "step": 2360 }, { "grad_norm": 0.6585530042648315, "learning_rate": 9.872448884858624e-05, "loss": 0.0588, "step": 2370 }, { "grad_norm": 0.6803420782089233, "learning_rate": 9.870586768353815e-05, "loss": 0.0567, "step": 2380 }, { "grad_norm": 0.7465255856513977, "learning_rate": 9.868711335850764e-05, "loss": 0.0561, "step": 2390 }, { "grad_norm": 0.5848642587661743, "learning_rate": 9.866822592476833e-05, "loss": 0.0588, "step": 2400 }, { "grad_norm": 0.5381805300712585, "learning_rate": 9.86492054339577e-05, "loss": 0.0573, "step": 2410 }, { "grad_norm": 0.6431410908699036, "learning_rate": 9.863005193807711e-05, "loss": 0.0564, "step": 2420 }, { "grad_norm": 0.5730747580528259, "learning_rate": 9.861076548949143e-05, "loss": 0.0575, "step": 2430 }, { "grad_norm": 0.5729357600212097, "learning_rate": 9.859134614092912e-05, "loss": 0.0577, "step": 2440 }, { "grad_norm": 0.5879770517349243, "learning_rate": 9.857179394548191e-05, "loss": 0.0571, "step": 2450 }, { "grad_norm": 0.5518467426300049, "learning_rate": 9.855210895660477e-05, "loss": 0.0546, "step": 2460 }, { "grad_norm": 0.6395381689071655, "learning_rate": 9.853229122811568e-05, "loss": 0.0569, "step": 2470 }, { "grad_norm": 0.6454721093177795, "learning_rate": 9.851234081419559e-05, "loss": 0.0587, "step": 2480 }, { "grad_norm": 0.6770858764648438, "learning_rate": 9.849225776938814e-05, "loss": 0.0586, "step": 2490 }, { "grad_norm": 0.6783500909805298, "learning_rate": 9.847204214859964e-05, "loss": 0.056, "step": 2500 }, { "grad_norm": 0.7126483917236328, "learning_rate": 9.845169400709879e-05, "loss": 0.0565, "step": 2510 }, { "grad_norm": 0.5490720272064209, "learning_rate": 9.843121340051664e-05, "loss": 0.0574, "step": 2520 }, { "grad_norm": 0.6126254796981812, "learning_rate": 9.841060038484641e-05, "loss": 0.0569, "step": 2530 }, { "grad_norm": 0.549883246421814, "learning_rate": 9.838985501644328e-05, "loss": 0.0519, "step": 2540 }, { "grad_norm": 0.610352635383606, "learning_rate": 9.83689773520243e-05, "loss": 0.0564, "step": 2550 }, { "grad_norm": 0.588969349861145, "learning_rate": 9.834796744866819e-05, "loss": 0.0544, "step": 2560 }, { "grad_norm": 0.6367093920707703, "learning_rate": 9.832682536381525e-05, "loss": 0.056, "step": 2570 }, { "grad_norm": 0.696803092956543, "learning_rate": 9.830555115526711e-05, "loss": 0.0565, "step": 2580 }, { "grad_norm": 0.67190021276474, "learning_rate": 9.828414488118667e-05, "loss": 0.0576, "step": 2590 }, { "grad_norm": 0.5457078814506531, "learning_rate": 9.826260660009785e-05, "loss": 0.0555, "step": 2600 }, { "grad_norm": 0.5594436526298523, "learning_rate": 9.824093637088547e-05, "loss": 0.0528, "step": 2610 }, { "grad_norm": 0.5211487412452698, "learning_rate": 9.821913425279514e-05, "loss": 0.0538, "step": 2620 }, { "grad_norm": 0.5177614092826843, "learning_rate": 9.8197200305433e-05, "loss": 0.0545, "step": 2630 }, { "grad_norm": 0.5182483792304993, "learning_rate": 9.817513458876564e-05, "loss": 0.0556, "step": 2640 }, { "grad_norm": 0.6016985774040222, "learning_rate": 9.815293716311987e-05, "loss": 0.0527, "step": 2650 }, { "grad_norm": 0.5641167759895325, "learning_rate": 9.813060808918262e-05, "loss": 0.0561, "step": 2660 }, { "grad_norm": 0.6382514834403992, "learning_rate": 9.810814742800069e-05, "loss": 0.0544, "step": 2670 }, { "grad_norm": 0.7021539211273193, "learning_rate": 9.808555524098074e-05, "loss": 0.0559, "step": 2680 }, { "grad_norm": 0.5652006268501282, "learning_rate": 9.806283158988887e-05, "loss": 0.0544, "step": 2690 }, { "grad_norm": 0.5768893361091614, "learning_rate": 9.803997653685072e-05, "loss": 0.0516, "step": 2700 }, { "grad_norm": 0.6968298554420471, "learning_rate": 9.801699014435112e-05, "loss": 0.0534, "step": 2710 }, { "grad_norm": 0.618663489818573, "learning_rate": 9.799387247523398e-05, "loss": 0.0535, "step": 2720 }, { "grad_norm": 0.5248475670814514, "learning_rate": 9.797062359270215e-05, "loss": 0.0529, "step": 2730 }, { "grad_norm": 0.6241867542266846, "learning_rate": 9.794724356031715e-05, "loss": 0.0536, "step": 2740 }, { "grad_norm": 0.6985629200935364, "learning_rate": 9.792373244199913e-05, "loss": 0.0558, "step": 2750 }, { "grad_norm": 0.6230672001838684, "learning_rate": 9.790009030202658e-05, "loss": 0.0537, "step": 2760 }, { "grad_norm": 0.5703837871551514, "learning_rate": 9.78763172050362e-05, "loss": 0.0545, "step": 2770 }, { "grad_norm": 0.5615924596786499, "learning_rate": 9.785241321602274e-05, "loss": 0.0535, "step": 2780 }, { "grad_norm": 0.6150487065315247, "learning_rate": 9.782837840033879e-05, "loss": 0.0541, "step": 2790 }, { "grad_norm": 0.6442948579788208, "learning_rate": 9.780421282369461e-05, "loss": 0.0566, "step": 2800 }, { "grad_norm": 0.5317158102989197, "learning_rate": 9.777991655215797e-05, "loss": 0.0564, "step": 2810 }, { "grad_norm": 0.5907642245292664, "learning_rate": 9.775548965215394e-05, "loss": 0.0521, "step": 2820 }, { "grad_norm": 0.5747297406196594, "learning_rate": 9.773093219046474e-05, "loss": 0.0495, "step": 2830 }, { "grad_norm": 0.6061862111091614, "learning_rate": 9.770624423422954e-05, "loss": 0.0549, "step": 2840 }, { "grad_norm": 0.5587713122367859, "learning_rate": 9.768142585094426e-05, "loss": 0.056, "step": 2850 }, { "grad_norm": 0.646105170249939, "learning_rate": 9.765647710846142e-05, "loss": 0.0518, "step": 2860 }, { "grad_norm": 0.5185126066207886, "learning_rate": 9.763139807498991e-05, "loss": 0.0537, "step": 2870 }, { "grad_norm": 0.5357843637466431, "learning_rate": 9.760618881909487e-05, "loss": 0.0534, "step": 2880 }, { "grad_norm": 0.5286585688591003, "learning_rate": 9.758084940969744e-05, "loss": 0.0562, "step": 2890 }, { "grad_norm": 0.5728244781494141, "learning_rate": 9.755537991607459e-05, "loss": 0.0521, "step": 2900 }, { "grad_norm": 0.6184394955635071, "learning_rate": 9.752978040785895e-05, "loss": 0.0574, "step": 2910 }, { "grad_norm": 0.5012248754501343, "learning_rate": 9.750405095503859e-05, "loss": 0.0528, "step": 2920 }, { "grad_norm": 0.5458285212516785, "learning_rate": 9.747819162795686e-05, "loss": 0.0524, "step": 2930 }, { "grad_norm": 0.6669784188270569, "learning_rate": 9.745220249731217e-05, "loss": 0.0554, "step": 2940 }, { "grad_norm": 0.5164629817008972, "learning_rate": 9.742608363415781e-05, "loss": 0.0519, "step": 2950 }, { "grad_norm": 0.5527239441871643, "learning_rate": 9.739983510990176e-05, "loss": 0.0553, "step": 2960 }, { "grad_norm": 0.6122348308563232, "learning_rate": 9.737345699630647e-05, "loss": 0.0532, "step": 2970 }, { "grad_norm": 0.5157252550125122, "learning_rate": 9.734694936548869e-05, "loss": 0.0533, "step": 2980 }, { "grad_norm": 0.5645245313644409, "learning_rate": 9.732031228991932e-05, "loss": 0.0577, "step": 2990 }, { "grad_norm": 0.7678781747817993, "learning_rate": 9.729354584242302e-05, "loss": 0.0547, "step": 3000 }, { "grad_norm": 0.6804419755935669, "learning_rate": 9.726665009617832e-05, "loss": 0.0559, "step": 3010 }, { "grad_norm": 0.5836779475212097, "learning_rate": 9.723962512471714e-05, "loss": 0.0548, "step": 3020 }, { "grad_norm": 0.5465885996818542, "learning_rate": 9.72124710019247e-05, "loss": 0.0562, "step": 3030 }, { "grad_norm": 0.5152290463447571, "learning_rate": 9.718518780203934e-05, "loss": 0.0521, "step": 3040 }, { "grad_norm": 0.5115591883659363, "learning_rate": 9.715777559965228e-05, "loss": 0.0525, "step": 3050 }, { "grad_norm": 0.49400827288627625, "learning_rate": 9.713023446970746e-05, "loss": 0.0521, "step": 3060 }, { "grad_norm": 0.6357924342155457, "learning_rate": 9.710256448750126e-05, "loss": 0.05, "step": 3070 }, { "grad_norm": 0.6152380108833313, "learning_rate": 9.707476572868235e-05, "loss": 0.0542, "step": 3080 }, { "grad_norm": 0.6186644434928894, "learning_rate": 9.704683826925149e-05, "loss": 0.0519, "step": 3090 }, { "grad_norm": 0.5993762612342834, "learning_rate": 9.701878218556129e-05, "loss": 0.0501, "step": 3100 }, { "grad_norm": 0.578052818775177, "learning_rate": 9.699059755431598e-05, "loss": 0.0509, "step": 3110 }, { "grad_norm": 0.5247306227684021, "learning_rate": 9.696228445257132e-05, "loss": 0.051, "step": 3120 }, { "grad_norm": 0.6611071825027466, "learning_rate": 9.693384295773419e-05, "loss": 0.0524, "step": 3130 }, { "grad_norm": 0.5699040293693542, "learning_rate": 9.690527314756259e-05, "loss": 0.0526, "step": 3140 }, { "grad_norm": 0.5631635785102844, "learning_rate": 9.687657510016527e-05, "loss": 0.0507, "step": 3150 }, { "grad_norm": 0.4601803123950958, "learning_rate": 9.684774889400161e-05, "loss": 0.05, "step": 3160 }, { "grad_norm": 0.5025836825370789, "learning_rate": 9.681879460788135e-05, "loss": 0.049, "step": 3170 }, { "grad_norm": 0.5727059245109558, "learning_rate": 9.67897123209644e-05, "loss": 0.0526, "step": 3180 }, { "grad_norm": 0.649380087852478, "learning_rate": 9.676050211276062e-05, "loss": 0.0537, "step": 3190 }, { "grad_norm": 0.5575081706047058, "learning_rate": 9.673116406312962e-05, "loss": 0.0539, "step": 3200 }, { "grad_norm": 0.4866867959499359, "learning_rate": 9.67016982522805e-05, "loss": 0.0487, "step": 3210 }, { "grad_norm": 0.5518966913223267, "learning_rate": 9.667210476077164e-05, "loss": 0.0509, "step": 3220 }, { "grad_norm": 0.5296523571014404, "learning_rate": 9.664238366951055e-05, "loss": 0.0497, "step": 3230 }, { "grad_norm": 0.5967239141464233, "learning_rate": 9.661253505975355e-05, "loss": 0.0524, "step": 3240 }, { "grad_norm": 0.5917206406593323, "learning_rate": 9.658255901310557e-05, "loss": 0.0522, "step": 3250 }, { "grad_norm": 0.5756188035011292, "learning_rate": 9.655245561152e-05, "loss": 0.0504, "step": 3260 }, { "grad_norm": 0.5686755776405334, "learning_rate": 9.65222249372984e-05, "loss": 0.0536, "step": 3270 }, { "grad_norm": 0.6282801628112793, "learning_rate": 9.649186707309026e-05, "loss": 0.052, "step": 3280 }, { "grad_norm": 0.5592830777168274, "learning_rate": 9.646138210189283e-05, "loss": 0.0492, "step": 3290 }, { "grad_norm": 0.6329211592674255, "learning_rate": 9.643077010705087e-05, "loss": 0.0525, "step": 3300 }, { "grad_norm": 0.5022810697555542, "learning_rate": 9.640003117225637e-05, "loss": 0.0514, "step": 3310 }, { "grad_norm": 0.6291565895080566, "learning_rate": 9.636916538154846e-05, "loss": 0.0512, "step": 3320 }, { "grad_norm": 0.491485059261322, "learning_rate": 9.633817281931296e-05, "loss": 0.0524, "step": 3330 }, { "grad_norm": 0.5268076658248901, "learning_rate": 9.630705357028242e-05, "loss": 0.0517, "step": 3340 }, { "grad_norm": 0.5763431787490845, "learning_rate": 9.627580771953563e-05, "loss": 0.0496, "step": 3350 }, { "grad_norm": 0.5327632427215576, "learning_rate": 9.624443535249759e-05, "loss": 0.0538, "step": 3360 }, { "grad_norm": 0.5271499156951904, "learning_rate": 9.621293655493913e-05, "loss": 0.0493, "step": 3370 }, { "grad_norm": 0.5392346978187561, "learning_rate": 9.618131141297675e-05, "loss": 0.048, "step": 3380 }, { "grad_norm": 0.557559072971344, "learning_rate": 9.614956001307242e-05, "loss": 0.0495, "step": 3390 }, { "grad_norm": 0.514748215675354, "learning_rate": 9.611768244203321e-05, "loss": 0.051, "step": 3400 }, { "grad_norm": 0.6630014181137085, "learning_rate": 9.60856787870112e-05, "loss": 0.0477, "step": 3410 }, { "grad_norm": 0.48811548948287964, "learning_rate": 9.605354913550318e-05, "loss": 0.0524, "step": 3420 }, { "grad_norm": 0.47521334886550903, "learning_rate": 9.602129357535037e-05, "loss": 0.0495, "step": 3430 }, { "grad_norm": 0.46967071294784546, "learning_rate": 9.598891219473825e-05, "loss": 0.0519, "step": 3440 }, { "grad_norm": 0.5551022887229919, "learning_rate": 9.595640508219625e-05, "loss": 0.0468, "step": 3450 }, { "grad_norm": 0.5609787106513977, "learning_rate": 9.592377232659761e-05, "loss": 0.0506, "step": 3460 }, { "grad_norm": 0.4883131980895996, "learning_rate": 9.589101401715904e-05, "loss": 0.0498, "step": 3470 }, { "grad_norm": 0.5301612615585327, "learning_rate": 9.585813024344045e-05, "loss": 0.0509, "step": 3480 }, { "grad_norm": 0.5042786002159119, "learning_rate": 9.58251210953449e-05, "loss": 0.0485, "step": 3490 }, { "grad_norm": 0.5025367140769958, "learning_rate": 9.579198666311809e-05, "loss": 0.0498, "step": 3500 }, { "grad_norm": 0.5075255632400513, "learning_rate": 9.575872703734832e-05, "loss": 0.0487, "step": 3510 }, { "grad_norm": 0.43919530510902405, "learning_rate": 9.572534230896611e-05, "loss": 0.0465, "step": 3520 }, { "grad_norm": 0.5095721483230591, "learning_rate": 9.569183256924403e-05, "loss": 0.0487, "step": 3530 }, { "grad_norm": 0.512020468711853, "learning_rate": 9.565819790979646e-05, "loss": 0.0486, "step": 3540 }, { "grad_norm": 0.5008028149604797, "learning_rate": 9.562443842257925e-05, "loss": 0.0511, "step": 3550 }, { "grad_norm": 0.5364949107170105, "learning_rate": 9.559055419988956e-05, "loss": 0.0512, "step": 3560 }, { "grad_norm": 0.5370563864707947, "learning_rate": 9.555654533436557e-05, "loss": 0.0499, "step": 3570 }, { "grad_norm": 0.5545650124549866, "learning_rate": 9.552241191898621e-05, "loss": 0.0493, "step": 3580 }, { "grad_norm": 0.5982073545455933, "learning_rate": 9.548815404707092e-05, "loss": 0.0496, "step": 3590 }, { "grad_norm": 0.5401698350906372, "learning_rate": 9.545377181227942e-05, "loss": 0.0489, "step": 3600 }, { "grad_norm": 0.43389976024627686, "learning_rate": 9.541926530861145e-05, "loss": 0.0464, "step": 3610 }, { "grad_norm": 0.608935534954071, "learning_rate": 9.538463463040645e-05, "loss": 0.0467, "step": 3620 }, { "grad_norm": 0.4709791839122772, "learning_rate": 9.534987987234337e-05, "loss": 0.048, "step": 3630 }, { "grad_norm": 0.4395516812801361, "learning_rate": 9.53150011294404e-05, "loss": 0.0496, "step": 3640 }, { "grad_norm": 0.5766939520835876, "learning_rate": 9.527999849705471e-05, "loss": 0.0462, "step": 3650 }, { "grad_norm": 0.4720335602760315, "learning_rate": 9.524487207088213e-05, "loss": 0.0466, "step": 3660 }, { "grad_norm": 0.5506165027618408, "learning_rate": 9.520962194695698e-05, "loss": 0.0471, "step": 3670 }, { "grad_norm": 0.540799617767334, "learning_rate": 9.517424822165175e-05, "loss": 0.0471, "step": 3680 }, { "grad_norm": 0.49528366327285767, "learning_rate": 9.513875099167685e-05, "loss": 0.0483, "step": 3690 }, { "grad_norm": 0.5733321309089661, "learning_rate": 9.510313035408035e-05, "loss": 0.0471, "step": 3700 }, { "grad_norm": 0.5157331228256226, "learning_rate": 9.506738640624775e-05, "loss": 0.0474, "step": 3710 }, { "grad_norm": 0.5828023552894592, "learning_rate": 9.50315192459016e-05, "loss": 0.0485, "step": 3720 }, { "grad_norm": 0.5534822344779968, "learning_rate": 9.499552897110136e-05, "loss": 0.0497, "step": 3730 }, { "grad_norm": 0.4209916293621063, "learning_rate": 9.495941568024304e-05, "loss": 0.0491, "step": 3740 }, { "grad_norm": 0.4925754964351654, "learning_rate": 9.492317947205904e-05, "loss": 0.048, "step": 3750 }, { "grad_norm": 0.4909178912639618, "learning_rate": 9.488682044561775e-05, "loss": 0.0466, "step": 3760 }, { "grad_norm": 0.5199701189994812, "learning_rate": 9.485033870032335e-05, "loss": 0.0495, "step": 3770 }, { "grad_norm": 0.44645825028419495, "learning_rate": 9.481373433591556e-05, "loss": 0.0493, "step": 3780 }, { "grad_norm": 0.6020624041557312, "learning_rate": 9.47770074524693e-05, "loss": 0.0455, "step": 3790 }, { "grad_norm": 0.488620400428772, "learning_rate": 9.474015815039446e-05, "loss": 0.0491, "step": 3800 }, { "grad_norm": 0.5846852660179138, "learning_rate": 9.470318653043565e-05, "loss": 0.0488, "step": 3810 }, { "grad_norm": 0.518318235874176, "learning_rate": 9.466609269367185e-05, "loss": 0.0487, "step": 3820 }, { "grad_norm": 0.5965957641601562, "learning_rate": 9.46288767415162e-05, "loss": 0.0464, "step": 3830 }, { "grad_norm": 0.5559269189834595, "learning_rate": 9.459153877571567e-05, "loss": 0.0475, "step": 3840 }, { "grad_norm": 0.4260229766368866, "learning_rate": 9.455407889835087e-05, "loss": 0.0488, "step": 3850 }, { "grad_norm": 0.5213231444358826, "learning_rate": 9.451649721183564e-05, "loss": 0.0461, "step": 3860 }, { "grad_norm": 0.4626673460006714, "learning_rate": 9.447879381891692e-05, "loss": 0.0446, "step": 3870 }, { "grad_norm": 0.5997720956802368, "learning_rate": 9.444096882267428e-05, "loss": 0.0494, "step": 3880 }, { "grad_norm": 0.5883204936981201, "learning_rate": 9.440302232651988e-05, "loss": 0.0501, "step": 3890 }, { "grad_norm": 0.4910527765750885, "learning_rate": 9.436495443419795e-05, "loss": 0.0472, "step": 3900 }, { "grad_norm": 0.4996890127658844, "learning_rate": 9.432676524978466e-05, "loss": 0.0455, "step": 3910 }, { "grad_norm": 0.508808970451355, "learning_rate": 9.42884548776878e-05, "loss": 0.0448, "step": 3920 }, { "grad_norm": 0.48599308729171753, "learning_rate": 9.425002342264646e-05, "loss": 0.0444, "step": 3930 }, { "grad_norm": 0.48764660954475403, "learning_rate": 9.421147098973077e-05, "loss": 0.0457, "step": 3940 }, { "grad_norm": 0.5945606231689453, "learning_rate": 9.41727976843416e-05, "loss": 0.0501, "step": 3950 }, { "grad_norm": 0.42392799258232117, "learning_rate": 9.413400361221029e-05, "loss": 0.0466, "step": 3960 }, { "grad_norm": 0.5303824543952942, "learning_rate": 9.409508887939835e-05, "loss": 0.0497, "step": 3970 }, { "grad_norm": 0.5664520263671875, "learning_rate": 9.40560535922972e-05, "loss": 0.0488, "step": 3980 }, { "grad_norm": 0.49102678894996643, "learning_rate": 9.40168978576278e-05, "loss": 0.0461, "step": 3990 }, { "grad_norm": 0.45616212487220764, "learning_rate": 9.397762178244043e-05, "loss": 0.0454, "step": 4000 } ], "logging_steps": 10, "max_steps": 20000, "num_input_tokens_seen": 0, "num_train_epochs": 9223372036854775807, "save_steps": 4000, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": false }, "attributes": {} } }, "total_flos": 0.0, "train_batch_size": 256, "trial_name": null, "trial_params": null }